diff --git a/.claude/skills/syft-pr-review/SKILL.md b/.claude/skills/syft-pr-review/SKILL.md
index 9950e1da080..ecd4c810f53 100644
--- a/.claude/skills/syft-pr-review/SKILL.md
+++ b/.claude/skills/syft-pr-review/SKILL.md
@@ -60,8 +60,9 @@ package, and whether each claim in `body.md` is still true.
 
 ## Step 3 — Write it
 
-Save to `koen/pr-reviews/pr-<N>-<slug>.md` when a git-ignored `koen/` folder exists, otherwise
-`pr-reviews/` in the repo root — and say the file is untracked.
+Save to `koen/reviews/<N>/summary.md` when a git-ignored `koen/` folder exists, otherwise
+`reviews/<N>/summary.md` in the repo root — and say the file is untracked. One folder per PR, so
+anything else about that review lands beside the summary later.
 
 Flows come first, because they are why the reader opened the document.
 
@@ -87,36 +88,42 @@ Flows come first, because they are why the reader opened the document.
 
 - [ ] **2. What is new** — additions only
 
-  - [ ] **NEW CLASS `Name`** — `path/file.py:<line>` — <what it is for>. Built by `A.b()`. (Flow 1.2)
-  - [ ] **NEW MODULE `path/file.py`** — <what it is for>. Defines `file.py: func()`. (Flow 1.1)
-  - [ ] **NEW helpers in `path/file.py`** — `f()` <does X>, `g()` <does Y>. (Flow 1.1)
+  - [ ] **2.1 NEW CLASS `Name`** — `path/file.py:<line>` — <what it is for>. Built by `A.b()`. (Flow 1.2)
+  - [ ] **2.2 NEW MODULE `path/file.py`** — <what it is for>. Defines `file.py: func()`. (Flow 1.1)
+  - [ ] **2.3 NEW helpers in `path/file.py`** — `f()` <does X>, `g()` <does Y>. (Flow 1.1)
 
 - [ ] **3. Changes** — everything changed or deleted except tests, grouped by theme
 
-  - [ ] **A — <Theme>**
-    - [ ] **A1 <label>** (Flow 1.2) — When we <do X>, we previously <did Z>, now we <do A>.
+  - [ ] **3.1 <Theme>**
+    - [ ] **3.1.1 <label>** (Flow 1.2) — When we <do X>, we previously <did Z>, now we <do A>.
           `path/file.py: Class.method()`
-    - [ ] **A2 <label>** — DELETED `Class.old()` and `file.py: helper()`, because <reason>.
+    - [ ] **3.1.2 <label>** — DELETED `Class.old()` and `file.py: helper()`, because <reason>.
 
 - [ ] **4. Tests**
 
-  - [ ] **NEW `path/test_file.py`** — <n> tests<, and how they are set up, when it is worth a clause>
+  - [ ] **4.1 NEW `path/test_file.py`** — <n> tests<, and how they are set up, when it is worth a clause>
     - [ ] `test_a()` — <the condition, what runs, and what is then true>
     - [ ] `test_b()` — <the condition, what runs, and what is then true>
-  - [ ] **REWRITTEN `test_c()`** — `path/test_file.py` — now asserts <X> instead of <Y>, because
+  - [ ] **4.2 REWRITTEN `test_c()`** — `path/test_file.py` — now asserts <X> instead of <Y>, because
         <reason>.
-  - [ ] **UPDATED for the new code** — <n> tests across <n> files follow the new
+  - [ ] **4.3 UPDATED for the new code** — <n> tests across <n> files follow the new
         `Class.method()` signature. Nothing is asserted differently.
   - [ ] Decision: <only when a test settles something a reader would otherwise wonder about>
 
 - [ ] **5. Code standards** — only where the new code breaks one; drop the section when it does not
 
-  - [ ] `path/file.py: Class.method()` — <the problem, one line>
+  - [ ] **5.1** `path/file.py: Class.method()` — <the problem, one line>
 
 - [ ] **6. Blocked on** — drop this section unless something real stops the merge
-  - [ ] <the problem, and what has to happen before this can go in>
+
+  - [ ] **6.1** <the problem, and what has to happen before this can go in>
+
+- [ ] **7. Comments for the author** — leave it empty; the review UI writes into it
 ```
 
+Number the comments section after the sections that survived, so it is 5, 6 or 7 depending on which
+of those were dropped. It is always last, and always there.
+
 ### What counts as a code-standards problem
 
 Judge only the lines this PR touched. Do not review the code around them, and do not turn this into
@@ -131,11 +138,13 @@ line, and write nothing when there is nothing wrong. Look for:
 - string building that is not an f-string
 - a repeated or magic value that belongs in a module-level constant
 - two functions or methods that do substantially the same work; one of them belongs, called from
-  both places. Name the pair and what they share
+  both places. Name the pair and what they share. Let a repeated line or two go — the copy has to be
+  big enough that pulling it out is worth a helper
 - an import inside a function; fine only to break a circular import, and worth one short note
-- a test name that does not say what it checks, or pads a name that does. Length is free when every
-  word earns it, so drop articles and filler: `test_no_hint_when_do_owns_no_jobs`, not
-  `test_no_hint_when_the_do_owns_none_of_the_jobs`
+- a name that does not say what the thing is, or pads a name that does. Length is free when every
+  word earns it, so drop articles and filler. Say it as a plain sentence and stop there:
+  `test_no_hint_when_the_do_owns_none_of_the_jobs() violates the no-filler rule, rename to
+test_no_hint_when_do_owns_no_jobs()`
 
 ## Rules
 
@@ -181,8 +190,11 @@ line, and write nothing when there is nothing wrong. Look for:
       decisions that change how someone reads the code; skip the rest.
 - [ ] **Budget reading time:** roughly 40 bullets for a normal PR, 120 for a very large one.
 - [ ] Do not repeat the PR description. If your text matches it, you read the wrong thing.
-- [ ] Cross-reference by name, e.g. `(Flow 1.2)`. Links such as `[x](#slug)` do not work: GitHub
-      gives headings no `id` inside a comment, so there is nothing to point at.
+- [ ] **Number every section and subsection**, `**<N>.<M> <label>**`, in every section and to the
+      same depth as the template shows. A leaf bullet is not numbered — its position under a
+      numbered parent already names it. This is what lets a comment say which bullet it is about.
+- [ ] Cross-reference by number, e.g. `(Flow 1.2)` or `(3.1.2)`. Links such as `[x](#slug)` do not
+      work: GitHub gives headings no `id` inside a comment, so there is nothing to point at.
 
 ## Finally
 
diff --git a/README.md b/README.md
index ed5d93695e0..595478706db 100644
--- a/README.md
+++ b/README.md
@@ -97,14 +97,15 @@ Submit the job and retrieve results:
 ds.submit_python_job(
     user="do@org.com",
     code_path="analysis.py",
+    job_name="analysis",
 )
 ds.sync(); do.sync()
 
 # Data owner Approves & runs job
-do.jobs[0].approve()
+do.jobs["ds@org.com"]["analysis"].approve()
 do.process_approved_jobs(share_outputs_with_submitter=True)
 do.sync(); ds.sync()
-result = open(ds.jobs[-1].output_paths[0]).read()
+result = open(ds.jobs["do@org.com"]["analysis"].output_paths[0]).read()
 ```
 
 ## Packages
diff --git a/docs/API.md b/docs/API.md
index 32b0ab2841d..3e34b96d729 100644
--- a/docs/API.md
+++ b/docs/API.md
@@ -59,7 +59,43 @@ Returns a `PeerList`.
 
 Get the list of jobs. Auto-syncs before returning.
 
-Returns a `JobsList`.
+Returns a `JobsList`. Address a job by **email, then name** — a job name is
+unique per datasite and submitter, so the email is what makes the name resolve
+to one job, and both parts stay the same as jobs are added:
+
+```python
+# as the data scientist, naming the data owner
+client.jobs["do@org.com"]["analysis"].output_paths
+
+# as the data owner, naming the submitter
+client.jobs["ds@org.com"]["analysis"].approve()
+```
+
+**An email keeps the jobs it is a party to**, on either side: the datasite they
+sit on, or the person who submitted them. Usually that is the other party — a
+data scientist names the data owner, a data owner names the submitter — but
+naming yourself works and keeps your own, which is what a `PermissionError` on
+someone else's job suggests. Job names cannot contain `@`, so the two kinds of
+key never collide.
+
+Chain both emails when one submitter sent the same name to two datasites:
+
+```python
+client.jobs["do@org.com"]["ds@org.com"]["analysis"].approve()
+```
+
+A bare name (`client.jobs["analysis"]`) searches every datasite at once. It
+still works, but it raises when more than one job answers to it, and the message
+names whichever key separates them.
+
+A job name can no longer hold an `@`. A job submitted before that rule still
+resolves by name, with a `DeprecationWarning`; the next version will not resolve
+it, so rename such a job.
+
+Positional indexing (`client.jobs[0]`) also works and matches the Index column
+in the table, but positions shift as jobs are added, so prefer an email and a
+name. A position is worth using in one case: two jobs that share a datasite, a
+submitter and a name, which no email separates.
 
 ### `client.datasets`
 
@@ -174,6 +210,7 @@ Submit a Python job to a Data Owner. **DS only.**
 ds_client.submit_python_job(
     user="owner@example.com",
     code_path="/path/to/script.py",
+    job_name="analysis",
 )
 ```
 
@@ -195,11 +232,23 @@ Run all approved jobs. **DO only.**
 - `stream_output`: Stream stdout/stderr in real-time.
 - `timeout`: Timeout in seconds per job (default: 300).
 - `force_execution`: Skip version compatibility checks.
+- `ignore_peer_version`: Run jobs from peers whose version is incompatible.
 
