random-claude-skills
claudeThree skills for working with Claude Code: context-budgeted subagent delegation (fanout), driving your own logged-in Chrome (my-chrome), and running a task to completion without check-in pauses (autopilot).
3 skills
Run an agentic engineering team: team-lead routing by cost of being wrong, agent-ready task packets, verify-against-reality before done, and guard hooks (no git add -A, no gh pr merge, no force-push, worktree-confined writes, DoD stop gate).
Plugin agent-team from marketplace embenzekri in embenzekri/random-claude-skills. Commands read from the repo's manifest — 24 days ago.
This plugin runs code on your machine — its hooks fire automatically on events like tool use. The exact commands are listed below; read them before installing.
codebase-mapMap a big existing repository for agents, produce an impact map for a proposed change, and keep the map fresh. Use when the user says "map this codebase", "onboard agents to this repo", "codemap", "AGENTS.md", "impact map", "what does this change touch", "big repo", "refresh the map", or when team-l
review-laneIndependent, read-only review of one ticket's branch or diff before merge, ending in a JSON verdict (merge-ready, merge-with-follow-up, blocked) and a short human summary. Use when the user or team-lead says "review this ticket", "review the branch", "is it ready to merge", "review lane", "independe
task-packetWrite an agent-ready ticket (task packet) before any code is written, and set up the ticket marker when dispatching an implementer into a worktree. Use when the user says "write a ticket", "task packet", "split this into tasks", "dispatch an implementer", "brief an agent", "make this agent-ready", o
team-leadEntry point for running an agent team on a new app or a big existing codebase. Use when the user says "build this app with agents", "agent team", "run a team on this", "plan the work", "dispatch", "split this across agents", "big feature in a large repo", "migrate/refactor with agents", or hands ove
verify-against-realityThe pre-close proof every agent runs before saying "done". Use when asked to verify, "is it done", "prove it works", "before closing", "check it's live", "did the deploy work", or to meet the "definition of done"; also before any success claim, commit, PR or ticket close. Supersedes and extends supe
implementerWrites the code for ONE ticket inside its own git worktree. The lead dispatches it after writing the task packet, re-verifying its premise, creating the worktree and writing .agent-team/ticket.json. Default is one implementer at a time; a second only when touches are disjoint. Sonnet by default; the
integratorThe single merge owner. The lead dispatches it with a list of review-approved branches/PRs to rebase, gate, merge one at a time, and re-gate on main. Handles conflict-free chores (rebase, CI fixes, changelog assembly) on sonnet; the lead overrides to opus when resolving semantic conflicts or when AP
mapperRead-only codebase mapper. The lead dispatches it before any change in a repo it does not know, to produce a repo map, an impact map for a proposed change (callers, tests, contracts, config, deploy paths), or a mechanical sweep (grep-and-list, counts, renames inventory). Sweeps run on haiku; the lea
planner-architectTop-tier design role. The lead dispatches it for greenfield specs, ADRs, shared contracts (API, schema, constants, env names), ticket splits, migration plans, and for resolving API/migration/auth conflicts the integrator sends back. Also dispatched for the root cause of a bug that already beat one a
releaseShips merged work and proves it is running. The lead dispatches it after the integrator reports a green main, to build, verify locally, deploy, verify live (running revision equals main HEAD), run migrations in expand/backfill/switch/contract order, and record the result. Sonnet by default; the lead
reviewerFresh-context, read-only reviewer that never saw the implementation. The lead dispatches it on every PR/branch before merge with the diff and the ticket (not the lead's opinion), and it writes a JSON verdict file. Sonnet for ordinary review; the lead overrides to opus when the diff touches auth/secr
testerWrites and repairs tests for ONE ticket inside its own worktree, tests only. The lead dispatches it when a change needs characterisation tests before a refactor or migration, a failing test that reproduces a bug, or independent tests for an implementer's change. Never the implementer of the same tic
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/guard.py"
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/dod_gate.py"
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/dod_gate.py"
Three skills for working with Claude Code: context-budgeted subagent delegation (fanout), driving your own logged-in Chrome (my-chrome), and running a task to completion without check-in pauses (autopilot).
3 skills
Matt Pocock's agent skills for real engineering (grilling, spec/ticket flows, TDD, code review, domain modelling and more).
39 skills
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
15 skills · hooks
Persistent memory for Claude Code: the next session starts where the last one left off.
29 skills · 1 MCP server · hooks
Cross-session memory and token savings for coding agents.
6 skills · 1 agent · 1 MCP server · hooks
Up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your coding agent's context.
1 MCP server
Build powerful, stateful CLI interfaces for any GUI application using the cli-anything harness methodology.
5 commands
This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.