How do I install the Lint skill?

Run npx skillmds@latest add flyaibox/lint in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

What does the Lint skill do?

Audit wiki structure, evidence and knowledge health. It is listed under AI & ML on SkillMD.

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Lint?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Yes. Installing skills from SkillMD is free. This skill is licensed under MIT.

FlyAIBox (@flyaibox) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Lint by flyaibox — Agent Skill

Name: Lint
Author: FlyAIBox

File contents

Lint Skill

Audit structural integrity and the quality of saved knowledge. Distinguish computed findings from semantic conclusions that require reading and judgment.

When to Use

The user asks about wiki health, statistics, broken links, missing pages or stale claims.

Prerequisites

Read purpose and schema; use the host's file and execution tools. In Hermes use read_file, search_files, terminal and patch as appropriate.

How to Run

Use ../../scripts/wiki_tool.py with status, graph or sync in preview mode. See the helper API.

Quick Reference

Structural report → evidence audit → severity and suggested fixes → authorized repairs.

Procedure

Use status for page/source counts, metadata, index, original hashes and pending reviews. Use graph for communities, hubs, zero-inbound orphans, wanted pages and ambiguous links. Keep requested JSON intact; explain findings in prose otherwise.
Inspect orphan and wanted pages in context. A wanted page may be an intentional research gap; an orphan may simply need a useful relationship. Do not fabricate links.
Read low-confidence or contested pages, high-impact claims and stale evidence. Check citations, taxonomy consistency, contradictions, outdated versions and scope. Structural status cannot prove factual correctness or semantic freshness.
Inspect pending cognition reviews and scheduling state. A local config saying active must be checked against the native scheduler before claiming a job is healthy.
If asked for repairs, apply reversible structural fixes supported by evidence, checkpoint before knowledge edits, rebuild index, log and track changes. Preserve disputed claims and record evidence; do not settle them merely to obtain a clean report.
For a pure audit, report issues without editing pages or committing tracking state.

Pitfalls

Do not count indexes, raw views, images or archives as knowledge nodes. Never repair a source hash mismatch by silently accepting a new hash; investigate the changed original.

Verification

Recheck the affected findings after repairs. Keep unresolved semantic concerns visible even when all structural checks pass. Follow cognition for findings that belong in future briefings.

---name: lintdescription: Audit wiki structure, evidence and knowledge health.license: MIT---# Lint SkillAudit structural integrity and the quality of saved knowledge. Distinguish computedfindings from semantic conclusions that require reading and judgment.## When to UseThe user asks about wiki health, statistics, broken links, missing pages or stale claims.## PrerequisitesRead purpose and schema; use the host's file and execution tools. In Hermes use`read_file`, `search_files`, `terminal` and `patch` as appropriate.## How to RunUse `../../scripts/wiki_tool.py` with `status`, `graph` or `sync` in preview mode.See the [helper API](../../references/helper-api.md).## Quick ReferenceStructural report → evidence audit → severity and suggested fixes → authorized repairs.## Procedure1. Use `status` for page/source counts, metadata, index, original hashes and pending reviews. Use `graph` for communities, hubs, zero-inbound orphans, wanted pages and ambiguous links. Keep requested JSON intact; explain findings in prose otherwise.2. Inspect orphan and wanted pages in context. A wanted page may be an intentional research gap; an orphan may simply need a useful relationship. Do not fabricate links.3. Read low-confidence or contested pages, high-impact claims and stale evidence. Check citations, taxonomy consistency, contradictions, outdated versions and scope. Structural status cannot prove factual correctness or semantic freshness.4. Inspect pending cognition reviews and scheduling state. A local config saying active must be checked against the native scheduler before claiming a job is healthy.5. If asked for repairs, apply reversible structural fixes supported by evidence, checkpoint before knowledge edits, rebuild index, log and track changes. Preserve disputed claims and record evidence; do not settle them merely to obtain a clean report.6. For a pure audit, report issues without editing pages or committing tracking state.## PitfallsDo not count indexes, raw views, images or archives as knowledge nodes. Never repair asource hash mismatch by silently accepting a new hash; investigate the changed original.## VerificationRecheck the affected findings after repairs. Keep unresolved semantic concerns visibleeven when all structural checks pass. Follow [cognition](../../references/cognition.md)for findings that belong in future briefings.

Frequently asked questions

Lint