Skills
96,770 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Turn a formal SPEC into tracer-bullet vertical slices with blocking edges, then publish them as single-file-per-ticket issues on the local markdown tracker. Use only when the user explicitly invokes $project-tickets; it verifies the SPEC handoff, drafts a quiz-able breakdown, and hands execution to the frontier via light-implement
Operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, skill sharing, worktree comments, and Orca's embedded browser through the `orca` CLI. Use when the user says "$orca-cli", "Orca worktree", "child worktree", "spawn codex/claude in a worktree", "read/wait/send Orca terminal", "handoff" / "handover" / "give this to another agent", "Orca browser", "orca artifacts", or "share skills". Prefer it over raw git worktree, ad hoc PTYs, or Computer Use when Orca state is involved. Use Computer Use only when a visible window needs GUI control that a CLI, filesystem, or API cannot do.
Audits whether AGENTS.md and/or CLAUDE.md in the current repository still reflect the project's reality (structure, commands, conventions, scope) and reports discrepancies without editing them. Use when the user runs /audit-agent-context, asks to verify or update agent context documentation, or suspects AGENTS.md/CLAUDE.md is outdated
Use when a session needs events from any polled system delivered as messages instead of polls, or when writing the source that polls it, so one daemon the session owns carries each poll's alerts in the peer envelope, keeps its notes in the log, and quotes nothing from the polled system.
Operate Veer-managed worktrees, folder contexts, terminals, repos, automations, artifacts, skill sharing, worktree comments, and Veer's embedded browser through the `veer` CLI. Use when the user says "$veer-cli", "Veer worktree", "child worktree", "spawn codex/claude in a worktree", "read/wait/send Veer terminal", "handoff" / "handover" / "give this to another agent", "Veer browser", "orca artifacts", or "share skills". Prefer it over raw git worktree, ad hoc PTYs, or Computer Use when Veer state is involved. Use Computer Use only when a visible window needs GUI control that a CLI, filesystem, or API cannot do.
Test how coding agents discover and use your product. Use when asked to run agent evaluations, test whether agents can use an SDK, API, CLI, or MCP server, find where agents struggle with onboarding, or measure whether docs and skills improve agent success. Runs real agent sessions and scores outcomes through the Gauge CLI. Also use for Gauge Agents tasks; not ordinary unit tests or AI-search visibility tracking.
Operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, skill sharing, worktree comments, and Orca's embedded browser through the `orca` CLI. Use when the user says "$orca-cli", "Orca worktree", "child worktree", "spawn codex/claude in a worktree", "read/wait/send Orca terminal", "handoff" / "handover" / "give this to another agent", "Orca browser", "orca artifacts", or "share skills". Prefer it over raw git worktree, ad hoc PTYs, or Computer Use when Orca state is involved. Use Computer Use only when a visible window needs GUI control that a CLI, filesystem, or API cannot do.
Draft 2–3 firegoat evals (firegoat.yaml) for the current repo: tests that a coding agent runs locally with `npx -y firegoat@latest eval`, each with a prompt, an end_goal, and script checks. Use when the user says "write firegoat evals", "draft evals", "set up firegoat", "firegoat init", or asks for local agent-experience (AX) evals of their SDK, CLI, API, or docs.
Coordinate a project from discovery through architecture, planning, implementation, verification, delivery and monitoring using an OWL knowledge mediator. Use for OntoLink, ontology-mediated project flow, sequential skills, compact context, knowledge relationships, recommendations, communication, warnings, progress, blockers, confidence and likelihood assessments.
Run the project's suggested-task ledger, the chips. /chips alone brings every open task up to date: it applies what the ledger helper says, settles stale and split tasks, cards what can start and reports the board. Use for /chips, /chips offer, start or drop, "what chips are open", "suggest a task for X", "update the chips", "clean up the chips", "which chips are stale", and whenever pstack would file a tracked issue, a follow-up, a reflect Backlog item or an out-of-band PR.
Subjects every non-trivial decision to a fresh-context adversarial review before it stands. Use when you want every assumption cross-examined before proceeding, when stress-testing a plan for hidden failure modes, when correctness matters more than speed, when working in unfamiliar code, when stakes are high (production auth, security-sensitive logic, a high-stakes migration, irreversible operations), or any time a confident output would be cheaper to verify now than to debug later.
Contextualizes an existing repository (brownfield) for AI-first work on it: verifies how to start and test it, creates the CodeGraph index, reconstructs the actual layer map, documents inherited decisions as ADRs, assembles the pattern catalog (canon vs. legacy), the debt map and verifiable build rules, all with file:line evidence, then pauses for human approval. Use when the ARCH role starts on an existing codebase without a PRD, or when asked to contextualize, onboard, document, map or understand a legacy/brownfield repository before working on it
Find the mistakes agents keep repeating in this repo and make each one impossible. Try architecture first, then types, then a lint whose error names the fix, then a test, and write docs last. Prove each check fails on a real past mistake. Repeat this each time the operator corrects you. Use for /correct.
Operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, skill sharing, worktree comments, and Orca's embedded browser through the `orca` CLI. Use when the user says "$orca-cli", "Orca worktree", "child worktree", "spawn codex/claude in a worktree", "read/wait/send Orca terminal", "handoff" / "handover" / "give this to another agent", "Orca browser", "orca artifacts", or "share skills". Prefer it over raw git worktree, ad hoc PTYs, or Computer Use when Orca state is involved. Use Computer Use only when a visible window needs GUI control that a CLI, filesystem, or API cannot do.
