MCP servers

Local environment doctor: runtime, tools, env files, and services, pass or fail.

Catalogue every error a codebase throws, and whether each message is documented.

Snapshot diff for agent and tool outputs: changed cases, score regressions, pass/fail verdict.

Check whether an exit IP works for AI services: ASN, network type, risk, AI availability.

Shared failure knowledge for agents: check known tool failures, report failures and outcomes.

Provably fair raffle draws: winners picked from public drand randomness, verifiable by anyone.

Mine supplied closed issues for repeated questions; draft an FAQ with counts and links.

Fetch and parse RSS or Atom feeds from a URL.

Flaky-test evidence tracker: per-test pass/fail history, flakiness rates, and a quarantine list.

Flashcard decks and spaced repetition via AnkiConnect: study, decks, notes, and stats.

Least-privilege auditor for GitHub Actions workflows: permissions, triggers, secrets.

Assess dependency staleness: behind level per dependency, grouped by risk from abandoned to current.

Find untested functions: link tests to code, score coverage, record proven runs locally.

AI-crawler readiness checks and fixes: GEO score, robots.txt, schema.org, FAQ and llms.txt.

Plan stale and merged branch cleanup: keep / review / delete per branch, with reasons.

Build a project glossary from supplied code and doc terms, merged, with source links.

Extract a repo's guardrails — DCO, CLA, changesets, sign-off — into one checkable report.

Build a cold-start handoff brief for the next agent from supplied commits, PRs, and TODOs.

Browser automation: navigate, click, fill, screenshot and evaluate in a headless browser.

Host, Docker, Compose, SSH, logs and ZFS operations via two action-router tools.

Detect circular imports, report the exact cycle per component, and suggest a break point.

Scan what AI agents read for prompt injection: hidden unicode, comment instructions, exfil, lures.

Install-script inventory for a dependency tree: what runs when you install.

Dad jokes on demand: fetch a random joke, search by keyword, or look one up by ID.

Compare two JSON documents deeply, ignoring key and array order, with every difference pathed.