 ```python
 do_client.process_approved_jobs()
 ```
 
+A job whose submitter runs an incompatible version is not run. Each one is
+reported by name, with the submitter and the reason:
+
+```
+⏭️  1 approved job(s) did not run:
+   • analysis (submitted by ds@example.com): Skipping peer ds@example.com: incompatible version.
+   Pass ignore_peer_version=True to run them anyway.
+```
+
+The job stays at `approved`, so it runs on the next call once the versions match.
+
 ---
 
 ## Cleanup
diff --git a/packages/syft-bg/src/syft_bg/__init__.py b/packages/syft-bg/src/syft_bg/__init__.py
index 89e5239ffa2..e293ece76b9 100644
--- a/packages/syft-bg/src/syft_bg/__init__.py
+++ b/packages/syft-bg/src/syft_bg/__init__.py
@@ -1,5 +1,7 @@
 __version__ = "0.2.2"
 
+from syft_job.logging_config import configure_package_logger
+
 from syft_bg.api import (
     AuthResult,
     AutoApproveResult,
@@ -61,3 +63,7 @@ def __getattr__(name: str):
             print("No config file found, run syft_bg.init() first")
             return
     raise AttributeError(f"module 'syft_bg' has no attribute {name!r}")
+
+
+# Last, so the imports stay at the top. Nothing here logs at import time.
+configure_package_logger(__name__)
diff --git a/packages/syft-enclave/pyproject.toml b/packages/syft-enclave/pyproject.toml
index b2c5f2b99b0..a8e8c78b2db 100644
--- a/packages/syft-enclave/pyproject.toml
+++ b/packages/syft-enclave/pyproject.toml
@@ -10,6 +10,7 @@ requires-python = ">=3.10"
 dependencies = [
     "syft>=0.10.0",  # floor: the `syft` PyPI project also hosts legacy PySyft <=0.9
     "syft-rds>=0.6.1",
+    "syft-job==0.1.40",  # imported directly for logging_config
     "pydantic-settings>=2.11.0",
     "requests>=2.32.0",
     "google-auth[pyjwt]>=2.22.0",
@@ -22,6 +23,7 @@ build-backend = "hatchling.build"
 [tool.uv.sources]
 "syft" = { workspace = true }
 "syft-rds" = { workspace = true }
+"syft-job" = { workspace = true }
 
 [tool.hatch.build.targets.wheel]
 packages = ["src/syft_enclaves"]
diff --git a/packages/syft-enclave/src/syft_enclaves/__init__.py b/packages/syft-enclave/src/syft_enclaves/__init__.py
index c7c5e509ef7..35855ad8b98 100644
--- a/packages/syft-enclave/src/syft_enclaves/__init__.py
+++ b/packages/syft-enclave/src/syft_enclaves/__init__.py
@@ -1,3 +1,4 @@
+from syft_job.logging_config import configure_package_logger
 from syft_enclaves.client import SyftEnclaveClient
 from syft_enclaves.login import login_do, login_ds
 from syft_enclaves.runner import EnclaveRunner
@@ -10,3 +11,6 @@
     "login_do",
     "login_ds",
 ]
+
+# Last, so the imports stay at the top. Nothing here logs at import time.
+configure_package_logger(__name__)
diff --git a/packages/syft-enclave/src/syft_enclaves/client.py b/packages/syft-enclave/src/syft_enclaves/client.py
index 8dce7d1e4ff..a0a39c790ba 100644
--- a/packages/syft-enclave/src/syft_enclaves/client.py
+++ b/packages/syft-enclave/src/syft_enclaves/client.py
@@ -153,7 +153,7 @@ def jobs(self) -> JobsList:
             else j
             for j in jobs_list
         ]
-        return JobsList(wrapped, jobs_list._root_email)
+        return JobsList(wrapped, jobs_list._root_email, jobs_list._has_do_role)
 
     def submit_python_job(
         self,
@@ -266,13 +266,10 @@ def approve_job(self, job: JobInfo) -> None:
         if os.environ.get("PRE_SYNC", "true").lower() == "true":
             self._rds.sync()
 
+        # approve() refuses when the party's approval file is missing, so
+        # reaching the next line means there is a file to sync.
         job.approve()
-        file_name = enclave_approval_file_name(self.email)
-        approval_file = job.job_review_path / file_name
-        if not approval_file.exists():
-            print(
-                "🟠 Approval file does not exist yet. Kindly wait until enclave sends it."
-            )
+        approval_file = job.job_review_path / enclave_approval_file_name(self.email)
         relative_path = approval_file.relative_to(self._rds.syftbox_folder)
         self._rds.sync_engine.datasite_watcher_syncer.on_file_change(
             relative_path, process_now=True
diff --git a/packages/syft-enclave/src/syft_enclaves/enclave_job_info.py b/packages/syft-enclave/src/syft_enclaves/enclave_job_info.py
index 287487a8e84..f086427cbfc 100644
--- a/packages/syft-enclave/src/syft_enclaves/enclave_job_info.py
+++ b/packages/syft-enclave/src/syft_enclaves/enclave_job_info.py
@@ -72,8 +72,11 @@ def approve(self) -> None:
         approval_file = self.job_review_path / file_name
         if not approval_file.exists():
             raise PermissionError(
-                f"No approval file found for {self.current_user_email}. "
-                f"You may not be a designated party for this job."
+                f"No approval file for {self.current_user_email} on job "
+                f"'{self.name}'. The enclave writes one per designated party "
+                f"when it distributes the job, so either it has not distributed "
+                f"this job yet — run client.sync() and retry — or you are not a "
+                f"party to it."
             )
         approval = PartyApprovalStatus.load_json(approval_file)
         if approval.status != JobStatus.PENDING:
diff --git a/packages/syft-enclave/tests/test_enclave_job_info.py b/packages/syft-enclave/tests/test_enclave_job_info.py
new file mode 100644
index 00000000000..2af7fd55813
--- /dev/null
+++ b/packages/syft-enclave/tests/test_enclave_job_info.py
@@ -0,0 +1,82 @@
+"""Unit tests for EnclaveJobInfo, the per-party approval gate.
+
+The gate lives here rather than in SyftEnclaveClient.approve_job, so these
+build a job on a tmp_path SyftBox folder instead of a four-party enclave flow.
+"""
+
+from datetime import datetime, timezone
+from pathlib import Path
+
+import pytest
+from syft_enclaves.enclave_job_info import (
+    EnclaveJobInfo,
+    PartyApprovalStatus,
+    enclave_approval_file_name,
+)
+from syft_job.client import JobClient
+from syft_job.config import SyftJobConfig
+from syft_job.job import JobInfo
+from syft_job.job_storage import JobRef
+from syft_job.models import JobState, JobStatus, JobSubmissionMetadata
+
+DO_EMAIL = "do@test.org"
+DS_EMAIL = "ds@test.org"
+
+
+def _make_enclave_job(tmp_path: Path, job_name: str = "test_job") -> EnclaveJobInfo:
+    """An enclave job on the DO's datasite, with no approval file written yet."""
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DO_EMAIL)
+    )
+    ref = JobRef(
+        datasite_email=DO_EMAIL,
+        ds_email=DS_EMAIL,
+        job_name=job_name,
+        protocol_version="1",
+    )
+    job = JobInfo(
+        job_metadata=JobSubmissionMetadata(
+            name=job_name,
+            type="python",
+            submitted_by=DS_EMAIL,
+            datasite_email=DO_EMAIL,
+            submitted_at=datetime.now(timezone.utc),
+        ),
+        state=JobState(status=JobStatus.PENDING),
+        client=client,
+        current_user_email=DO_EMAIL,
+        ref=ref,
+    )
+    return EnclaveJobInfo.from_job_info(job)
+
+
+def test_approve_refuses_when_approval_file_missing(tmp_path: Path):
+    """No approval file means the enclave has not distributed the job yet.
+
+    The message used to say the caller may not be a designated party, which is
+    the wrong cause for the common case and offers nothing to do about it.
+    """
+    job = _make_enclave_job(tmp_path)
+
+    with pytest.raises(PermissionError) as exc:
+        job.approve()
+
+    message = str(exc.value)
+    assert DO_EMAIL in message
+    assert "test_job" in message
+    assert "client.sync()" in message
+
+
+def test_approve_refuses_when_already_approved(tmp_path: Path):
+    """A second approval must not overwrite the first one's timestamp."""
+    job = _make_enclave_job(tmp_path)
+    approval_file = job.job_review_path / enclave_approval_file_name(DO_EMAIL)
+    PartyApprovalStatus(party=DO_EMAIL).save_json(approval_file)
+
+    job.approve()
+    assert PartyApprovalStatus.load_json(approval_file).status == JobStatus.APPROVED
+
+    with pytest.raises(ValueError, match="Already in status: approved"):
+        job.approve()
diff --git a/packages/syft-job/src/syft_job/__init__.py b/packages/syft-job/src/syft_job/__init__.py
index 1341d77d48f..4bd49e496e3 100644
--- a/packages/syft-job/src/syft_job/__init__.py
+++ b/packages/syft-job/src/syft_job/__init__.py
@@ -1,5 +1,6 @@
 # __version__ comes from the installed distribution metadata (see version.py).
 from .version import __version__
+from .logging_config import configure_package_logger
 
 from .client import BaseJobClient, JobClient, get_client
 from .config import SyftJobConfig
@@ -32,3 +33,6 @@
     # Migration registry
     "job_registry",
 ]
+
+# Last, so the imports stay at the top. Nothing here logs at import time.
+configure_package_logger(__name__)
diff --git a/packages/syft-job/src/syft_job/client.py b/packages/syft-job/src/syft_job/client.py
index f4a1f384fcb..662d5cb0b2e 100644
--- a/packages/syft-job/src/syft_job/client.py
+++ b/packages/syft-job/src/syft_job/client.py
@@ -604,8 +604,18 @@ def jobs(self) -> JobsList:
         """
         Get all jobs from all peer directories as an indexable list grouped by user.
 
+        Order is the same as the jobs table: jobs on your own datasite first,
+        then the other datasites alphabetically, newest-first within each one.
+        ``jobs[N]`` is the row labelled ``[N]``. Prefer an email and a name
+        (``jobs["do@org.com"]["analysis"]``); both parts stay the same as jobs
+        are added, and the email is what makes the name resolve to one job.
+
         Returns a JobsList object that can be:
-        - Indexed: jobs[0], jobs[1], etc.
+        - Indexed by email, then name (preferred): jobs["do@org.com"]["analysis"]
+        - Indexed by datasite and submitter, for a name they share:
+          jobs["do@org.com"]["ds@org.com"]["analysis"]
+        - Indexed by name across every datasite: jobs["analysis"]
+        - Indexed by position: jobs[0], jobs[1] — matches the table Index column
         - Iterated: for job in jobs
         - Displayed: print(jobs) shows separate tables for each user
         - HTML display: in Jupyter, shows separate tables for each user with jobs
@@ -618,25 +628,18 @@ def jobs(self) -> JobsList:
 
         current_jobs = self._get_all_jobs()
 
-        # Sort jobs by recent submissions first (newest first), then by user/status
         def job_sort_key(job):
-            # Parse submitted_at timestamp for sorting (most recent first)
+            # Root owner first, then peers, newest-first within each owner.
+            # This list is the authority for both jobs[N] and the table's [N].
             try:
                 if job.submitted_at:
-                    from datetime import datetime as dt
-
-                    # Parse ISO format timestamp
-                    ts = dt.fromisoformat(job.submitted_at.replace("Z", "+00:00"))
-                    # Use negative timestamp for reverse chronological order (newest first)
+                    ts = datetime.fromisoformat(job.submitted_at.replace("Z", "+00:00"))
                     time_priority = -ts.timestamp()
                 else:
-                    # Jobs without submitted_at go to the end
                     time_priority = float("inf")
             except Exception:
-                # Invalid timestamps go to the end
                 time_priority = float("inf")
 
-            # Secondary sorting: user priority (root first), then user name, then status
             user_priority = (
                 0 if job.datasite_owner_email == self.current_user_email else 1
             )
@@ -652,9 +655,9 @@ def job_sort_key(job):