Use when a human wants to create, hire, activate, or set up a new AI employee from a plain-English description, e.g. wanting something that writes a weekly recap or replies to guest messages. Drives the Hiring Dozal's multi-turn setup wizard (the hiring chat), relays its button questions (show me first or just send it) to the human, and hires the finished Dozal. Use it for any new-employee request. Load operate-ai-employees first for auth and base-URL setup.
Use when a human wants to see, list, pause, reactivate, rename or inspect their existing AI employees (config, status active/paused/draft), or manage connected apps, schedules, handoffs, members and invitations, usage and bills, e.g. what employees do I have, pause the recap bot, how did last Friday go. Not for running one now (run-ai-employee), a task's output, approval or logs (monitor-and-review-ai-employee-work), or creating one (hire-ai-employee). Load operate-ai-employees first.
Use when a human asks whether an AI employee task is done, wants to see its output, or wants to approve or reject its work: is the recap ready, show me what it wrote, approve it, reject and shorten it, why did that run go wrong, show its logs, watch it live. Reads task status via phase_category, drives the approve/reject decision, and reads run logs, activity and live status. Load operate-ai-employees first for auth setup.
Use when a human asks you to operate their AI Employee platform account: hiring an employee, running one, checking on a task, approving or rejecting work, listing or pausing employees, or managing connections, schedules, handoffs, members, usage, bills, activity and run logs. Load this first to route between the four job skills. The umbrella skill: dze_ API key auth and its level, the production base URL, tenant resolution, the async model, and person-only steps.
Use the public `orca` CLI to operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, worktree comments, and the browser embedded inside the Orca app. Use when the user says "$orca-cli", "use orca cli", "Orca worktree", "child worktree", "cardStatus", "spawn codex/claude in a worktree", "read/wait/send Orca terminal", "terminal send", "full handoff", "handover", "give this to another agent", "another worktree", "Orca browser", "orca artifacts", "share HTML/Markdown", "public artifact link", or "control the browser inside Orca". Prefer this over raw `git worktree`, ad hoc PTYs, Playwright, or Computer Use when the task touches Orca-managed state. Use Computer Use for browser windows, webviews, or desktop UI outside Orca's embedded browser.
Find where an agent pipeline can swap frontier models for cheaper ones without losing quality. Drive the installed rightmodeler TypeScript CLI through its resumable pipeline, interpret its machine protocol, present per-family verdicts, and open a reviewed model-swap pull request when the evidence supports one. Use when the user wants to cut model spend, right-size models per task, benchmark cheaper model substitutions, or analyze an agentic pipeline for cost optimization.
Use when a human wants to run, trigger, start, or test-run an existing AI employee, e.g. run the recap bot now, kick off the guest responder, or run it once per topic from a list of inputs. Triggers the employee, can dry-run to list required inputs, and hands back task handles without blocking. Load operate-ai-employees first for auth; use monitor-and-review-ai-employee-work to check the result.
Write the observable contract for a User Story so the TESTER and DEV can work in parallel without seeing each other: assemble the context package (HU + story ADRs + Tech Design section), define inputs, per-case outputs, errors, and non-negotiable names with the source of each line, list decisions the HU left open, and stop for human approval. Use when the LEAD takes a user story to build, or when asked to write, draft, or version the contract/spec of a user story
Operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, skill sharing, worktree comments, and Orca's embedded browser through the `orca` CLI. Use when the user says "$orca-cli", "Orca worktree", "child worktree", "spawn codex/claude in a worktree", "read/wait/send Orca terminal", "handoff" / "handover" / "give this to another agent", "Orca browser", "orca artifacts", or "share skills". Prefer it over raw git worktree, ad hoc PTYs, or Computer Use when Orca state is involved. Use Computer Use only when a visible window needs GUI control that a CLI, filesystem, or API cannot do.
Audit the skill inventory for quality, coverage, and staleness. Detect overlapping, incomplete, or outdated skills and produce a health report. Activate when: skill audit, stocktake, skill inventory, check skills, which skills exist, audit skills, skill quality, skill coverage.
Execute one bounded, already-decided work item — code, document, configuration, research artifact, Skill, or generic project task — by inspecting relevant context, offering agent-config routing when materially useful, verifying locally, and handing the result to review-loop with the appropriate specialist reviewer
Coordinate root Herdr orchestrators across the local server and enabled saved machines. Ask for their current work and progress, relay messages verbatim, and route task execution through workers. Pair with the shunk031-herdr-orchestrate-workers skill for worker dispatch. Use when one session oversees multiple Herdr orchestrators; use the herdr skill for a single-pane or single-agent task.
For Japanese writing and editing, decide whether to organize from the structure, locally refine AI-like wording while preserving the original meaning and strength of claims, concisely summarize business or technical documents, or turn content into an outline, telegraphic prose, or technical slides, then choose the appropriate skill. Use as an entry point for choosing a writing approach when asked to “write something,” “make it easier to read,” or “fix the AI-like tone.” Do not use for ordinary conversation or code-only editing requests.
Send an independent judge that is not the author: a fresh evaluator subagent that checks finished work against pass criteria written before it was built (evaluator mode), or audits a batch against its source of truth (audit mode), with evidence for every verdict and never a fix. Use when the owner asks for an evaluator, reviewer, auditor or judge, or says 'send a judge', 'evaluate this', 'check my work with a fresh agent', 'get a second agent to review this', 'audit this folder', 'audit my registry', 'verify this against the source of truth', or 'LLM-as-judge'. Runs only when asked; it does not fire on its own at the end of a task. Not for checks a machine can run (tests, schema validation, a search that must return zero), not for hundreds of identical units, and not for fixing anything.
Group a scope of Issues into Waves that can be implemented concurrently without stepping on each other — estimate what each Issue touches, exclude the ones that need a human, and print launch lines. Use when the user has several prepared Issues and wants to know which can run in parallel unattended
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.