lint
Skill2 starsChanged 7 months agoMITView on GitHub
Health check for the brain. Detects orphan pages, broken wiki-links, stale claims, contradictions, and index hygiene issues. Reports structured findings; does not auto-fix unless run with --fix. Karpathy's 3rd core operation alongside /ingest and /query.
80 other projects use a version of this.
Adopted by
Most recently taken up by
- PMDevSolutions/Flaviantheir own version, first seen Nov 8, 2026
- aiagentkit-labs/agentic-dev-kittheir own version, first seen Nov 8, 2026
- levi-qiao/obsidian-llm-wikitheir own version, first seen Nov 8, 2026
- elastic/docs-buildertheir own version, first seen Nov 4, 2026
- DataDog/dd-trace-pytheir own version, first seen Nov 1, 2026
- PMDevSolutions/Aureliustheir own version, first seen Nov 1, 2026
Used with
Projects with it also use coverage, save and 6 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.
PydanticSkillcoverage
Fetch coverage diff from Codecov for the current branch or a specific PR. Shows uncovered lines, patch coverage, and overall coverage change.
Also in 4 projects that use this
seongjaeryuSkillsave
Complete and save your current thinking to Bindly
Also in 5 projects that use this
BugRogerSkillproject
Project lifecycle management. Dispatches to Ada.
Also in 4 projects that use this
AnthropicSkilldigest
Generate a daily or weekly digest of activity across all connected sources. Use when catching up after time away, starting the day and wanting a summary of mentions and action items, or reviewing a week's decisions and document updates grouped by project.
Also in 3 projects that use this
AWSSkillissue
Implement a GitHub issue end-to-end — branch, code, test, docs, and commit.
Also in 4 projects that use this
agenticnotetakingSkillupgrade
Apply plugin knowledge base updates to an existing generated system. Consults the Ars Contexta research graph for methodology improvements, proposes skill upgrades with research justification. Never auto-implements. Triggers on "/upgrade", "upgrade skills", "check for improvements", "update methodology".
Also in 3 projects that use this
Other versions80 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 80
- ray-project/ray★ 44ksame name
- FalkorDB/FalkorDB★ 10ksame name
- FastLED/FastLED★ 7.5ksame name
- fleetdm/fleet★ 7.2ksame name
- aaddrick/claude-desktop-debian★ 5.5ksame name
- cloudposse/atmos★ 1.4ksame name
- ethereum/execution-specs★ 1.2ksame name
- adobe/S3Mock★ 1.2ksame name
- maxGraph/maxGraph★ 1.2ksame name
- jason-effi-lab/karpathy-llm-wiki-vault★ 718same name
70 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository12 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- absorbSkill
- captureSkill
- deep-syncSkill
- extract-entitiesSkill
- helpSkill
- ingestSkill
- new-pageSkill
- orientSkill
- publish-contextSkill
- querySkill
- quick-syncSkill
The file
We split this skill into the rules it gives an agent
6 rules. 0 are common, 2 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 67 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 24
No other wiki page links to it via [[name]] markers
- Very specific: line 42
If count ≥ 5 AND most recent is > 90 days old → flag as STALE_CANDIDATE
- Distinctive: line 49
Look for negation patterns: one section says "X is Y", another says "X is not Y" or "X is Z"
- Very specific: line 50
Look for decision chains: a decision marked status: decided with a later decision touching the same topic but no ## Superseded by: link
- Distinctive: line 60
Every page should have a row in the index table
- Very specific: line 67
Read wiki/index.md. Verify every page under wiki/ appears in its ## Page catalog section.
“Common” is measured: kept by at least 6 in 10 of the 67 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.
Files
Audit the brain for rot. Run weekly, after /deep-sync, or on demand when something feels off.
Usage
/lint # report only
/lint --fix # attempt auto-fixes where unambiguous (orphan index entries, broken wiki-links with single candidate)
/lint --scope wiki # only check wiki/, skip decisions/Checks
1. Orphan pages — wiki pages no source has touched
A page is orphaned if:
- It exists under
wiki/ordecisions/ - Its
## Sources that shaped this pagetable is empty or absent - No other wiki page links to it via
[[name]]markers - Its
touched_byfrontmatter field is[]or missing
Orphans are not automatically wrong — you might be documenting something the brain hasn't ingested yet. But an unlinked, unsourced page six months after creation is usually dead weight.
