agentleFS
meyverickmeyverick/agentic/skills/check/SKILL.md

check

Skill0 starsChanged 13 days agoISCView on GitHub

Execute and interpret local workspace and submodule check-gates to verify project integrity before push. Use when verifying code changes, running pre-flight gates, diagnosing gate failures or submodule pointer desync, or confirming clean-clone buildability. Do NOT use when only reading or exploring code without modifications, or when planning changes before implementation.

  • Reads credentials

80 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use daily-arxiv, exp-eval and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • LambenthanSkill

    daily-arxiv

    每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新

    Also in 5 projects that use this

  • LambenthanSkill

    exp-eval

    实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges

    Also in 5 projects that use this

  • JuliusBrusseeSkill

    backprop

    Bug → spec protocol. When a bug is found or a test fails, trace the cause, decide whether a new §V invariant would catch recurrence, append to §B. This is the one non-obvious thing SDD does that plan-then-execute doesn't. Triggers on test failure, bug report, post-mortem, or explicit user ask.

    Also in 4 projects that use this

  • skyllwtSkill

    exp-pilot-eval

    Pilot result evaluation — read pilot results, apply success criteria, update idea page (pilot_result, failure_reason if failed), generate PILOT_VERDICT_REPORT. Called by /ideate Phase 5 after /exp-pilot-run.

    Also in 4 projects that use this

  • skyllwtSkill

    exp-pilot-run

    Pilot experiment execution — read Pilot Spec YAML, write pilot code, run experiment(Confirm with the user before operation and require the applicant to conduct manual inspection), return results. Called by /ideate Phase 5. Does NOT modify wiki pages or judge pass/fail.

    Also in 4 projects that use this

  • +6 more from Lambenthan
  • +1 more from skyllwt

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 3 are distinctive, and 6 are very specific to this project.

  • 0

    Common

    Most of the 69 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use when verifying code changes, running pre-flight gates, diagnosing gate failures or submodule pointer desync, or confirming clean-clone …

    line 5
  2. Very specific:

    Do NOT use when only reading or exploring code without modifications, or when planning changes before implementation.

    line 6
  3. Distinctive:

    Only reading, searching, or exploring the codebase without making mutations.

    line 27
  4. Very specific:

    Always execute the unflagged gate before marking any code-modifying task [Completed]:

    line 58
  5. Very specific:

    When bun scripts/check.ts exits non-zero, identify the failure class and apply its self-healing recipe:

    line 77
  6. Distinctive:

    Do not halt or prompt the user for routine gate failures:

    line 94
  7. Very specific:

    Never bypass failures with --no-verify: If the gate fails, the change is incomplete.

    line 95
  8. Very specific:

    Submodule pointer bumps: Always check git submodule status after committing inside a submodule.

    line 97
  9. Very specific:

    Re-verify after healing: Always re-execute bun scripts/check.ts after applying a fix to guarantee exit code 0.

    line 98

“Common” is measured: kept by at least 6 in 10 of the 69 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Check Gate Handbook

A specialized reference manual and textbook instructing AI agents on how to execute local workspace check-gates, observe the two-speed verification protocol, interpret gate failure diagnostics, and apply autonomous self-healing recipes.

Activation Boundary

Consult this textbook when:

  • Executing bun scripts/check.ts during the Pre-response self-audit (./AGENTS.md Chapters 7 & 1).
  • Diagnosing why a local check gate, clean-clone build, or submodule pointer verification failed.
  • Confirming that all subprojects and orchestrators pass before reporting task completion.

Do NOT consult when:

  • Only reading, searching, or exploring the codebase without making mutations.
  • Authoring plans, proposals, or designs prior to implementation.

The Two-Speed Gate Protocol

Local verification operates at two distinct speeds to balance development velocity with release hermeticity:

Fast Loop (--quick)                     Full Pre-Flight Gate (Default)
Speed: 3 - 10 seconds                   Speed: 20 - 45 seconds
When: Active development iteration      When: Task completion / Pre-push
Runs:                                   Runs:
  [1] Code format verification            [1] Code format verification
  [2] Strict linter execution             [2] Strict linter execution
  [3] Fast unit/integration tests         [3] Fast unit/integration tests
  [4] Static cross-boundary contracts     [4] Static cross-boundary contracts
  [5] Tracked source & assets            [5] Tracked source & assets
                                          [6] Hermetic clean-clone sandbox
                                          [7] Runtime smoke / browser test

1. The Fast Loop (--quick)

Pass --quick as the first argument when validating intermediate code edits:

bun scripts/check.ts --quick
# Or for a specific submodule:
bun <submodule-path>/scripts/check.ts --quick

This executes all native linters, compilers, and static asset checks, but skips the clean-clone build and browser smoke tests, returning actionable feedback in seconds.

