Gate scoped memory audits with exact conflict and lifecycle regression fixtures

BENCHMARKS.md

@@ -94,14 +94,14 @@ interpretation and do not count as additional benchmark-quality gains.
 ### Public numeric evidence registry
 
 Every exact public aggregate retained below comes from the checked-in, public-safe
-[`offline-fixtures-pr262-git-bytes-20261014.json`](docs/benchmark-evidence/offline-fixtures-pr262-git-bytes-20261014.json) artifact. Its
+[`offline-fixtures-pr263-git-bytes-20261014.json`](docs/benchmark-evidence/offline-fixtures-pr263-git-bytes-20261014.json) artifact. Its
 SHA-256 is
-`1b4ba1d95ae003bab4c5d15a7d619fc7a7b813d66387bc9530a74babad5221eb`, also recorded in the
+`9050c5b88343bd05984372975da0c2a25090dfe081c39e4fd2c95940f865177a`, also recorded in the
 adjacent `.sha256` file. The artifact contains no raw questions, answers, prompts, customer data,
 or per-record content fingerprints.
 
 The fixture-suite digest is
-`e5f78cc61a584b64051cc8650c8e2f68cff9a3ccb096df57e8a621d376f829b8`. The artifact defines
+`7c8079520618e40c47d81354ea30408e8c9288b4ce6dd0f629539d2317c21ec8`. The artifact defines
 the digest algorithm and records the SHA-256 of every suite and dataset file. Each evidence ID
 also binds its exact command through `sha256(UTF-8 exact command)`:
 
@@ -126,13 +126,13 @@ grounded checks in separate panels; provider billing and MCP transport are not m
 the SVG and matching PNG with:
 
 ```bash
-python scripts/render_benchmark_report.py --report docs/benchmark-evidence/offline-fixtures-pr262-git-bytes-20261014.json --output docs/images/context-efficiency.svg --png-output docs/images/context-efficiency.png
+python scripts/render_benchmark_report.py --report docs/benchmark-evidence/offline-fixtures-pr263-git-bytes-20261014.json --output docs/images/context-efficiency.svg --png-output docs/images/context-efficiency.png
 ```
 
 The companion examples are also generated from that artifact with:
 
 ```bash
-python -m scripts.render_benchmark_examples --report docs/benchmark-evidence/offline-fixtures-pr262-git-bytes-20261014.json --output docs/images/evidence-backed-agent-examples.svg --png-output docs/images/evidence-backed-agent-examples.png
+python -m scripts.render_benchmark_examples --report docs/benchmark-evidence/offline-fixtures-pr263-git-bytes-20261014.json --output docs/images/evidence-backed-agent-examples.svg --png-output docs/images/evidence-backed-agent-examples.png
 ```
 The historical-to-executable mapping is in
 [`docs/BENCHMARK_CHANGE_COVERAGE.md`](docs/BENCHMARK_CHANGE_COVERAGE.md).

README.md

@@ -50,7 +50,7 @@ Connect a coding agent over MCP with the [agent setup guide](https://github.com/
 The current registered artifact contains three deterministic offline fixture runs. It separates context size, retrieval quality, and grounded decision checks; these small fixtures do not establish general task performance.
 
 <p align="center">
-  <img src="https://raw.githubusercontent.com/Coding-Dev-Tools/engraphis/dfa3151d06ca3c226b50bc2e43305f4b23d42903/docs/images/context-efficiency.svg" alt="Three registered offline fixtures: structure-aware chunking reduced mean retrieved context from 740.3 to 214.3 tokens per question (71.1%, 18 questions); the serialized JSON-shape proxy fell from 24,590 to 11,138 tokens across 26 payload samples and 260 recalls. Candidate and packed retrieval quality (Recall@5, Hit@5, and answer-token recall) are shown separately (each 1.000). Grounded checks show 5/5 answerable queries grounded and 6/6 abstention queries rejected, including a 1/1 quarantined-evidence probe; 11/11 decisions were correct. MCP transport and provider billing were not measured. Artifact SHA-256 prefix 1b4ba1d95ae0; see the benchmark guide for the full checksum." width="100%">
+  <img src="https://raw.githubusercontent.com/Coding-Dev-Tools/engraphis/5f94cd2a1f01c5627cb62104015d184ca9c30f01/docs/images/context-efficiency.svg" alt="Three registered offline fixtures: structure-aware chunking reduced mean retrieved context from 740.3 to 214.3 tokens per question (71.1%, 18 questions); the serialized JSON-shape proxy fell from 24,590 to 11,138 tokens across 26 payload samples and 260 recalls. Candidate and packed retrieval quality (Recall@5, Hit@5, and answer-token recall) are shown separately (each 1.000). Grounded checks show 5/5 answerable queries grounded and 6/6 abstention queries rejected, including a 1/1 quarantined-evidence probe; 11/11 decisions were correct. MCP transport and provider billing were not measured. Artifact SHA-256 prefix 9050c5b88343; see the benchmark guide for the full checksum." width="100%">
   <br>
   <sup>Three offline fixtures separate context reduction, candidate and packed retrieval quality, and grounded behavior. The chart shows the artifact checksum prefix; see the benchmark guide for the full checksum and reproduction steps.</sup>
 </p>
@@ -61,7 +61,7 @@ The current registered artifact contains three deterministic offline fixture run
 | Recall payload proxy | JSON-shape proxy: 24,590 → 11,138 tokens (54.71% lower, 26 samples; 260 timed recalls) | Candidate and packed Recall@5, Hit@5, and answer-token recall are each 1.000 |
 | Grounded decisions | 5/5 answerable queries grounded; 6/6 abstention queries rejected, including 1/1 quarantined-evidence check | 11/11 decisions correct |
 
-The payload figure is a serialized JSON-shape estimate, not an MCP transport measurement or provider billing total. See the [Benchmark methodology](https://github.com/Coding-Dev-Tools/engraphis/blob/dfa3151d06ca3c226b50bc2e43305f4b23d42903/BENCHMARKS.md) for artifact identity, counting methods, reproduction commands, external-evaluation boundaries, and limitations.
+The payload figure is a serialized JSON-shape estimate, not an MCP transport measurement or provider billing total. See the [Benchmark methodology](https://github.com/Coding-Dev-Tools/engraphis/blob/5f94cd2a1f01c5627cb62104015d184ca9c30f01/BENCHMARKS.md) for artifact identity, counting methods, reproduction commands, external-evaluation boundaries, and limitations.
 
 ## Optional Jev assistance
 

docs/SPEC_CRAWL.md

@@ -185,6 +185,8 @@ These four Classic MCP tools are also discoverable and callable through Smart MC
 
 CI verifies determinism, axis classification, flag recall, qualified-token avoidance, section counts, and score boundaries in both the full stack and NumPy-only jobs. Qualified-token avoidance measures the explicitly labelled negative examples; it is not a general false-positive precision estimate:
 
+The same gate checks eight synthetic memory-audit contracts in both input orders: unrelated-subject noise, agreeing and opposing nested directives, subject and claim-kind separation, stored orphan connectivity, unknown effective dates, and effective-time ordering. It compares exact conflict pairs and retirement targets, requires compact node responses, and reports the fixture digest. These bounded regression fixtures do not establish general classifier quality.
+
 ```bash
 python -m eval.spec_crawl
 ```