2. Broken wiki-links — [[name]] with no target
Grep all markdown files for [[...]] patterns. For each match, check if:
- A file exists at the referenced path (try:
wiki/**/<name>.md,decisions/<name>.md, and the direct path if it has slashes) - OR it's a
[[Source: basename]]marker pointing to aninbox/orraw/file (those can be gitignored, so only flag if the basename clearly won't resolve)
Report each broken link with: source file, line number, the broken [[...]] text, and candidates (if any close matches exist).
3. Stale claims — pages enriched ≥5 times, none recent
For each wiki/ page:
- Count rows in
## Sources that shaped this page - Find the most recent row's date
- If count ≥ 5 AND most recent is > 90 days old → flag as
STALE_CANDIDATE
These are high-value pages that haven't absorbed anything new — worth manually reviewing.
4. Contradictions — conflicting facts
For each wiki/ page with multiple enrichment sections:
- Look for negation patterns: one section says "X is Y", another says "X is not Y" or "X is Z"
- Look for decision chains: a decision marked
status: decidedwith a later decision touching the same topic but no## Superseded by:link - Flag for human review — never auto-resolve
This check is necessarily heuristic. False positives are OK; false negatives silently rot the brain.
5. Index hygiene — pages missing from folder index.md
For each folder in wiki/:
- Read
wiki/<folder>/index.md - List all non-
index.mdpages in that folder - Every page should have a row in the index table
- Flag missing rows
With --fix, auto-append missing rows using the page's frontmatter name + created date.
6. Global wiki/index.md completeness
Read wiki/index.md. Verify every page under wiki/ appears in its ## Page catalog section. Flag missing entries. With --fix, append the missing rows using each page's frontmatter (kind, created) and first body paragraph (one-line summary).
Output format
/lint report — 2026-05-01
ORPHANS (2)
wiki/concepts/cold-start.md — created 2026-03-15, no sources, not linked
wiki/people/old-contact.md — created 2025-11-02, 1 source 5 months ago, not linked
BROKEN LINKS (3)
wiki/people/me.md:42 [[Priya-Sharma]] — no target; close match: wiki/people/Priya-S.md
decisions/2026-03-12-... [[pricing-strategy]] — no target; no close matches
wiki/concepts/tool-vs-platform.md:18 [[2026-04-18-offsite]] — target missing
STALE CANDIDATES (1)
wiki/concepts/strategy.md — 7 sources, last absorbed 2026-01-10 (112 days)
CONTRADICTIONS (1)
wiki/people/ravi.md
2026-02-01: "Ravi prefers seat-based pricing" [[2026-02-01-pricing-call]]
2026-04-18: "Ravi pushed for platform-shaped product" [[2026-04-18-offsite]]
— no explicit reconciliation. Review.
INDEX HYGIENE (2)
wiki/partners/index.md missing: meridian.md
wiki/concepts/index.md missing: first-five-minutes.md
GLOBAL INDEX (0)
(clean)
SUMMARY: 9 findings, 2 auto-fixable (re-run with --fix)With --fix:
/lint --fix — 2026-05-01
AUTO-FIXED (2)
wiki/partners/index.md + meridian.md row
wiki/concepts/index.md + first-five-minutes.md row
STILL MANUAL (7)
[orphans, broken links, stale, contradictions per above]Log
Append to log.md:
## [YYYY-MM-DD] lint
Findings: N (orphans: O, broken: B, stale: S, contradictions: C, index: I)
Auto-fixed: FFailure modes
- Empty brain (no wiki pages yet) → print "Brain has no wiki pages to lint. Run /ingest first."
- Huge brain (>500 pages) → lint runs in scope-limited mode by default. Warn if total pages > 500 and suggest
--scope <folder>. - Broken link with multiple candidates → under
--fix, do NOT auto-rewrite. Report and let the user choose.
Non-goals
/lintdoes NOT rewrite prose. It flags, it hygiene-fixes, it reports. Semantic merging of contradictions is a human call./lintis not RAG or semantic search. It uses grep + file structure./lintis not CI. It is an on-demand or weekly-cadence tool.
Relationship to other skills
/deep-synccalls/lintinternally before the retro step/new-pageshould fail-safe into the folder index, soindex hygienefindings should be rare/ingestcompounding gate already prevents the worst orphan-creation cases
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
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.

