Herdr LLM Lint — AI skill for Claude Code
Lints CLAUDE.md, AGENTS.md, and other agent instruction files for stale paths, commands, facts, and drift.
How to install Herdr LLM Lint
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open shindakun/herdr-llm-lint and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Herdr LLM Lint does
Lints CLAUDE.md, AGENTS.md, and other agent instruction files for stale paths, commands, facts, and drift.
Alternatives in AI
- Wiki Lint — Health-check the LLM Wiki for issues 2.3k ★
- Part 4 — Generate AGENTS.md And AI Agent Configuration Files — I'll help you create the instruction files that will guide your AI coding assistant to build your MVP 2.1k ★
- Fact Checker — Verify claims and facts with sources 653 ★
README
herdr-llm-lint
Lints CLAUDE.md, AGENTS.md, and other agent instruction files for stale paths, commands, facts, and drift. Runs as a CLI and as a [Herdr](https://herdr.dev) plugin. Rust, one binary. Findings look like compiler output and carry a file and line. Covers `CLAUDE.md`, `AGENTS.md`, `GEMINI.md`, `.cursorrules`, `.cursor/rules/*.mdc`, `.github/copilot-instructions.md`, and nested `CLAUDE.md` and `AGENTS.md`.
Instruction files rot. A path moves, a Makefile target is renamed, a rule gets pasted into three files and edited in one. The agent reads the stale version every session and nobody notices until it does something wrong.
The design is in [docs/PLAN.md](docs/PLAN.md).
Install
herdr plugin install shindakun/herdr-llm-lint
Needs `cargo`; the install step builds the binary. Linux and macOS.
For local development, link the checkout instead:
cargo build --release
herdr plugin link /path/to/herdr-llm-lint
Use it
herdr-llm-lint lint # the nearest git root, text output
herdr-llm-lint lint path/to/repo
herdr-llm-lint lint --format json
herdr-llm-lint lint --fail-on error # exit 1 only on errors
Output:
CLAUDE.md:7: ref-path: `internal/httpx/` does not exist
CLAUDE.md:9: ref-target: `make check` is not a Makefile target (have: build lint test)
CLAUDE.md:11: ref-command: `frobnicate` is not on PATH
CLAUDE.md: size-bytes: 9400 bytes, budget is 8192
Exit 1 when any finding is at or above `--fail-on` (default `warn`, so any finding).
`--fix` applies the safe subset, prints each action to stderr, then lints again and prints what remains:
- long lines reported by
shape-linesorshape-bodyare wrapped atline_chars, with a list item's continuation indented under its text and backtick spans kept whole; headings are left alone - a line reported by
content-dupis dropped when it is an exact repeat (after trimming) of an earlier line; near-duplicates stay - in each directory, an
Related Skills
Mobbin Claude Rodeo
Claude Code skill: find real-app UI references on Mobbin and build to them faithfully — no LLM drift, no stale
AI Safe Driver
AI agent drift detection and recovery for Claude Code and OpenAI Codex—stop repeated tool-call, instruction-fo
Agent Doc Lint
Your CLAUDE.md / AGENTS.md is too long? This skill helps your AI agent analyze it, split it into skills, and k
Bead Sweep
Scan open beads for stale trackers already implemented — deterministic checks then LLM verification.
Agent Skills Deep Dive
A Python implementation of the Claude Agent Skills system — a sophisticated prompt-based meta-tool architectur
Agent Config Harness
Agent instruction files, scored and regression-tested like code. CLAUDE.md / AGENTS.md / skills managed via a
Related Agents
Docs Drift Review
Given a session's commits or diff, finds the docs that the change made wrong — stale paths and symbols, checkl
Review Workflow Consistency
Reviews .claude/ workflow machinery for cross-file consistency: phantom references, mismatched threshold table
Light Md DriftUpdate
Scan a Delphi project's markdown docs (CLAUDE.md, README.md, docs/.md) for drift against the current code — st