             status_priority = status_order.get(job.status, 7)
 
             return (
-                time_priority,
                 user_priority,
                 job.datasite_owner_email,
+                time_priority,
                 status_priority,
                 job.name.lower(),
             )
diff --git a/packages/syft-job/src/syft_job/job.py b/packages/syft-job/src/syft_job/job.py
index 0eed2bbf7ad..f2c99a7d65a 100644
--- a/packages/syft-job/src/syft_job/job.py
+++ b/packages/syft-job/src/syft_job/job.py
@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import shutil
+import warnings
 from datetime import datetime, timezone
 from pathlib import Path
 from typing import TYPE_CHECKING, List, Optional
@@ -120,7 +121,7 @@ def _list_output_files() -> List[Path]:
                     for item in outputs_dir.iterdir()
                     if item.name != PERMISSION_FILE_NAME
                 ]
-            except Exception:
+            except OSError:
                 return []
 
         if status == JobStatus.FAILED:
@@ -174,7 +175,7 @@ def files(self) -> List[Path]:
                     ):
                         continue
                     all_files.append(f)
-        except Exception:
+        except OSError:
             pass
         return all_files
 
@@ -213,8 +214,10 @@ def approve(
 
         if self.datasite_owner_email != self.current_user_email:
             raise PermissionError(
-                f"Only the admin user ({self.datasite_owner_email}) can approve jobs in their folder. "
-                f"Current job is in {self.datasite_owner_email}'s folder."
+                f"You are {self.current_user_email}, and job '{self.name}' is on "
+                f"{self.datasite_owner_email}'s datasite. Only they can approve "
+                f"it. If you meant one of your own, select your datasite "
+                f'first: jobs["{self.current_user_email}"]["<name>"].'
             )
 
         self._state.status = JobStatus.APPROVED
@@ -246,7 +249,8 @@ def reject(self, reason: Optional[str] = None) -> None:
 
         if self.datasite_owner_email != self.current_user_email:
             raise PermissionError(
-                f"Only the admin user ({self.datasite_owner_email}) can reject jobs."
+                f"You are {self.current_user_email}, and job '{self.name}' is on "
+                f"{self.datasite_owner_email}'s datasite. Only they can reject it."
             )
 
         self._state.status = JobStatus.REJECTED
@@ -423,6 +427,27 @@ def _repr_html_(self) -> str:
         return job_info_repr_html(self)
 
 
+def _with_party(jobs: List[JobInfo], email: str) -> List[JobInfo]:
+    """The jobs ``email`` is a party to: on its datasite, or submitted by it."""
+    return [j for j in jobs if email in (j.datasite_owner_email, j.submitted_by)]
+
+
+def _with_name(jobs: List[JobInfo], name: str) -> List[JobInfo]:
+    """The jobs called ``name``."""
+    return [j for j in jobs if j.name == name]
+
+
+def _keep(jobs: List[JobInfo], key: str) -> List[JobInfo]:
+    """The jobs one subscript key keeps, read the way ``__getitem__`` reads it.
+
+    The '@' tells the two kinds of key apart, so the usage hint can measure a
+    chain before offering it. A deprecated job name holding an '@' reads here
+    as an email and keeps nothing, where ``__getitem__`` falls back to the
+    name; the hint then offers a position, which is the safe direction to err.
+    """
+    return _with_party(jobs, key) if "@" in key else _with_name(jobs, key)
+
+
 class JobsList:
     """A list-like container for JobInfo objects with nice display."""
 
@@ -431,17 +456,134 @@ def __init__(self, jobs: List[JobInfo], root_email: str, has_do_role: bool = Fal
         self._root_email = root_email
         self._has_do_role = has_do_role
 
-    def __getitem__(self, index: int | str) -> JobInfo:
+    def __getitem__(self, index: int | str) -> "JobInfo | JobsList":
+        """A job by position or name, or the jobs of one party by email.
+
+        An email keeps the jobs it is a party to, on either side. That is
+        usually the other party — a data scientist names the data owner, a data
+        owner names the submitter — but naming yourself keeps your own. So
+        ``jobs["do@x.org"]["analysis"]`` reads as one job, and chaining both —
+        ``jobs["do@x.org"]["ds@y.org"]["analysis"]`` — pins the datasite and the
+        submitter, the pair a job name is unique under. A bare name searches
+        every datasite at once and raises when more than one job answers to it.
+        Job names cannot contain ``@``, so the two kinds of key never collide.
+        """
         if isinstance(index, int):
             return self._jobs[index]
         elif isinstance(index, str):
-            for job in self._jobs:
-                if job.name == index:
-                    return job
-            raise ValueError(f"Job with name '{index}' not found")
+            if "@" in index:
+                return self._by_email(index)
+            return self._by_name(index)
         else:
             raise TypeError(f"Invalid index type: {type(index)}")
 
+    def _by_email(self, email: str) -> "JobInfo | JobsList":
+        """The jobs this email is a party to: on its datasite, or submitted by it.
+
+        One key covers both roles because which one narrows depends on who is
+        asking. A data scientist names the data owner's datasite; a data owner
+        names the submitter. Chaining the two pins the pair.
+
+        A job submitted before names could not hold an '@' answers to no party,
+        and reading it is the one thing this key must not take away, so a key
+        that names no party falls back to the name.
+        """
+        matches = _with_party(self._jobs, email)
+        if matches:
+            return JobsList(matches, self._root_email, self._has_do_role)
+
+        if any(job.name == email for job in self._jobs):
+            warnings.warn(
+                f"Job name {email!r} holds an '@', which now marks a datasite or "
+                "submitter email. Such names are deprecated and the next version "
+                "will not resolve them. Rename the job.",
+                DeprecationWarning,
+                stacklevel=3,
+            )
+            return self._by_name(email)
+
+        datasites = ", ".join(sorted({job.datasite_owner_email for job in self._jobs}))
+        submitters = ", ".join(sorted({job.submitted_by for job in self._jobs}))
+        raise ValueError(
+            f"No jobs involving {email}. These jobs are on: {datasites}. "
+            f"They were submitted by: {submitters}."
+        )
+
+    def _by_name(self, name: str) -> JobInfo:
+        matches = _with_name(self._jobs, name)
+        if not matches:
+            raise ValueError(f"Job with name '{name}' not found")
+        if len(matches) > 1:
+            raise ValueError(self._ambiguous_name_message(name, matches))
+        return matches[0]
+
+    def _ambiguous_name_message(self, name: str, matches: List[JobInfo]) -> str:
+        """Why the name did not resolve, and the narrower subscript to use.
+
+        A name is unique per datasite and submitter, not across the list, so
+        the remedy names whichever of the two separates these candidates. One
+        submitter holding a name twice across protocol layouts shares both, and
+        naming either would send the caller back to this same error, so the
+        position is the only thing left to offer.
+        """
+        locations = ", ".join(
+            f"[{i}] on {job.datasite_owner_email} from {job.submitted_by}"
+            for i, job in enumerate(self._jobs)
+            if job.name == name
+        )
+        if len({job.datasite_owner_email for job in matches}) > 1:
+            remedy = f'Select the datasite first: jobs["<datasite email>"]["{name}"].'
+        elif len({job.submitted_by for job in matches}) > 1:
+            remedy = f'Select the submitter first: jobs["<submitter email>"]["{name}"].'
+        else:
+            remedy = "One datasite and one submitter hold both, so no email "
+            remedy += "narrows them: select one by position."
+        return f"Multiple jobs are named '{name}': {locations}. {remedy}"
+
+    def hint_accessor(self) -> str | None:
+        """The subscript chain the jobs table tells a data owner to type.
+
+        Names a pending job on the DO's own datasite and gives the shortest
+        chain that reaches it and nothing else. The email in a chain is the
+        other party, so the submitter comes first; the datasite joins it only
+        when that submitter used the name on another datasite too. One
+        submitter holding a name twice across protocol layouts defeats every
+        chain, which is what the position is for.
+
+        Returns None when no such job is there to name, because the hint's
+        ``approve()`` takes only a pending job on your own datasite: a client
+        with no data-owner role, a DO who owns none of these jobs, or a DO
+        whose own jobs have all been reviewed already. The hint also offers
+        ``accept_by_depositing_result()``, which an approved job would still
+        take; withholding both is the cost of never printing a command that
+        raises.
+        """
+        if not self._has_do_role:
+            return None
+        owned = [
+            j
+            for j in self._jobs
+            if j.datasite_owner_email == self._root_email and j.status == "pending"
+        ]
+        if not owned:
+            return None
+        pick = owned[0]
+        chains = (
+            (pick.submitted_by, pick.name),
+            (self._root_email, pick.submitted_by, pick.name),
+        )
+        for keys in chains:
+            if self._reaches_only(keys, pick):
+                return "".join(f'["{key}"]' for key in keys)
+        return f"[{self._jobs.index(pick)}]"
+
+    def _reaches_only(self, keys: tuple[str, ...], job: JobInfo) -> bool:
+        """Whether subscripting by ``keys`` in turn reaches ``job`` and nothing else."""
+        reached = self._jobs
+        for key in keys:
+            reached = _keep(reached, key)
+        return len(reached) == 1 and reached[0] is job
+
     def __len__(self) -> int:
         return len(self._jobs)
 
@@ -449,10 +591,10 @@ def __iter__(self):
         return iter(self._jobs)
 
     def __str__(self) -> str:
-        return jobs_list_str(self._jobs, self._root_email, self._has_do_role)
+        return jobs_list_str(self._jobs, self.hint_accessor())
 
     def __repr__(self) -> str:
         return f"JobsList({len(self._jobs)} jobs)"
 
     def _repr_html_(self) -> str:
-        return jobs_list_repr_html(self._jobs, self._root_email, self._has_do_role)
+        return jobs_list_repr_html(self._jobs, self.hint_accessor())
diff --git a/packages/syft-job/src/syft_job/job_repr.py b/packages/syft-job/src/syft_job/job_repr.py
index a12507aaab7..8c900b55281 100644
--- a/packages/syft-job/src/syft_job/job_repr.py
+++ b/packages/syft-job/src/syft_job/job_repr.py
@@ -731,18 +731,30 @@ def job_info_repr_html(job: "JobInfo") -> str:
         """
 
 
-def jobs_list_str(
-    jobs: List["JobInfo"], root_email: str, has_do_role: bool = False
-) -> str:
+def _owner_groups(jobs: List["JobInfo"]) -> List[tuple[str, List["JobInfo"]]]:
+    """Split jobs into consecutive owner sections without reordering.
+
+    JobClient.jobs is the authority for order. Re-sorting here is what
+    made the table's [N] disagree with jobs[N]. A list that is not already
+    grouped by owner gives an owner more than one section; the row indexes
+    stay correct.
+    """
+    groups: List[tuple[str, List["JobInfo"]]] = []
