Cyber Audit
Read-only exposure audit of THIS MACHINE against a specific external advisory — a named CVE, malicious package, or supply-chain advisory — then write a report. Use when the user brings an advisory and asks "am I affected by X", "is this package/CVE on my system", "scan my machine for Y". Not for code/repo security review, and not for account/credential-breach response.
Agents Pair
Pair with a peer AI agent. Use when the user asks to pair with Codex or Claude, delegate a scoped slice, request peer review, run an adversarial challenge, mirror a long goal loop, or use live herdr pairing. Routes Claude Code<->Codex through the codex plugin, Codex<->Claude through headless `claude -p`, and live panes through `herdr-pair`. For paired UI work, use `impeccable`.
Ship It
Ship a PR end to end. Use when the user asks to ship, commit, push, open or update a PR, handle CI, handle reviewer comments, or get a branch ready for review. Splits changes into conventional commits, runs the quality gate, opens/updates the PR, and loops until CI and review are clean.
Delegate
Delegate work across models like a tech lead: plan and freeze specs, route building to Codex, deep reasoning to Opus subagents, taste-sensitive light work to Sonnet, then verify and synthesize. Use when the user says 'you're the lead', asks to orchestrate or delegate a task across models or subagents, is burning through the top model's usage or quota, or wants independent parallel takes on a high…
Check Logs
Check logs in an existing turborepo dev TUI inside herdr. Use when the user asks about an app's dev output, whether an app is running, or an error shown by a `bun dev` / `turbo run dev` pane. Read-only: navigate the existing TUI; never start, stop, or restart the server.
Review Pr Comments
Fetch PR review comments from bots and humans, triage as fix/false-positive/out-of-scope/needs-discussion/informational, apply valid same-PR fixes, reply to processed threads, push to the PR branch, and keep rechecking until the latest head is stably clean. Use when the user asks to review PR comments, check PR feedback, handle reviews, address comments, triage a PR, or fix PR comments.
Herdr Pair
Live Claude/Codex pairing inside herdr. Use when the user invokes `/herdr-pair` or asks for live pairing in herdr. Use when input begins with `[agent <name> -> <name> kind=<kind> sid=<...>]`; that is partner-agent traffic, so validate pane/session state before responding.
Goal Loop
Goal-driven improvement loop run by Claude + Codex as a pair: the human names one target and what 'better' means, and the agents iteratively test it for real (browser/computer-use, dev logs, runtime debugging), improve it — correctness, behavior, performance, UX, polish, whatever the goal is — and ship autonomous PRs, looping until the goal's quality bar and the peer both pass. Use when the user…
Debug Mode
Evidence debugging for unresolved bugs. Use when behavior is broken, flaky, environment-specific, or unexplained and the root cause is not proven. Build hypotheses, collect runtime evidence, and fix only after evidence identifies the cause.
Art Director
Art-direct a long design-exploration loop: you direct a batch on one axis, a peer generator (Codex) produces the mockups, you curate and wipe the generator, then redirect on a fresh axis until the gallery converges on a winning direction. Use when the user invokes /art-director, wants to explore many design directions or generate mockups for a surface over a long session, or wants to brainstorm o…
Cmux Pair
Legacy cmux-based pair programming bootstrap for Claude and Codex talking through cmux surfaces with an XML protocol. Keep this for historical reference and migration only. Prefer herdr-pair for current Claude/Codex collaboration; use this skill only when the user explicitly asks for the old cmux workflow.
Setup Help
Guided setup — walk through configuring or installing something one atomic step at a time, always showing what's left. Invoke by hand when you want the hand-held pace.