[TheJeje20]

Summary

Generate the three documentation-driven skills from code_review_graph/skills.py and keep the checked-in SKILL.md files synced with that source.

Changes

• Added a generator for build-graph, review-delta, and review-pr
• Regenerated the three checked-in skills/*/SKILL.md files
• Added regression tests to keep source and generated files aligned

Validation

• python -m compileall code_review_graph tests
• pytest tests/test_document_skills.py tests/test_skills.py -q
• 194 passed

[tirth8205]

The generator matches the three checked-in files byte for byte and the sync test does catch drift, verified by mutating skills/build-graph/SKILL.md. Tests and mypy are green. But ruff fails with 22 E501 line-too-long errors in the new DOCUMENTARY_SKILLS dict, skills.py lines 829-914, and CI runs ruff check on code_review_graph/ as a hard gate. Please reflow those long string literals under 100 chars using implicit concatenation; your own sync test will confirm the rendered output stays identical. Two smaller things. The description says the three SKILL.md files were regenerated, but the diff does not touch them; they already matched, so please correct that. In tests/test_document_skills.py the tool-name loop in test_render_documentary_skill_markdown_contains_expected_sections is tautological since content always contains the body verbatim, and get_affected_flows_tool appears in no body. Assert against the frontmatter and title instead, or drop the loop. No conflict with #779 or #821; those edit the other four skills and merge cleanly in either order.

[tirth8205]

Outside this sweep's mission of verified coding-session token savings: the new documentary generator has byte-identical output and no runtime callers. It does not repair Qoder installation, because neither the installer nor wheel packaging is wired to it; the separate #909 repair uses the existing seven skill files directly. No code from this PR has been retained.

[tirth8205]

This fails the checks once merged into staging.

pytest (py3.13, staging+PR merge): "FAILED tests/test_document_skills.py::test_checked_in_documentary_skills_match_generator" / "1 failed, 3225 passed, 9 skipped, 2 xpassed, 1 warning in 172.01s (0:02:52)".

To fix:

• Rebase onto staging and regenerate DOCUMENTARY_SKILLS bodies/descriptions from the current skills/build-graph, review-delta and review-pr SKILL.md (staging 1e64982); as submitted the PR's own sync test fails on staging.
• Fix the 22 ruff E501 violations in code_review_graph/skills.py (PR-added lines; wrap the long string literals to <=100 chars).
• Test for render_documentary_skill_markdown with an unknown slug (KeyError or a clearer error).
• Test for the argument_hint-absent branch of render_documentary_skill_markdown.

Merge origin/staging into your branch first to reproduce it.

PRs now target staging, not main. Yours was retargeted already, so nothing to do there.

[github-actions]

code-review-graph review

Overall risk: 0.40 (MEDIUM) — 5 changed function(s)/class(es), 1 affected flow(s), 0 test gap(s)

Risk-scored changes

Risk	Level	Symbol	Location	Tested
0.40	medium	code_review_graph/skills.py::render_documentary_skill_markdown	code_review_graph/skills.py:1513	yes
0.35	low	code_review_graph/skills.py::generate_documentary_skills	code_review_graph/skills.py:1533	yes
0.20	low	tests/test_document_skills.py::test_render_documentary_skill_markdown_contains_expected_sections	tests/test_document_skills.py:12	(test)
0.20	low	tests/test_document_skills.py::test_checked_in_documentary_skills_match_generator	tests/test_document_skills.py:47	(test)
0.15	low	tests/test_document_skills.py::test_generate_documentary_skills_writes_three_skill_dirs	tests/test_document_skills.py:32	(test)

Affected execution flows

• install_hermes_skills — criticality 0.26, 3 node(s) across 1 file(s)

Token savings: this graph-backed report used ~30,753 fewer tokens (~96%) than reading every changed file in full (estimated, chars/4 approximation).

---

Powered by code-review-graph — local-first analysis; no code leaves the CI runner.

[tirth8205]

Thank you for this PR.

I tested your branch with the latest staging, the new tests and ruff.

Before I can merge it, please fix these:

1. The new test test_checked_in_documentary_skills_match_generator fails. The text in DOCUMENTARY_SKILLS is older than the files in skills/. See code_review_graph/skills.py:1395. All three skills differ: build-graph, review-delta and review-pr. For example, the generator output does not have get_minimal_context_tool. Please copy the current skills/*/SKILL.md text into it.
2. Ruff finds 22 lines longer than 100 characters (E501). They are in code_review_graph/skills.py:1397 to 1482. Please wrap these lines.
3. Nothing calls generate_documentary_skills yet. Please call it from a script or command, or remove it.

The first two problems make the lint and test (3.13) CI checks fail.

After you push the fix, I will test it again.

[Copilot]

🟡 Changes recommended

The new regression test fails because all three checked-in skill files differ from the generated output, requiring the templates and skill files to be synchronized.

1 open issue

• Synchronize checked-in skills with generated template output · New

What changed in this PR

This PR centralizes the definitions of three documentation-driven skills in code_review_graph/skills.py, with generation helpers and regression checks intended to keep checked-in Markdown synchronized.

Changes:

• Defines templates for build-graph, review-delta, and review-pr.
• Adds Markdown rendering and skill-directory generation helpers.
• Adds tests for expected content, generated files, and checked-in file parity.

File	Description
tests/​test_document_skills.py	Adds rendering, generation, and synchronization tests.
code_review_graph/​skills.py	Adds shared skill definitions and Markdown generation helpers.

_{Review effort: Balanced}