+    for job in jobs:
+        owner = job.datasite_owner_email
+        if groups and groups[-1][0] == owner:
+            groups[-1][1].append(job)
+        else:
+            groups.append((owner, [job]))
+    return groups
+
+
+def jobs_list_str(jobs: List["JobInfo"], hint_accessor: str | None = None) -> str:
     """Format jobs list as separate tables grouped by user."""
     if not jobs:
         return "📭 No jobs found.\n"
 
-    jobs_by_user: dict[str, list["JobInfo"]] = {}
-    for job in jobs:
-        if job.datasite_owner_email not in jobs_by_user:
-            jobs_by_user[job.datasite_owner_email] = []
-        jobs_by_user[job.datasite_owner_email].append(job)
+    owner_groups = _owner_groups(jobs)
 
     status_emojis = {
         "received": "📨",
@@ -758,25 +770,12 @@ def jobs_list_str(
     lines.append("📊 Jobs Overview")
     lines.append("=" * 50)
 
-    total_jobs = 0
+    total_jobs = len(jobs)
     global_status_counts: dict[str, int] = {}
 
-    def user_sort_key(item):
-        user_email, _user_jobs = item
-        if user_email == root_email:
-            return (0, user_email)
-        return (1, user_email)
-
-    sorted_users = sorted(jobs_by_user.items(), key=user_sort_key)
-
     job_index = 0
 
-    for user_email, user_jobs in sorted_users:
-        if not user_jobs:
-            continue
-
-        total_jobs += len(user_jobs)
-
+    for user_email, user_jobs in owner_groups:
         lines.append("")
         lines.append(f"👤 {user_email}")
         lines.append("-" * 60)
@@ -795,9 +794,7 @@ def user_sort_key(item):
         lines.append(header)
         lines.append("-" * len(header))
 
-        sorted_jobs = user_jobs
-
-        for job in sorted_jobs:
+        for job in user_jobs:
             emoji = status_emojis.get(job.status, "❓")
             status_display = f"{emoji} {job.status}"
             approval_display = job.approval_method or "—"
@@ -824,7 +821,8 @@ def user_sort_key(item):
 
     lines.append("")
     lines.append("=" * 50)
-    lines.append(f"📈 Total: {total_jobs} jobs across {len(jobs_by_user)} users")
+    owner_count = len({job.datasite_owner_email for job in jobs})
+    lines.append(f"📈 Total: {total_jobs} jobs across {owner_count} users")
 
     global_summary_parts = []
     for status, count in global_status_counts.items():
@@ -834,18 +832,17 @@ def user_sort_key(item):
     if global_summary_parts:
         lines.append("📋 Global: " + " | ".join(global_summary_parts))
 
-    if has_do_role:
+    if hint_accessor is not None:
         lines.append("")
         lines.append(
-            "💡 Use job_client.jobs[0].approve() to approve jobs or job_client.jobs[0].accept_by_depositing_result('file_or_folder') to complete jobs"
+            f"💡 Use job_client.jobs{hint_accessor}.approve() to approve jobs or "
+            f"job_client.jobs{hint_accessor}.accept_by_depositing_result('file_or_folder') to complete jobs"
         )
 
     return "\n".join(lines)
 
 
-def jobs_list_repr_html(
-    jobs: List["JobInfo"], root_email: str, has_do_role: bool = False
-) -> str:
+def jobs_list_repr_html(jobs: List["JobInfo"], hint_accessor: str | None = None) -> str:
     """HTML representation for Jupyter notebooks with enhanced visual appeal."""
     if not jobs:
         return """
@@ -919,11 +916,7 @@ def jobs_list_repr_html(
             </div>
             """
 
-    jobs_by_user: dict[str, list["JobInfo"]] = {}
-    for job in jobs:
-        if job.datasite_owner_email not in jobs_by_user:
-            jobs_by_user[job.datasite_owner_email] = []
-        jobs_by_user[job.datasite_owner_email].append(job)
+    owner_groups = _owner_groups(jobs)
 
     status_styles = {
         "received": {
@@ -964,6 +957,7 @@ def jobs_list_repr_html(
     }
 
     total_jobs = len(jobs)
+    owner_count = len({job.datasite_owner_email for job in jobs})
     global_status_counts: dict[str, int] = {}
     for job in jobs:
         global_status_counts[job.status] = global_status_counts.get(job.status, 0) + 1
@@ -1217,25 +1211,12 @@ def jobs_list_repr_html(
         <div class="syftjob-overview">
             <div class="syftjob-global-header">
                 <h3>📊 Jobs Overview</h3>
-                <p>Total: {total_jobs} jobs across {len(jobs_by_user)} users</p>
+                <p>Total: {total_jobs} jobs across {owner_count} users</p>
             </div>
         """
 
-    def user_sort_key(item):
-        user_email, _user_jobs = item
-        if user_email == root_email:
-            return (0, user_email)
-        return (1, user_email)
-
-    sorted_users = sorted(jobs_by_user.items(), key=user_sort_key)
-
     job_index = 0
-    for user_email, user_jobs in sorted_users:
-        if not user_jobs:
-            continue
-
-        sorted_user_jobs = user_jobs
-
+    for user_email, user_jobs in owner_groups:
         user_status_counts: dict[str, int] = {}
         for job in user_jobs:
             user_status_counts[job.status] = user_status_counts.get(job.status, 0) + 1
@@ -1264,7 +1245,7 @@ def user_sort_key(item):
                     <tbody>
             """
 
-        for i, job in enumerate(sorted_user_jobs):
+        for i, job in enumerate(user_jobs):
             style_info = status_styles.get(job.status, {"emoji": "❓"})
             row_class = "syftjob-row-even" if i % 2 == 0 else "syftjob-row-odd"
 
@@ -1314,10 +1295,10 @@ def user_sort_key(item):
     html += """
                 </div>
         """
-    if has_do_role:
-        html += """
+    if hint_accessor is not None:
+        html += f"""
                 <div class="syftjob-hint">
-                    💡 Use <code class="syftjob-code">jobs[0].approve()</code> to approve jobs or <code class="syftjob-code">jobs[0].accept_by_depositing_result('file_or_folder')</code> to complete jobs
+                    💡 Use <code class="syftjob-code">jobs{hint_accessor}.approve()</code> to approve jobs or <code class="syftjob-code">jobs{hint_accessor}.accept_by_depositing_result('file_or_folder')</code> to complete jobs
                 </div>
         """
     html += """
diff --git a/packages/syft-job/src/syft_job/job_runner.py b/packages/syft-job/src/syft_job/job_runner.py
index 471eca5d811..21a56a63f59 100644
--- a/packages/syft-job/src/syft_job/job_runner.py
+++ b/packages/syft-job/src/syft_job/job_runner.py
@@ -2,6 +2,7 @@
 import shutil
 import subprocess
 import time
+import warnings
 from datetime import datetime, timezone
 from pathlib import Path
 from typing import List, Set
@@ -515,32 +516,76 @@ def _get_job_info(self, ref: JobRef) -> JobInfo:
             ref=ref,
         )
 
+    @staticmethod
+    def _resolve_skip_jobs(
+        approved_jobs: List[JobRef],
+        skip_jobs: list[tuple[str, str]] | None,
+        skip_job_names: list[str] | None,
+    ) -> Set[tuple[str, str]]:
+        """The (job_name, ds_email) pairs to skip, accepting the old name-only form.
+
+        A bare name expands to every submitter who used it, which is what
+        ``skip_job_names`` did. Bare names in ``skip_jobs`` are read the same
+        way, so a caller who passed the old third positional argument keeps the
+        behaviour it had.
+        """
+        pairs: Set[tuple[str, str]] = set()
+        names: List[str] = list(skip_job_names or [])
+        for entry in skip_jobs or []:
+            if isinstance(entry, str):
+                names.append(entry)
+            else:
+                pairs.add((entry[0], entry[1]))
+
+        if names:
+            warnings.warn(
+                "Skipping jobs by name alone is deprecated: a name is unique "
+                "per datasite and submitter, so it also drops other "
+                "submitters' jobs of the same name. Pass "
+                "skip_jobs=[(job_name, ds_email), ...] instead.",
+                DeprecationWarning,
+                stacklevel=3,
+            )
+            wanted = set(names)
+            pairs.update(
+                (ref.job_name, ref.ds_email)
+                for ref in approved_jobs
+                if ref.job_name in wanted
+            )
+        return pairs
+
     def process_approved_jobs(
         self,
         stream_output: bool = True,
         timeout: int | None = None,
-        skip_job_names: list[str] | None = None,
+        skip_jobs: list[tuple[str, str]] | None = None,
         share_outputs_with_submitter: bool = False,
         share_logs_with_submitter: bool = False,
+        skip_job_names: list[str] | None = None,
     ) -> None:
         """Process all jobs in approved status.
 
         Args:
             stream_output: If True (default), stream output in real-time.
             timeout: Timeout in seconds per job. Defaults to 300 (5 minutes).
-            skip_job_names: Optional list of job names to skip.
+            skip_jobs: Optional (job_name, ds_email) pairs to skip. A name alone
+                does not identify a job — it is unique per datasite and
+                submitter — so skipping by name drops every other job that
+                shares it.
             share_outputs_with_submitter: If True, grant read access on outputs to submitter.
             share_logs_with_submitter: If True, grant read access on logs to submitter.
+            skip_job_names: Deprecated. The name-only form of ``skip_jobs``.
         """
         approved_jobs = self._get_jobs_in_approved()
 
         if not approved_jobs:
             return
 
-        # Filter out jobs to skip
-        if skip_job_names:
-            skip_set = set(skip_job_names)
-            approved_jobs = [j for j in approved_jobs if j.job_name not in skip_set]
+        skip_set = self._resolve_skip_jobs(approved_jobs, skip_jobs, skip_job_names)
+        if skip_set:
+            approved_jobs = [
+                j for j in approved_jobs if (j.job_name, j.ds_email) not in skip_set
+            ]
 
         if not approved_jobs:
             return
diff --git a/packages/syft-job/src/syft_job/job_storage.py b/packages/syft-job/src/syft_job/job_storage.py
index 51a96103025..795656bfd77 100644
--- a/packages/syft-job/src/syft_job/job_storage.py
+++ b/packages/syft-job/src/syft_job/job_storage.py
@@ -120,6 +120,10 @@ def validate_job_name(job_name: str) -> None:
             raise ValueError(
                 f"Job name {job_name!r} is reserved for protocol version directories"
             )
+        if "@" in job_name:
+            # An '@' marks a datasite email in JobsList.__getitem__, which is
+            # how jobs["do@x.org"]["analysis"] tells the two keys apart.
+            raise ValueError(f"Job name {job_name!r} cannot contain '@'")
 
     # -- scanning (union over all protocol layouts) -------------------------------
     def iter_submission_refs(self, datasite_email: str) -> Iterator[JobRef]:
diff --git a/packages/syft-job/src/syft_job/logging_config.py b/packages/syft-job/src/syft_job/logging_config.py
new file mode 100644
index 00000000000..6096c3a9e92
--- /dev/null
+++ b/packages/syft-job/src/syft_job/logging_config.py
@@ -0,0 +1,31 @@
+"""Shared logger setup for syft-job and the packages built on it.
+
+``syft`` configures the ``syft`` logger only. ``syft_job``, ``syft_rds``,
+``syft_enclaves`` and ``syft_bg`` are sibling top-level namespaces, so they
+inherit the root logger, which has no handler. INFO records are then dropped,
+and WARNING and above fall back to ``logging.lastResort`` — bare text on
+stderr, with no way to raise or lower the level. Each package calls
+``configure_package_logger`` once, at the end of its ``__init__``.
+
+The helper lives here, not in ``syft``, because syft-job does not depend on
+syft, while the other three packages depend on syft-job.
+"""
+
