By jezweb
Audit a codebase so every comment, docstring, and doc file tells the truth about the code — for the next AI agent (or human) who reads it before changing anything. Finds stale, inaccurate, misleading, and missing comments; ships fixes through a confidence- and blast-radius-gated auto-fix vs review queue.
Audit a codebase so every comment, docstring, and doc file tells the truth about the code — for the next AI agent (or human) who reads it before changing anything.
A comment is a promise about what the code does. When it drifts, it becomes worse than no comment:
the next reader trusts it, builds a wrong mental model, and ships a mistake. And the next reader is
increasingly an AI coding agent — it reads a docstring that says a function returns null, writes a
dead null-guard against a function that actually throws, and skips the try/catch. The comment meant
to help is the thing that makes it wrong.
Honesty finds those broken promises and either corrects them or flags them. It's a Claude Code plugin: one skill that audits a codebase module by module and tells you exactly which comments lie, which are missing, and which can be fixed automatically versus which need your eye.
scope → bundle → audit (one agent per module) → structured findings →
disposition (auto-fix vs review) → apply + verify → review queue → roll
It checks every surface the code makes claims on — inline + block comments, docstrings/JSDoc, doc
files (README, CLAUDE.md/AGENTS.md, ARCHITECTURE.md, DATABASE_SCHEMA.md), and config-file
comments — against a tight rubric:
It's deliberately high-signal — an all-sound module returns nothing, and that's a good result.
Confidence alone doesn't make a fix safe to apply — a confident edit to a contract docstring or to
CLAUDE.md can mislead far beyond one line. Honesty auto-applies a finding only when it's high
confidence AND a mechanical factual correction AND local in blast radius AND not a doc/config file.
Everything else — rewrites, gap-fills, contract docstrings, every doc-file change — goes to a review
queue with the suggested fix attached. New codebases default to findings-first; auto-fix is a knob you
turn on once a calibration batch proves the signal.
ARCHITECTURE.md, the generator
regenerates it).One lane: are the claims the code makes about itself true, and present where they matter.
See docs/pattern.md for the methodology, the disposition gate, and the evidence
from real runs (including a real bug found hiding behind a technically-true comment). The skill itself
is in skills/honesty/SKILL.md; reusable audit prompt + findings schema are
in templates/.
Part of a family of Claude Code plugins: fixer, perfection, walkabout.
MIT © Jeremy Dawes / Jezweb
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub jezweb/honesty --plugin honestyScaffold Cloudflare Workers, Hono APIs, D1/Drizzle schemas, D1 migration workflows, full-stack Vite+Workers apps, and TanStack Start SSR dashboards.
Social media content creation — platform-formatted posts for LinkedIn, Facebook, Instagram, and Reddit with character limits, hashtag strategies, and image specs.
Manage Shopify stores: product creation, bulk imports, content pages, blog posts, and SEO metadata.
Local business SEO setup with JSON-LD schema, meta tags, robots.txt, and sitemap.
Generate colour palettes, favicon packages, custom SVG icon sets, and image processing (resize, convert, optimise).
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Build and maintain an LLM-curated personal knowledge base in your project — Andrej Karpathy's LLM Wiki pattern, designed to scale to thousands of pages without becoming a context bottleneck. Now with an optional compiled graph layer for typed, provenance-backed relationships.
Connect to Atlassian products including Jira and Confluence. Search and create issues, access documentation, manage sprints, and integrate your development workflow with Atlassian's collaboration tools.
AI-powered wiki generator for code repositories. Generates comprehensive, Mermaid-rich documentation with dark-mode VitePress sites, onboarding guides, deep research, and source citations. Inspired by OpenDeepWiki and deepwiki-open.
20 modular skills for idiomatic Go — each under 225 lines, backed by 48 reference files, 8 automation scripts (all with --json, --limit, --force), and 4 asset templates. Covers error handling, naming, testing, concurrency, interfaces, generics, documentation, logging, performance, and more. Activates automatically with progressive disclosure and conditional cross-references.
Claude + Obsidian knowledge companion. Sets up a persistent, compounding wiki vault (Karpathy's LLM Wiki pattern). v1.7 "Compound Vault" + v1.8 methodology modes close 5 of 5 priority gaps from the May 2026 compass artifact. Ships: substrate alignment with kepano/obsidian-skills, default Obsidian CLI transport, hybrid retrieval (contextual prefix + BM25 + cosine rerank per Anthropic's Sept 2024 research), per-file advisory locking for multi-writer safety, pre-commit verifier agent, AND methodology modes (LYT / PARA / Zettelkasten / Generic) for first-class organizational support no other Claude+Obsidian competitor offers. v1.7.x audit closure: every BLOCKER + HIGH + MEDIUM + LOW finding from the v1.7.0 audit is CLOSED or DEFERRED-with-rationale. Optional DragonScale Memory extension (log folds, deterministic addresses, semantic tiling lint, boundary-first autoresearch).