feat(references): contributing to repositories you do not own

[CybotTM]

New reference: first contribution into a repository you do not own. Every rule in it exists because skipping it produced a public artifact that had to be walked back, all in one upstream engagement (FriendsOfTYPO3/fractor, 2026-08-25):

Rule	What happened without it
Read the open issue AND PR lists, not just keyword searches	Three searches empty; the repo's #427 described the same defect in other words, #430 already fixed it — duplicate issue and PR both closed with thanks
Measure comment style, don't import house style	Two 5-line comment blocks were the longest in their entire codebase; a reviewer asked why
Draft-first	The PR went out ready; the reviewer expected draft
Run ALL contributing steps	composer docs was skipped
Revert what their fixers touched	Rectify wanted to reformat four unrelated files into the PR
Assess by running, not by reading	Running the competing fix against the own reproduction settled the maintainer's "does it fix your case?" in one comment

SKILL.md gains the When-to-Use trigger and the reference row. main sat at 499 of the 500-word cap, so existing prose is compressed to make room — the commit message lists each compression; no guidance is removed. Validator green (was the blocker), markdownlint green.

Follow-through once merged: the project-local memory feedback_read_the_issue_list_before_filing.md in the nr-llm session memory gets re-homed to point here (promote-then-tombstone).

Assisted by claude-code:claude-opus-5 — Session

[github-actions]

Dependency Review

✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.

Scanned Files

None

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[CybotTM]

Copilot review is unavailable — account-wide monthly quota. Reviewed the diff myself; what I checked beyond reading it:

• The word-cap compressions are lossless. main sat at 499 of the 500-word cap, so adding the reference row and trigger required trimming existing prose. Each trim is listed in the commit message; I re-read every compressed line against its original — no pointer and no rule disappeared (the one removed parenthetical reference is carried by the table row that already listed it).
• The validator that rejected the first commit passes now (Validate skill repo structure: Passed, 498 words), and markdownlint-cli2 is clean on both files — the same two gates CI runs.
• The reference's claims are session-verified, not invented: every rule cites the concrete failure it comes from, and the two shell snippets in it were run, not sketched.

What this cannot do: a reference only helps if it is read before the first artifact goes out — that is why the When-to-Use trigger exists, and it is the part I would most want a human eye on: is the trigger phrasing findable enough?

Assisted by claude-code:claude-opus-5 — Session

[CybotTM]

Self-review: 839286d

The review this pull request demands is unsatisfiable (Copilot quota wall or repeated bot failures on this head). Per the documented fallback, the diff on this head was reviewed by the PR author; this comment is the on-the-record attestation the merge gate reads back. It stops matching on the next push.