+import logging
+
+
+def configure_package_logger(name: str, level: int = logging.INFO) -> logging.Logger:
+    """Give the ``name`` logger a handler and a level, but only if it has none.
+
+    A caller who sets up their own logging keeps full control. Records still
+    propagate to the root logger, so pytest's caplog and any root handler still
+    see them; in default Python the root logger has no handler, so nothing is
+    printed twice.
+    """
+    logger = logging.getLogger(name)
+    if not logger.handlers:
+        handler = logging.StreamHandler()
+        handler.setFormatter(logging.Formatter("[%(levelname)s] %(message)s"))
+        logger.addHandler(handler)
+        logger.setLevel(level)
+    return logger
diff --git a/packages/syft-job/tests/test_job_flow.py b/packages/syft-job/tests/test_job_flow.py
index d5022e76417..2aba9c2d9cc 100644
--- a/packages/syft-job/tests/test_job_flow.py
+++ b/packages/syft-job/tests/test_job_flow.py
@@ -3,15 +3,17 @@
 import time
 from pathlib import Path
 
+import pytest
 from syft_job.client import JobClient
 from syft_job.config import SyftJobConfig
 from syft_job.job_runner import SyftJobRunner
-from syft_perms import SyftPermContext
 from syft_job.models import JobState
-
+from syft_perms import SyftPermContext
 
 DO_EMAIL = "do@test.org"
 DS_EMAIL = "ds@test.org"
+PEER_EMAIL = "peer@test.org"
+DS2_EMAIL = "ds2@test.org"
 
 MAIN_PY = """\
 import os
@@ -252,3 +254,172 @@ def test_timeout_does_not_hang_runner(tmp_path: Path):
     # 3s job timeout + venv setup + tree-kill cleanup should fit well under 60s.
     assert elapsed < 60, f"process_approved_jobs took {elapsed:.1f}s — likely hung"
     assert do_client.jobs[0].status == "failed"
+
+
+def test_skipping_one_job_spares_its_same_named_sibling(tmp_path: Path):
+    """A skip must name the submitter as well as the job.
+
+    Two data scientists can submit the same job name to one data owner. Given
+    only the name, the runner dropped both — so a job its peer was compatible
+    with never ran, and nothing said so.
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_config = SyftJobConfig(syftbox_folder=syftbox, current_user_email=DO_EMAIL)
+    do_client = JobClient(config=do_config)
+    do_runner = SyftJobRunner(config=do_config)
+
+    for ds_email in (DS_EMAIL, DS2_EMAIL):
+        JobClient(
+            config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=ds_email)
+        ).submit_python_job(
+            user=DO_EMAIL, code_path=str(code_file), job_name="shared.job"
+        )
+
+    for job in do_client.jobs:
+        job.approve()
+
+    do_runner.process_approved_jobs(
+        stream_output=False, timeout=60, skip_jobs=[("shared.job", DS_EMAIL)]
+    )
+
+    final = {job.submitted_by: job.status for job in do_client.jobs}
+    assert final[DS_EMAIL] == "approved", "the skipped job must not run"
+    assert final[DS2_EMAIL] == "done", "its same-named sibling must still run"
+
+
+def test_runner_ignores_approved_jobs_on_another_datasite(tmp_path: Path):
+    """`jobs` spans every datasite in the folder; the runner runs only its own.
+
+    A peer's approved job is visible but was never a candidate, which is why the
+    skip report in syft-rds filters on the datasite owner before naming a job as
+    one that did not run.
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_config = SyftJobConfig(syftbox_folder=syftbox, current_user_email=DO_EMAIL)
+    peer_config = SyftJobConfig(syftbox_folder=syftbox, current_user_email=PEER_EMAIL)
+    do_client = JobClient(config=do_config)
+    peer_client = JobClient(config=peer_config)
+    do_runner = SyftJobRunner(config=do_config)
+
+    ds_client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    )
+    ds_client.submit_python_job(
+        user=DO_EMAIL, code_path=str(code_file), job_name="mine.job"
+    )
+    ds_client.submit_python_job(
+        user=PEER_EMAIL, code_path=str(code_file), job_name="theirs.job"
+    )
+
+    do_client.jobs["mine.job"].approve()
+    peer_client.jobs["theirs.job"].approve()
+
+    # The DO sees both, including the peer's approved job.
+    visible = {job.name: job.status for job in do_client.jobs}
+    assert visible == {"mine.job": "approved", "theirs.job": "approved"}
+
+    do_runner.process_approved_jobs(stream_output=False, timeout=60)
+
+    after = {job.name: job.status for job in do_client.jobs}
+    assert after["mine.job"] == "done"
+    assert after["theirs.job"] == "approved", "a peer's job is not this runner's to run"
+
+
+def test_deprecated_skip_job_names_still_skips_by_name(tmp_path: Path):
+    """The old name-only argument keeps its old behaviour, and says it is old.
+
+    It drops every submitter's job of that name — which is the bug — so it warns
+    rather than silently changing what an existing caller gets.
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_config = SyftJobConfig(syftbox_folder=syftbox, current_user_email=DO_EMAIL)
+    do_client = JobClient(config=do_config)
+    do_runner = SyftJobRunner(config=do_config)
+
+    for ds_email in (DS_EMAIL, DS2_EMAIL):
+        JobClient(
+            config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=ds_email)
+        ).submit_python_job(
+            user=DO_EMAIL, code_path=str(code_file), job_name="shared.job"
+        )
+    for job in do_client.jobs:
+        job.approve()
+
+    with pytest.warns(DeprecationWarning, match="Skipping jobs by name alone"):
+        do_runner.process_approved_jobs(
+            stream_output=False, timeout=60, skip_job_names=["shared.job"]
+        )
+
+    assert [job.status for job in do_client.jobs] == ["approved", "approved"]
+
+
+def test_approval_error_names_both_parties(tmp_path: Path):
+    """The old message called the peer "the admin user" and never named you."""
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DO_EMAIL)
+    )
+    JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    ).submit_python_job(
+        user=PEER_EMAIL, code_path=str(code_file), job_name="peer-owned.job"
+    )
+    # The owner scans it into pending, so approve() reaches the ownership check.
+    JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=PEER_EMAIL)
+    ).scan_inbox()
+
+    job = do_client.jobs["peer-owned.job"]
+    assert job.status == "pending"
+    with pytest.raises(PermissionError) as exc:
+        job.approve()
+
+    message = str(exc.value)
+    assert DO_EMAIL in message, "must say who you are"
+    assert PEER_EMAIL in message, "must say whose datasite it is"
+    assert "peer-owned.job" in message
+    assert "admin user" not in message
+    # The remedy must not send them back to the same job: a bare name searches
+    # every datasite, and this job is the only one that answers to this one.
+    assert f'jobs["{DO_EMAIL}"]' in message, "must point at your own datasite"
+
+
+def test_job_files_do_not_hide_non_filesystem_error(tmp_path: Path, monkeypatch):
+    """Only the filesystem may truncate the file list; other errors propagate."""
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DO_EMAIL)
+    )
+    JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    ).submit_python_job(user=DO_EMAIL, code_path=str(code_file), job_name="files.job")
+
+    job = do_client.jobs["files.job"]
+    assert job.files, "sanity: the job has files"
+
+    def boom(*args, **kwargs):
+        raise RuntimeError("not a filesystem problem")
+
+    monkeypatch.setattr(Path, "rglob", boom)
+    with pytest.raises(RuntimeError, match="not a filesystem problem"):
+        job.files
diff --git a/packages/syft-job/tests/test_jobs_addressing.py b/packages/syft-job/tests/test_jobs_addressing.py
new file mode 100644
index 00000000000..2f6be963f04
--- /dev/null
+++ b/packages/syft-job/tests/test_jobs_addressing.py
@@ -0,0 +1,338 @@
+"""How a job is addressed: the subscript chain, and the hint that teaches it.
+
+These drive JobClient over a tmp_path SyftBox folder, like the lifecycle tests
+next door, but assert on what jobs[...] returns and what the table's hint says
+rather than on a job running.
+"""
+
+import re
+from pathlib import Path
+
+import pytest
+from syft_job.client import JobClient
+from syft_job.config import SyftJobConfig
+
+DO_EMAIL = "do@test.org"
+DS_EMAIL = "ds@test.org"
+PEER_EMAIL = "peer@test.org"
+DS2_EMAIL = "ds2@test.org"
+
+MAIN_PY = """\
+import os
+
+print("hello from job")
+os.makedirs("outputs", exist_ok=True)
+with open("outputs/result.txt", "w") as f:
+    f.write("done")
+"""
+
+
+def test_hint_names_submitter_and_job(tmp_path: Path):
+    """The DO hint must point at jobs["submitter"]["name"], not jobs[0].
+
+    Positional indexing is not safe to recommend: positions shift as jobs are
+    added. A bare name is not safe either — it searches every datasite, so a
+    peer's job of the same name can answer instead. The email in a chain is the
+    other party, which for a data owner is the submitter.
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    # has_do_role gates the hint — it is only shown to a data owner.
+    do_config = SyftJobConfig(
+        syftbox_folder=syftbox, current_user_email=DO_EMAIL, has_do_role=True
+    )
+    ds_config = SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    ds_client = JobClient(config=ds_config)
+    do_client = JobClient(config=do_config)
+
+    ds_client.submit_python_job(
+        user=DO_EMAIL, code_path=str(code_file), job_name="analysis.job"
+    )
+
+    jobs = do_client.jobs
+    text = str(jobs)
+    html = jobs._repr_html_()
+
+    for rendering in (text, html):
+        assert f'jobs["{DS_EMAIL}"]["analysis.job"].approve()' in rendering
+        assert "jobs[0]" not in rendering
+        assert DO_EMAIL not in rendering.split("💡")[1], "the DO's own email is noise"
+
+    # The hint names a job the DO can actually approve, as written.
+    assert do_client.jobs[DS_EMAIL]["analysis.job"].status == "pending"
+
+
+def _index_name_pairs_from_text(text: str) -> list[tuple[int, str]]:
+    return [(int(i), name) for i, name in re.findall(r"\[(\d+)\s*\]\s+(\S+)", text)]
+
+
+def _index_name_pairs_from_html(html: str) -> list[tuple[int, str]]:
+    return [
+        (int(i), name.strip())
+        for i, name in re.findall(
+            r'class="syftjob-index">\[(\d+)\]</span>.*?'
+            r'class="syftjob-td syftjob-job-name">\s*([^<]+)',
+            html,
+            flags=re.DOTALL,
+        )
+    ]
+
+
+def test_jobs_table_index_matches_getitem(tmp_path: Path):
+    """The [N] printed in the table must be the N that jobs[N] returns.
+
+    The table groups by datasite owner (root first, then peers). __getitem__
+    used to subscript a newest-first list, so with two owners the row labelled
+    [0] was not jobs[0].
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_config = SyftJobConfig(
+        syftbox_folder=syftbox, current_user_email=DO_EMAIL, has_do_role=True
+    )
+    ds_config = SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    ds_client = JobClient(config=ds_config)
+    do_client = JobClient(config=do_config)
+
+    # Older job on the root datasite, then a newer one on a peer datasite.
+    ds_client.submit_python_job(
+        user=DO_EMAIL, code_path=str(code_file), job_name="older-root-job"
+    )
+    ds_client.submit_python_job(
+        user=PEER_EMAIL, code_path=str(code_file), job_name="newest-peer-job"
+    )
+
+    jobs = do_client.jobs
+    assert [job.name for job in jobs] == ["older-root-job", "newest-peer-job"]
+
+    text_pairs = _index_name_pairs_from_text(str(jobs))
+    html_pairs = _index_name_pairs_from_html(jobs._repr_html_())
+    assert text_pairs == [(0, "older-root-job"), (1, "newest-peer-job")]