2. The Full Pre-Flight Gate

Always execute the unflagged gate before marking any code-modifying task [Completed]:

bun scripts/check.ts

This runs the full 6-slot pipeline, including cloning the committed tree into an isolated temporary directory to prove that uncommitted local files are not masking build breakages.

Workspace vs Submodule Topology

A check-gated repository organizes gates in a 2-tier hierarchy:

  1. Workspace Root (bun scripts/check.ts):
    • Checks orchestrator invariants: submodule pointer freshness (git submodule status) and tracked credential leaks (git ls-files).
    • Delegates to each registered submodule's gate: <module>/scripts/check.ts --quick.
  2. Codebase Submodule (bun <module>/scripts/check.ts):
    • Executes the component's native toolchain matrix (deduced autonomously via package manifests).
    • Validates compile-time asset tracking, the public funding ledger (.github/FUNDING.yml + DEPENDENCIES.md parity, offline), and performs the clean-clone test. The full (non---quick) lane adds funding-gate.ts --check --verify-links so every funding entry resolves (404/5xx fail, 403/429 warn as anti-bot walls). Skip the gate when gh repo view --json isPrivate -q .isPrivate prints true. Offline contract tests: bun scripts/funding-gate.test.ts.

Diagnostic Matrix: The 10 Failure Classes

When bun scripts/check.ts exits non-zero, identify the failure class and apply its self-healing recipe:

ClassObserved DiagnosticRoot CauseAutonomous Healing Action
1. Pointer Stalegit submodule status shows +<sha>Submodule committed, but root gitlink was not updatedIn workspace root: git add <submodule> && git commit -m "chore: bump <submodule> pointer"
2. Pointer Uninitializedgit submodule status shows -<sha>Submodule clone is missing or uninitializedRun git submodule update --init --recursive <submodule>
3. Credential Leakedtracked: .env / *.key / id_rsaSecret-shaped file added to Git indexgit rm --cached <file>, ensure pattern is in .gitignore, and re-run check
4. Unindexed AssetX unindexed: <path> (Slot 4)New or regenerated source/compile-time file not yet staged (include_str!, include_bytes!, or //go:embed targets included)Stage the missing asset: git add <path> and re-run check
5. Clean-Clone FailureBuild passes locally, fails in $tmp/cloneCode depends on an unstaged file, or a required file sits in a deny-listed path so git neither tracks nor lists itRun git status and git ls-files --others --exclude-standard, stage missing files (git add) or relocate them out of deny-listed paths
6. Cross-Boundary BreachSyntax error in embedded JS/DOM contractStatic frontend asset contains undeclared variable or broken handlerFix the client-side JavaScript/HTML contract in the source file
7. Native RegressionCompiler, clippy, eslint, or test failedCode broke typecheck, linter rules, or unit test assertionsRead compiler diagnostic, fix code or formatting, and re-run
8. Docs Contract BreachFAIL: missing Spotlight README.md / README.md contains technical code/terminal blocks / missing or untracked ./wiki/*.mdMissing README.md, fenced code block inside README.md, or missing/untracked wiki/*.md filesLoad .agents/skills/okf-docs/SKILL.md, move technical blocks from README.md into ./wiki/, stage git add README.md wiki/, and re-run check
10. LOC Ceilingloc-lane: <file>: N lines exceeds the 500-line ceilingA module the change touched grew past 500 physical linesProve behaviour green, extract without behaviour change, re-prove; revert the extraction if the second proof fails (see references/diagnostic-matrix.md)
9. Funding Ledger Stalefunding-gate: FAIL .github/FUNDING.yml: stale, manifests changedA dependency was added, removed, or bumped in a package manifest and the generated ledger was not regeneratedRun bun scripts/funding-gate.ts --write <repo-dir>, stage .github/FUNDING.yml DEPENDENCIES.md, and re-run check. Never hand-edit either file

Autonomous Self-Healing Playbook

Do not halt or prompt the user for routine gate failures:

  1. Never bypass failures with --no-verify: If the gate fails, the change is incomplete.
  2. Untracked build inputs: Compare the error output with git status. If a newly created asset was forgotten, stage it immediately.
  3. Submodule pointer bumps: Always check git submodule status after committing inside a submodule. If a + appears, commit the gitlink in the parent orchestrator before finishing.
  4. Re-verify after healing: Always re-execute bun scripts/check.ts after applying a fix to guarantee exit code 0.

Reference Depth

  • Script Anatomy & 6-Slot Skeleton: Detailed canonical bash patterns for orchestrators and submodules.
  • Failure Diagnostic Matrix: Deep-dive failure scenarios and step-by-step remediation commands.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.