fitchmultz

Add integration/E2E tests to existing codebase using Design Docs

Execute materialized task files in autonomous execution mode

Execute from codebase-scoped analysis through optional ADR decisions to complete Design Doc approval

Investigate problem, verify findings, and derive solutions

Adjust an already-implemented UI in-session with verification against the design source

Execute materialized frontend task files in autonomous execution mode

Execute from repository evidence through applicable UI Spec and optional ADR decisions to complete frontend Design Doc approval

Create frontend work plan from design document and obtain plan approval

Create and render polished technical diagrams as editable SVG or D2 plus PNG/SVG: architecture, sequence, data-flow, dependency, lifecycle, and before/after visuals. Use when a user asks for a diagram or rendered system/process image. Do not use for…

Use for the propose-then-ship pipeline in pi: scan a repo, propose a ranked #1 recommendation, stop for the user's direction, then implement in a worktree and drive the PR through validation and CI to merge. Do not use for plain research, an already-decided…

Use this skill when the agent is about to claim work is complete, commit, open a PR, or say tests/lint/build pass. Verify the exact success claim with current evidence from diff cleanup, requirement checks, validation commands, runtime/manual checks, and…

Pi extension/package runtime and bundled-resource install/discovery: tools, commands/events, providers, TUI, SDK/RPC, release/publish, debugging, and performance. Do not use for Pi core, Agent Skill content authoring, prompt-only work, Crabbox/cbx, platform…

Dogfood or exploratory-QA browser-visible products through real user flows: web apps, local URLs, visual bug hunts, jank/flicker/streaming, and terminal/TUI redraw via VFR. Do not use for backend/unit-test-only checks, code review, trivial visual tweaks, or…

Review PRs, designs, plans, and product behavior that affect a human user's experience, especially agentic or long-running workflows. Evaluate usability, end-to-end task completion, recovery, visible progress, truthful outcomes, capability boundaries, scope…

Use this skill when the user asks to build or fix code test-first, practice red-green-refactor, add behavior/integration tests before implementation, or explicitly wants TDD. Do not use for ordinary final test runs, broad QA, post-hoc coverage padding, or…

Clean AI-generated slop from branch/PR diffs while preserving behavior: noisy comments, debug/scaffold leftovers, odd defensive checks, evidence-laundering casts, spurious try/catch, needless wrappers/nesting, test-table ceremony, and changed-hunk style…

Pi subagent orchestration: delegate to builtin/custom agents; run single, parallel, chain, async/background, forked-context, acceptance, worktree, intercom, status/control, or agent-management workflows. Do not use for Agent Skill maintenance, spawned child…

Coordinate local Pi sessions with pi-intercom: list peers, send updates, ask blocking questions, reply to inbound asks, use contact_supervisor, or handle pi-subagents escalations. Do not use for generic chat, remote/cross-machine messaging, unrelated repos,…

Use for macuse pi tools: inspect, QA, dogfood, or safely control local macOS native apps through Codex Computer Use while preserving focus. Do not use for browser DOM automation, generic pi extension work, raw MCP probes, or…

Write mcp_script JavaScript for discovering, inspecting, and calling MCP tools.