+    assert html_pairs == [(0, "older-root-job"), (1, "newest-peer-job")]
+
+    for index, name in text_pairs + html_pairs:
+        assert jobs[index].name == name
+
+
+def test_hint_names_submitter_for_shared_name(tmp_path: Path):
+    """A name two submitters share needs the submitter to reach one job.
+
+    Job names are unique per datasite and submitter, so two data scientists can
+    submit "analysis" to the same data owner. The hint adds the submitter rather
+    than giving up on the name.
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_config = SyftJobConfig(
+        syftbox_folder=syftbox, current_user_email=DO_EMAIL, has_do_role=True
+    )
+    do_client = JobClient(config=do_config)
+    ds1_client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    )
+    ds2_client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS2_EMAIL)
+    )
+
+    # "solo" first, so the ambiguous name is the newest and would otherwise win.
+    ds1_client.submit_python_job(
+        user=DO_EMAIL, code_path=str(code_file), job_name="solo"
+    )
+    ds1_client.submit_python_job(
+        user=DO_EMAIL, code_path=str(code_file), job_name="analysis"
+    )
+    ds2_client.submit_python_job(
+        user=DO_EMAIL, code_path=str(code_file), job_name="analysis"
+    )
+
+    jobs = do_client.jobs
+    for rendering in (str(jobs), jobs._repr_html_()):
+        assert f'jobs["{DS2_EMAIL}"]["analysis"].approve()' in rendering
+
+    # The chain the hint gives reaches exactly one job, and the datasite alone
+    # does not, which is why it names the submitter.
+    job = jobs[DS2_EMAIL]["analysis"]
+    assert (job.submitted_by, job.status) == (DS2_EMAIL, "pending")
+    with pytest.raises(ValueError, match="Multiple jobs are named 'analysis'"):
+        jobs[DO_EMAIL]["analysis"]
+
+
+def test_hint_names_every_submitter_of_shared_name(tmp_path: Path):
+    """Every job in the table shares its name, so every chain needs a submitter.
+
+    The datasite alone reaches neither, and the hint still has to name something
+    the data owner can run.
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_config = SyftJobConfig(
+        syftbox_folder=syftbox, current_user_email=DO_EMAIL, has_do_role=True
+    )
+    do_client = JobClient(config=do_config)
+    for ds_email in (DS_EMAIL, DS2_EMAIL):
+        JobClient(
+            config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=ds_email)
+        ).submit_python_job(
+            user=DO_EMAIL, code_path=str(code_file), job_name="analysis"
+        )
+
+    jobs = do_client.jobs
+    for rendering in (str(jobs), jobs._repr_html_()):
+        assert f'jobs["{DS2_EMAIL}"]["analysis"].approve()' in rendering
+        assert "jobs[0]" not in rendering
+
+    for ds_email in (DS_EMAIL, DS2_EMAIL):
+        assert jobs[ds_email]["analysis"].submitted_by == ds_email
+
+
+def test_no_hint_when_do_owns_no_jobs(tmp_path: Path):
+    """Every subscript would name a job on another datasite, which approve() refuses.
+
+    A hint is worse than no hint when the only command it can give raises
+    PermissionError.
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_config = SyftJobConfig(
+        syftbox_folder=syftbox, current_user_email=DO_EMAIL, has_do_role=True
+    )
+    do_client = JobClient(config=do_config)
+    JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    ).submit_python_job(
+        user=PEER_EMAIL, code_path=str(code_file), job_name="peer-owned.job"
+    )
+
+    jobs = do_client.jobs
+    assert [job.datasite_owner_email for job in jobs] == [PEER_EMAIL]
+    for rendering in (str(jobs), jobs._repr_html_()):
+        assert "peer-owned.job" in rendering
+        assert "💡" not in rendering
+
+
+def test_email_key_resolves_name_shared_by_two_datasites(tmp_path: Path):
+    """One submitter can send the same job name to two data owners.
+
+    The bare name has no way to choose between them and raises. Selecting the
+    datasite first leaves one job, which is the whole point of the two-step
+    subscript.
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    ds_client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    )
+    for owner in (DO_EMAIL, PEER_EMAIL):
+        ds_client.submit_python_job(
+            user=owner, code_path=str(code_file), job_name="analysis.job"
+        )
+        JobClient(
+            config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=owner)
+        ).scan_inbox()
+
+    jobs = ds_client.jobs
+    with pytest.raises(ValueError, match="Multiple jobs are named 'analysis.job'"):
+        jobs["analysis.job"]
+
+    for owner in (DO_EMAIL, PEER_EMAIL):
+        job = jobs[owner]["analysis.job"]
+        assert job.datasite_owner_email == owner
+
+
+def test_email_key_names_emails_it_has(tmp_path: Path):
+    """An email with no jobs on it is a typo the message has to help with."""
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    ds_client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    )
+    ds_client.submit_python_job(
+        user=DO_EMAIL, code_path=str(code_file), job_name="analysis.job"
+    )
+
+    with pytest.raises(ValueError) as exc:
+        ds_client.jobs["nobody@test.org"]
+
+    message = str(exc.value)
+    assert "nobody@test.org" in message
+    assert DO_EMAIL in message, "must name the datasites that do have jobs"
+    assert DS_EMAIL in message, "must name the submitters too"
+
+
+def test_hint_adds_datasite_when_submitter_used_name_twice(
+    tmp_path: Path,
+):
+    """One submitter can send the same name to the DO and to a peer.
+
+    The submitter alone then reaches both, so this is the one case where the
+    DO's own datasite has to join the chain, and the only thing that earns the
+    third key.
+    """
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    do_client = JobClient(
+        config=SyftJobConfig(
+            syftbox_folder=syftbox, current_user_email=DO_EMAIL, has_do_role=True
+        )
+    )
+    ds_client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    )
+    for owner in (DO_EMAIL, PEER_EMAIL):
+        ds_client.submit_python_job(
+            user=owner, code_path=str(code_file), job_name="analysis.job"
+        )
+        JobClient(
+            config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=owner)
+        ).scan_inbox()
+
+    jobs = do_client.jobs
+    assert {job.datasite_owner_email for job in jobs} == {DO_EMAIL, PEER_EMAIL}
+    chain = f'jobs["{DO_EMAIL}"]["{DS_EMAIL}"]["analysis.job"]'
+    for rendering in (str(jobs), jobs._repr_html_()):
+        assert f"{chain}.approve()" in rendering
+
+    # The submitter alone reaches both datasites, so the chain needs both keys.
+    with pytest.raises(ValueError, match="Multiple jobs are named 'analysis.job'"):
+        jobs[DS_EMAIL]["analysis.job"]
+    assert jobs[DO_EMAIL][DS_EMAIL]["analysis.job"].datasite_owner_email == DO_EMAIL
+
+
+def test_job_name_cannot_contain_at_sign(tmp_path: Path):
+    """An '@' is how a subscript tells a datasite email from a job name."""
+    syftbox = tmp_path / "SyftBox"
+    syftbox.mkdir()
+    code_file = tmp_path / "main.py"
+    code_file.write_text(MAIN_PY)
+
+    ds_client = JobClient(
+        config=SyftJobConfig(syftbox_folder=syftbox, current_user_email=DS_EMAIL)
+    )
+    with pytest.raises(ValueError, match="cannot contain '@'"):
+        ds_client.submit_python_job(
+            user=DO_EMAIL, code_path=str(code_file), job_name="ds@test.org"
+        )
diff --git a/packages/syft-rds/src/syft_rds/__init__.py b/packages/syft-rds/src/syft_rds/__init__.py
index d4826ca6b8a..c86d6b6c62d 100644
--- a/packages/syft-rds/src/syft_rds/__init__.py
+++ b/packages/syft-rds/src/syft_rds/__init__.py
@@ -1,5 +1,7 @@
 """syft-rds: Remote Data Science product composed on top of syft."""
 
+from syft_job.logging_config import configure_package_logger
+
 from syft_rds.client import SyftRDSClient
 from syft_rds.config import SyftRDSClientConfig
 from syft_rds.job_auto_approval import auto_approve_and_run_jobs, job_matches_criteria
@@ -25,3 +27,6 @@
     "Environment",
     "check_env",
 ]
+
+# Last, so the imports stay at the top. Nothing here logs at import time.
+configure_package_logger(__name__)
diff --git a/packages/syft-rds/src/syft_rds/client.py b/packages/syft-rds/src/syft_rds/client.py
index e6cac4ee6bd..8b40f9fe476 100644
--- a/packages/syft-rds/src/syft_rds/client.py
+++ b/packages/syft-rds/src/syft_rds/client.py
@@ -25,6 +25,14 @@
 logger = logging.getLogger(__name__)
 
 
+def _print_skipped_jobs(skipped: list[tuple[str, str, str]]) -> None:
+    """Report the approved jobs that did not run, and why."""
+    print(f"\n⏭️  {len(skipped)} approved job(s) did not run:")
+    for job_name, peer_email, reason in skipped:
+        print(f"   • {job_name} (submitted by {peer_email}): {reason}")
+    print("   Pass ignore_peer_version=True to run them anyway.")
+
+
 class SyftRDSClient(BaseModel):
     # Holds live service objects (sync engine + RDS-owned managers), not
     # serializable data, so arbitrary types are allowed.
@@ -346,11 +354,18 @@ def process_approved_jobs(
         if self.job_runner is None:
             raise ValueError("Job runner is not configured for this client")
 
-        skip_job_names = []
+        skipped: list[tuple[str, str, str]] = []
 
         if not force_execution:
+            # Only jobs on this client's own datasite: job_client.jobs spans
+            # every datasite in the folder, but the runner never runs a job on
+            # a peer's. Reporting those as skipped would offer a remedy that
+            # cannot help.
             approved_jobs = [
-                job for job in self.job_client.jobs if job.status == "approved"
+                job
+                for job in self.job_client.jobs
+                if job.status == "approved"
+                and job.datasite_owner_email == self.job_client.current_user_email
             ]
             for job in approved_jobs:
                 result = self.sync_engine.peer_manager.get_peer_compatibility_status(
@@ -360,16 +375,26 @@ def process_approved_jobs(
                 )
                 result.maybe_warn()
                 if result.should_skip:
-                    skip_job_names.append(job.name)
+                    skipped.append(
+                        (
+                            job.name,
+                            job.submitted_by,
+                            result.explanation_skip or "peer version is not compatible",
+                        )
+                    )
 
         self.job_runner.process_approved_jobs(
             stream_output=stream_output,
             timeout=timeout,
-            skip_job_names=skip_job_names if skip_job_names else None,
+            skip_jobs=[(name, peer) for name, peer, _ in skipped] or None,
             share_outputs_with_submitter=share_outputs_with_submitter,
             share_logs_with_submitter=share_logs_with_submitter,
         )
 
+        # maybe_warn() above logs the peer, never the job.
+        if skipped:
+            _print_skipped_jobs(skipped)
+
         if self._pre_sync_enabled:
             self.sync_engine.sync()
 
diff --git a/packages/syft-rds/tests/test_datasets_jobs_repr.py b/packages/syft-rds/tests/test_datasets_jobs_repr.py
index 8d63b88cfca..18f5c0e7e8b 100644
--- a/packages/syft-rds/tests/test_datasets_jobs_repr.py
+++ b/packages/syft-rds/tests/test_datasets_jobs_repr.py
@@ -1,5 +1,7 @@
 """Tests for SyftDatasetManager and JobsList repr and indexing."""
 
+import warnings
+
 import pytest
 
 from syft_rds import SyftRDSClient
@@ -134,7 +136,12 @@ def test_dataset_repr_html_mentions_mock_files():
 # --- JobsList tests ---
 
 
-def _make_job_info(name: str, status: str = "pending") -> JobInfo:
