skills/maintain-project-constraints/SKILL.md
Reads the latest codebase and docs to update CLAUDE.md and AGENTS.md at both the project root and per sub-module (crate, package, etc.). CLAUDE.md targets Claude Code; AGENTS.md targets non-Claude agents (Codex, etc.).
npx skillsauth add laitszkin/apollo-toolkit maintain-project-constraintsInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Maintain project constraint files at both root and sub-module levels so that every agent modifying any part of the codebase has clear boundaries to follow.
Deliverables:
CLAUDE.md + AGENTS.md (three-section format: Commands / Business Goals / Documentation Index)CLAUDE.md + AGENTS.md (description / file list / non-violable rules)CLAUDE.md and AGENTS.md updated and consistent with actual codeRead the repository to understand:
Look across the project for implicit and explicit prohibitions. Concrete sources:
Each prohibition must be specific and actionable (e.g. "never auto-create database migrations" not "handle databases carefully"). Include them in root-level AGENTS.md and CLAUDE.md under a Prohibitions section.
Identify every sub-module that needs its own constraint files. A directory qualifies if EITHER:
Cargo.toml, package.json, pyproject.toml, go.mod, pom.xml, build.gradle, etc.packages/, crates/, apps/, lib/, modules/, components/ AND has a clear boundary (own entry point, README, or standalone deployment unit)For each sub-module record: relative path and a brief description (extract from the manifest's description field or README; infer from code if neither exists).
Create or update CLAUDE.md and AGENTS.md in the project root using assets/templates/root-AGENTS.md and assets/templates/root-CLAUDE.md as format skeletons. Both use exactly three sections in this order:
package.json, Makefile, scripts, CI config). Before using any apltk command, run its --help first. Append a one-line purpose per command.docs/features/, docs/architecture/, docs/principles/, plus important root files (README.md, etc.). Also include each sub-module's constraint file: <module_name>/CLAUDE.md in CLAUDE.md, <module_name>/AGENTS.md in AGENTS.md — each with a one-line description.AGENTS.md vs CLAUDE.md split:
| Aspect | AGENTS.md | CLAUDE.md | |--------|-----------|-----------| | Audience | Non-Claude agents (Codex, etc.) | Claude Code | | Syntax | Agent-neutral, no platform-specific syntax | May use Claude-specific syntax | | Claude-only content | Must not include | May include hooks, scheduled tasks, cowork, MCP | | Common base | Three sections + Prohibitions | Same as AGENTS.md |
Both files must stay under 100 lines. Prioritise brevity over completeness when constrained.
For each sub-module discovered in Step 3, create or update CLAUDE.md and AGENTS.md in its directory using assets/templates/module-CLAUDE.md as the format skeleton.
Key points:
CLAUDE.md and AGENTS.md have identical content — the format is simple and carries no platform-specific informationMODULE FILE LIST covers all source files; exclude generated output, dist, node_modules, target, etc.RULES SHOULD NOT BE VIOLATED derives from: module boundary rules, dependency direction, internal conventions, and historical error patternsSelf-review before delivery:
CLAUDE.md and AGENTS.mdRULES SHOULD NOT BE VIOLATED entry can be traced back to code structure or project history — no invented rulesassets/templates/root-AGENTS.md — format skeleton for root-level AGENTS.mdassets/templates/root-CLAUDE.md — format skeleton for root-level CLAUDE.mdassets/templates/module-CLAUDE.md — format skeleton for per-module constraint filesdevelopment
Review a pull request — interactive PR selection via `gh`, 4-dimension code review (hallucinated code, architecture, performance, test validity), then post severity-graded comments with fix suggestions on the PR. Not for spec-based review — use `review` instead.
development
Read a user-specified PDF that marks the week's key financial events, deeply research each marked event with current sources, capture any additional breaking financial developments, and produce a concise Chinese-capable PDF briefing that explains what happened and why it matters.
documentation
Generate long-form videos (more than 10 minutes) by following user instructions and invoking related skills only when needed (`openai-text-to-image-storyboard`, `docs-to-voice`, `remotion-best-practices`). For text inputs, extract a complete long-form story arc, generate fresh storyboard images (no reuse of previously generated pictures), and render a 16:9 animated long-form video.
tools
協助完成自動化版本發佈。同步文檔、更新版本號、推送 tag 並建立 GitHub Release。