+def _make_job_info(
+    name: str,
+    status: str = "pending",
+    ds_email: str = "ds@test.com",
+    owner_email: str = "test@test.com",
+) -> JobInfo:
     """Create a minimal JobInfo for testing."""
     from datetime import datetime, timezone
     from pathlib import Path
@@ -151,15 +158,15 @@ def _make_job_info(name: str, status: str = "pending") -> JobInfo:
     submission_config = JobSubmissionMetadata(
         name=name,
         type="python",
-        submitted_by="ds@test.com",
-        datasite_email="ds@test.com",
+        submitted_by=ds_email,
+        datasite_email=ds_email,
         submitted_at=datetime.now(timezone.utc),
     )
     state = JobState(status=JobStatus(status))
     # Identity (owner, submitter, name) comes from the path-derived ref.
     ref = JobRef(
-        datasite_email="test@test.com",
-        ds_email="ds@test.com",
+        datasite_email=owner_email,
+        ds_email=ds_email,
         job_name=name,
         protocol_version="1",
     )
@@ -198,6 +205,241 @@ def test_jobs_list_getitem_str_not_found():
         jobs["nonexistent"]
 
 
+def test_jobs_list_getitem_str_ambiguous():
+    """Two submitters can use one job name; the lookup must not guess between them.
+
+    Names are unique per datasite and submitter, so the same name can appear
+    more than once in one list. Returning the first match approves the wrong
+    job and says nothing.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info("analysis", ds_email="ds1@test.com"),
+            _make_job_info("analysis", ds_email="ds2@test.com"),
+        ],
+        root_email="test@test.com",
+    )
+    with pytest.raises(ValueError, match="Multiple jobs are named 'analysis'") as exc:
+        jobs["analysis"]
+
+    message = str(exc.value)
+    assert "[0] on test@test.com from ds1@test.com" in message
+    assert "[1] on test@test.com from ds2@test.com" in message
+    # One datasite holds both, so the submitter is what separates them.
+    assert 'jobs["<submitter email>"]["analysis"]' in message
+
+
+def test_jobs_list_getitem_str_ambiguous_names_the_datasite():
+    """One submitter can send the same job name to two data owners.
+
+    The submitter is then identical on every row, so the message has to name
+    the datasite as well. This is the DS-side call the README recommends.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info("analysis", owner_email="do1@test.com"),
+            _make_job_info("analysis", owner_email="do2@test.com"),
+        ],
+        root_email="ds@test.com",
+    )
+    with pytest.raises(ValueError) as exc:
+        jobs["analysis"]
+
+    message = str(exc.value)
+    assert "[0] on do1@test.com" in message
+    assert "[1] on do2@test.com" in message
+    assert 'jobs["<datasite email>"]["analysis"]' in message, (
+        "the datasite narrows this"
+    )
+
+
+def test_jobs_list_getitem_email_selects_datasite():
+    """An email key narrows to one datasite, where the name is unique."""
+    jobs = JobsList(
+        [
+            _make_job_info("analysis", owner_email="do1@test.com"),
+            _make_job_info("analysis", owner_email="do2@test.com"),
+        ],
+        root_email="ds@test.com",
+    )
+    on_do1 = jobs["do1@test.com"]
+    assert isinstance(on_do1, JobsList)
+    assert len(on_do1) == 1
+    assert on_do1["analysis"].datasite_owner_email == "do1@test.com"
+
+
+def test_jobs_list_getitem_email_not_found():
+    jobs = JobsList(
+        [_make_job_info("analysis", owner_email="do1@test.com")],
+        root_email="ds@test.com",
+    )
+    with pytest.raises(ValueError, match="No jobs involving nobody@test.com"):
+        jobs["nobody@test.com"]
+
+
+def test_jobs_list_getitem_submitter_narrows_one_datasite():
+    """Two submitters to one datasite share a name the datasite cannot separate.
+
+    Chaining the submitter is what reaches one job, and without it the lookup
+    must keep raising rather than guess.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info("analysis", ds_email="ds1@test.com"),
+            _make_job_info("analysis", ds_email="ds2@test.com"),
+        ],
+        root_email="test@test.com",
+    )
+    with pytest.raises(ValueError, match="Multiple jobs are named 'analysis'"):
+        jobs["test@test.com"]["analysis"]
+
+    job = jobs["test@test.com"]["ds2@test.com"]["analysis"]
+    assert job.submitted_by == "ds2@test.com"
+
+
+def test_jobs_list_getitem_legacy_name_with_at_warns():
+    """A job named before the '@' ban must stay readable, and say it is on notice.
+
+    Nothing warned the submitter at the time, so the name is already on disk.
+    The email key answers first; only a key no party answers to falls back.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info(
+                "report@2026-09", owner_email="do@test.com", ds_email="ds@test.com"
+            )
+        ],
+        root_email="do@test.com",
+    )
+    with pytest.warns(DeprecationWarning, match="deprecated"):
+        job = jobs["report@2026-09"]
+    assert job.name == "report@2026-09"
+
+    # A real party still wins the key, and warns about nothing.
+    with warnings.catch_warnings():
+        warnings.simplefilter("error")
+        assert len(jobs["do@test.com"]) == 1
+
+
+def test_hint_prefers_pending_job_over_unique_name():
+    """The hint's approve() only accepts a pending job, so status wins.
+
+    A finished job with a name no one shares once won the pick, and handed the
+    data owner a command that raises on status.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info("solo", ds_email="ds1@test.com", status="done"),
+            _make_job_info("analysis", ds_email="ds1@test.com", status="pending"),
+            _make_job_info("analysis", ds_email="ds2@test.com", status="pending"),
+        ],
+        root_email="test@test.com",
+        has_do_role=True,
+    )
+    assert jobs.hint_accessor() == '["ds1@test.com"]["analysis"]'
+
+
+def test_hint_chain_reaches_job_it_names():
+    """Whatever chain the hint gives, subscripting by it must return that job.
+
+    The hint and the lookup read a chain through the same evaluator, so this
+    holds for the two-key and three-key forms alike.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info(
+                "analysis", owner_email="do@test.com", ds_email="ds@test.com"
+            ),
+            _make_job_info(
+                "analysis", owner_email="peer@test.com", ds_email="ds@test.com"
+            ),
+        ],
+        root_email="do@test.com",
+        has_do_role=True,
+    )
+    # One submitter, two datasites: the chain needs all three keys.
+    assert jobs.hint_accessor() == '["do@test.com"]["ds@test.com"]["analysis"]'
+
+    reached = jobs["do@test.com"]["ds@test.com"]["analysis"]
+    assert reached.datasite_owner_email == "do@test.com"
+
+
+def test_hint_gives_position_when_no_chain_resolves():
+    """One submitter can hold a name twice across protocol layouts.
+
+    Datasite and submitter are then identical on both rows, so no chain reaches
+    one job and the position is all that is left.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info("analysis", ds_email="ds1@test.com"),
+            _make_job_info("analysis", ds_email="ds1@test.com"),
+        ],
+        root_email="test@test.com",
+        has_do_role=True,
+    )
+    assert jobs.hint_accessor() == "[0]"
+
+
+def test_ambiguous_name_offers_position_when_no_email_narrows():
+    """One submitter can hold a name twice across protocol layouts.
+
+    Datasite and submitter are identical on both rows, so naming either would
+    return the caller to this same error. Only the position separates them.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info("analysis", ds_email="ds1@test.com"),
+            _make_job_info("analysis", ds_email="ds1@test.com"),
+        ],
+        root_email="test@test.com",
+    )
+    with pytest.raises(ValueError) as exc:
+        jobs["analysis"]
+
+    message = str(exc.value)
+    assert "by position" in message
+    assert "Select the submitter" not in message, "the submitter does not narrow this"
+    assert "Select the datasite" not in message
+
+
+def test_no_hint_when_no_owned_job_is_pending():
+    """The hint's approve() takes a pending job, so a reviewed list gets no hint.
+
+    Every job here is finished, and naming one would print a command that
+    raises on status rather than on ownership.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info("analysis", status="done"),
+            _make_job_info("failed-one", status="failed"),
+            _make_job_info("approved-one", status="approved"),
+        ],
+        root_email="test@test.com",
+        has_do_role=True,
+    )
+    assert jobs.hint_accessor() is None
+
+
+def test_jobs_list_repr_counts_distinct_owners():
+    """The owner total counts owners, not table sections.
+
+    The renderers group consecutive rows to keep the table order equal to the
+    list order. A list that is not grouped by owner gives an owner more than
+    one section, which must not inflate the total.
+    """
+    jobs = JobsList(
+        [
+            _make_job_info("job-a", owner_email="do1@test.com"),
+            _make_job_info("job-b", owner_email="do2@test.com"),
+            _make_job_info("job-c", owner_email="do1@test.com"),
+        ],
+        root_email="do1@test.com",
+    )
+    assert "3 jobs across 2 users" in str(jobs)
+    assert "3 jobs across 2 users" in jobs._repr_html_()
+
+
 def test_jobs_list_getitem_invalid_type():
     jobs = JobsList(
         [_make_job_info("job-a")],
diff --git a/packages/syft-rds/tests/test_version_negotiation.py b/packages/syft-rds/tests/test_version_negotiation.py
index aaf422a92b5..56da05bd1b5 100644
--- a/packages/syft-rds/tests/test_version_negotiation.py
+++ b/packages/syft-rds/tests/test_version_negotiation.py
@@ -1,9 +1,12 @@
 """Version gating on the RDS job submission and execution paths."""
 
+from unittest.mock import patch
+
 import pytest
 
 from syft.sync.version.exceptions import VersionUnknownError
-from syft.sync.version.version_info import VersionInfo
+from syft.sync.version.peer_manager import PeerCompatibilityResult, PeerManager
+from syft.sync.version.version_info import CompatibilityStatus, VersionInfo
 from syft_rds import SyftRDSClient
 
 
@@ -113,9 +116,9 @@ def test_job_execution_forced_with_incompatible_version(self):
         executed_jobs = []
 
         def mock_process_approved_jobs(
-            stream_output=True, timeout=None, skip_job_names=None, **kwargs
+            stream_output=True, timeout=None, skip_jobs=None, **kwargs
         ):
-            executed_jobs.append(skip_job_names)
+            executed_jobs.append(skip_jobs)
 
         do.job_runner.process_approved_jobs = mock_process_approved_jobs
 
@@ -123,3 +126,44 @@ def mock_process_approved_jobs(
 
         assert len(executed_jobs) == 1
         assert executed_jobs[0] is None  # No jobs skipped when force=True
+
+
+class TestSkippedJobsAreReported:
+    """process_approved_jobs must say which approved jobs it did not run.
+
+    maybe_warn() logs the peer, not the job, so on its own it leaves the data
+    owner with a job stuck at 'approved' and nothing naming it.
+    """
+
+    def test_skipped_job_name_and_reason_reach_stdout(self, tmp_path, capfd):
+        ds, do = SyftRDSClient.pair_with_mock_drive_service_connection(
+            use_in_memory_cache=False,
+            sync_automatically=False,
+        )
+
+        code_path = tmp_path / "skipped.py"
+        code_path.write_text('print("hello")')
+        ds.submit_python_job(
+            user=do.email, code_path=str(code_path), job_name="skipped.job"
+        )
+        do.sync()
+        do.jobs["skipped.job"].approve()
+
+        skip = PeerCompatibilityResult(
+            peer_email=ds.email,
+            status=CompatibilityStatus.INCOMPATIBLE,
+            should_skip=True,
+            explanation_skip=f"Skipping peer {ds.email}: incompatible version.",
+        )
+        capfd.readouterr()
+        with patch.object(
+            PeerManager, "get_peer_compatibility_status", return_value=skip
+        ):
+            do.process_approved_jobs()
+        out, _ = capfd.readouterr()
+
+        assert "skipped.job" in out
+        assert ds.email in out
+        assert "incompatible version" in out
+        assert "ignore_peer_version=True" in out
+        assert do.jobs["skipped.job"].status == "approved"
diff --git a/syft/__init__.py b/syft/__init__.py
index d2fb13658d8..879688b2595 100644
--- a/syft/__init__.py
+++ b/syft/__init__.py
@@ -20,6 +20,7 @@
 
 from syft.version import SYFT_VERSION as __version__  # noqa: F401, E402
 from syft.sync.login import login_do, login_ds, login  # noqa: F401, E402
+from syft.sync.peers.exceptions import SyftPeerNotReadyError  # noqa: F401, E402
 from syft.utils import (  # noqa: F401, E402
     resolve_path,
     resolve_dataset_file_path,
diff --git a/syft/sync/peers/exceptions.py b/syft/sync/peers/exceptions.py
new file mode 100644
index 00000000000..db2aa7855c0
--- /dev/null
+++ b/syft/sync/peers/exceptions.py
@@ -0,0 +1,18 @@
+"""Peer-related exceptions for syft."""
+
+
+class SyftPeerNotReadyError(ValueError):
+    """A peer is known but not yet usable for the operation.
+
+    Carries the cause and the remedy separately so callers — a readiness
+    helper, or a diagnostic — can show either one on its own.
+
+    Subclasses ValueError because these conditions were reported as a bare
+    ValueError before, and callers catch that.
+    """
+
+    def __init__(self, peer_email: str, cause: str, remedy: str):
+        self.peer_email = peer_email
+        self.cause = cause
+        self.remedy = remedy
+        super().__init__(f"{cause} {remedy}")
diff --git a/syft/sync/peers/peer_store.py b/syft/sync/peers/peer_store.py
index d42c38a3328..577d3642eb2 100644
--- a/syft/sync/peers/peer_store.py
+++ b/syft/sync/peers/peer_store.py
@@ -11,7 +11,8 @@
 import syft_crypto_python as syc
 from pydantic import BaseModel, PrivateAttr
 
-from syft.sync.peers.peer import Peer
+from syft.sync.peers.exceptions import SyftPeerNotReadyError
+from syft.sync.peers.peer import Peer, PeerState
 
 # Encryption key bundles persist inside the participant's own SyftBox datasite
 # folder, under private/ (which is never synced to Drive). This scopes keys per
@@ -123,21 +124,65 @@ def set_peers(self, peers: List[Peer]) -> None:
 
     def _ensure_private_keys(self) -> syc.SyftPrivateKeys:
         if self._private_keys is None:
-            raise ValueError("No private keys — call generate_keys() first")
+            raise ValueError(
+                f"No private keys for {self.email}. Encryption is on for this "
+                "client, but it holds no key pair. Keys are created at login, "
+                "so log in again to create them."
+            )
         return self._private_keys
 
     def _ensure_peer(self, email: str) -> Peer:
         peer = self.get_cached_peer(email)
         if peer is None:
-            raise ValueError(f"No cached peer for {email}")
+            raise SyftPeerNotReadyError(
+                email,
+                cause=f"No cached peer for {email}.",
+                remedy=(
+                    "Run client.sync() to refresh your peers, or "
+                    f"client.add_peer('{email}') if you have not added them."
+                ),
+            )
         return peer
 
     def _ensure_peer_bundle(self, email: str) -> dict:
         peer = self._ensure_peer(email)
         if peer.public_encryption_bundle is None:
-            raise ValueError(f"No public encryption bundle for {email}")
+            cause, remedy = self._missing_bundle_reason(peer)
+            raise SyftPeerNotReadyError(email, cause=cause, remedy=remedy)
         return peer.public_encryption_bundle
 
+    @staticmethod
+    def _missing_bundle_reason(peer: Peer) -> tuple[str, str]:
+        """The cause and the remedy for a peer that has no encryption bundle.
+
+        A bundle is only stored for a peer who is accepted (or requested by
+        you) and who has already published one, so the four states below are
+        the four reasons the bundle can be missing.
+        """
+        email = peer.email
+        if peer.state == PeerState.REQUESTED_BY_PEER:
+            return (
+                f"{email} asked to peer with you, but you have not accepted yet.",
+                f"Run client.approve_peer_request('{email}').",
+            )
+        if peer.state == PeerState.REJECTED:
+            return (
+                f"You rejected {email}, so no encryption bundle was kept.",
+                f"Run client.add_peer('{email}') to peer with them again.",
+            )
+        if peer.state == PeerState.REQUESTED_BY_ME:
+            return (
+                f"You asked to peer with {email}, but they have not accepted yet.",
+                "Wait for them to accept, then run client.sync().",
+            )
+        return (
+            f"{email} is an accepted peer, but you do not have their public "
+            "encryption bundle. Either they have not published one yet, or you "
+            "have not synced since they did.",
+            "Run client.sync(). If it continues, the peer may not have "
+            "encryption enabled.",
+        )
+
     # ========== Crypto methods ==========
 
     def generate_keys(self) -> None:
diff --git a/tests/unit/test_encryption.py b/tests/unit/test_encryption.py
index cfc839d871f..9d0fb18200f 100644
--- a/tests/unit/test_encryption.py
+++ b/tests/unit/test_encryption.py
@@ -3,7 +3,8 @@
 
 import pytest
 
-from syft.sync.peers.peer import Peer
+from syft.sync.peers.exceptions import SyftPeerNotReadyError
+from syft.sync.peers.peer import Peer, PeerState
 from syft.sync.peers.peer_store import PeerStore
 from syft.sync.syftbox_manager import SyftboxManager
 from tests.unit.test_sync_manager import path_for_job
@@ -49,22 +50,62 @@ def test_encrypt_decrypt_roundtrip():
 def test_encrypt_without_keys_raises():
     ps = PeerStore(email="alice@example.com", use_encryption=True)
     ps.add_peer(Peer(email="bob@example.com"))
-    with pytest.raises(ValueError, match="No private keys"):
+    with pytest.raises(ValueError, match="Keys are created at login"):
         ps.encrypt("bob@example.com", b"data")
 
 
-def test_encrypt_without_peer_bundle_raises():
+@pytest.mark.parametrize(
+    "state, cause, remedy",
+    [
+        (
+            PeerState.ACCEPTED,
+            "accepted peer, but you do not have their public",
+            "Run client.sync()",
+        ),
+        (
+            PeerState.REQUESTED_BY_ME,
+            "they have not accepted yet",
+            "Wait for them to accept",
+        ),
+        (
+            PeerState.REQUESTED_BY_PEER,
+            "you have not accepted yet",
+            "client.approve_peer_request('bob@example.com')",
+        ),
+        (
+            PeerState.REJECTED,
+            "You rejected bob@example.com",
+            "client.add_peer('bob@example.com')",
+        ),
+    ],
+)
+def test_encrypt_without_peer_bundle_names_the_cause(state, cause, remedy):
+    """A missing bundle means one of four things; the message must say which.
+
+    Naming neither the cause nor the remedy is what left data owners stuck on
+    this error with nothing to act on.
+    """
     ps = PeerStore(email="alice@example.com", use_encryption=True)
     ps.generate_keys()
-    ps.add_peer(Peer(email="bob@example.com"))
-    with pytest.raises(ValueError, match="No public encryption bundle"):
+    ps.add_peer(Peer(email="bob@example.com", state=state))
+
+    with pytest.raises(SyftPeerNotReadyError) as exc:
         ps.encrypt("bob@example.com", b"data")
 
+    assert cause in exc.value.cause
+    assert remedy in exc.value.remedy
+    assert str(exc.value) == f"{exc.value.cause} {exc.value.remedy}"
+
+
+def test_peer_not_ready_error_is_a_value_error():
+    """Callers that catch ValueError keep working."""
+    assert issubclass(SyftPeerNotReadyError, ValueError)
+
 
 def test_try_decrypt_no_keys():
     ps = PeerStore(email="alice@example.com", use_encryption=True)
     data = b"some unencrypted data"
-    with pytest.raises(ValueError, match="No private keys"):
+    with pytest.raises(ValueError, match="Keys are created at login"):
         ps.decrypt("bob@example.com", data) == data
 
 
@@ -72,8 +113,10 @@ def test_try_decrypt_no_peer_bundle():
     ps = PeerStore(email="alice@example.com", use_encryption=True)
     ps.generate_keys()
     data = b"some unencrypted data"
-    with pytest.raises(ValueError, match="No cached peer for"):
-        ps.decrypt("bob@example.com", data) == data
+    with pytest.raises(SyftPeerNotReadyError) as exc:
+        ps.decrypt("bob@example.com", data)
+    assert "No cached peer for bob@example.com" in exc.value.cause
+    assert "client.add_peer('bob@example.com')" in exc.value.remedy
 
 
 def test_try_decrypt_invalid_envelope():
diff --git a/tests/unit/test_package_logging.py b/tests/unit/test_package_logging.py
new file mode 100644
index 00000000000..cabef30476d
--- /dev/null
+++ b/tests/unit/test_package_logging.py
@@ -0,0 +1,37 @@
+"""Each package must configure its own logger when imported on its own.
+
+`syft/__init__.py` configures only the `syft` logger. The other packages are
+sibling top-level namespaces, so without this they inherit the root logger —
+no handler, level WARNING — and every logger.info/warning/error call in them
+is dropped, swallowed tracebacks included.
+
+Each case runs in its own interpreter: the point is that importing the package
+alone is enough, with no dependence on `syft` being imported first.
+"""
+
+import subprocess
+import sys
+
+import pytest
+
+PACKAGES = ["syft", "syft_job", "syft_rds", "syft_enclaves", "syft_bg"]
+
+PROBE = """
+import importlib, logging, sys
+importlib.import_module({name!r})
+logger = logging.getLogger({name!r})
+print(logger.getEffectiveLevel(), bool(logger.handlers))
+"""
+
+
+@pytest.mark.parametrize("package", PACKAGES)
+def test_package_logger_is_audible(package):
+    result = subprocess.run(
+        [sys.executable, "-c", PROBE.format(name=package)],
+        capture_output=True,
+        text=True,
+    )
+    assert result.returncode == 0, result.stderr
+    level, has_handler = result.stdout.strip().splitlines()[-1].split()
+    assert int(level) == 20, f"{package} effective level is {level}, want INFO (20)"
+    assert has_handler == "True", f"{package} logger has no handler"
diff --git a/uv.lock b/uv.lock
index 8bfd2ad056b..cfad79d8bf1 100644
--- a/uv.lock
+++ b/uv.lock
@@ -4619,13 +4619,14 @@ dev = [{ name = "ipykernel", specifier = ">=7.1.0" }]
 
 [[package]]
 name = "syft-enclave"
-version = "0.1.0"
+version = "0.1.1"
 source = { editable = "packages/syft-enclave" }
 dependencies = [
     { name = "google-auth", extra = ["pyjwt"] },
     { name = "pydantic-settings" },
     { name = "requests" },
     { name = "syft" },
+    { name = "syft-job" },
     { name = "syft-rds" },
 ]
 
@@ -4635,6 +4636,7 @@ requires-dist = [
     { name = "pydantic-settings", specifier = ">=2.11.0" },
     { name = "requests", specifier = ">=2.32.0" },
     { name = "syft", editable = "." },
+    { name = "syft-job", editable = "packages/syft-job" },
     { name = "syft-rds", editable = "packages/syft-rds" },
 ]