joshrotenberg

Use when deciding whether to dispatch a runner or a worker for a subtask. Describes the decision boundary, the failure mode when runner is used where worker is intended, and the safe composition pattern.

When a repo has no GitHub remote (or is not yet a git repo) -- use this to route survey, sweep, and dispatch off the GitHub-centric path. Detect with `git remote` empty; treat blank GitHub columns as expected, not an error; route work to commits + CLAUDE.md instead of issues + PRs; degrade a maintenance sweep to the local-only axes.

When the directive is "status check," "what's new across my projects," or "sweep maintenance" -- run a per-project, read-mostly, non-PR sweep. For each project under the owner-prefixed layout, gather stars/forks, open issues/PRs split into mine vs community vs bots, package downloads, release-due (unreleased conventional commits), a test run, a stale-CLAUDE.md flag, and branch-cleanup candidates. Emit a fresh per-project report every sweep; NEVER cache status back into a workspace/manager CLAUDE.md.

Before making any non-trivial code, docs, or config change -- create a feature branch first, then open a PR. Never commit directly to main. Apply by default before any edit; this is the baseline branch discipline that all other git skills build on.

When the issue or PR queue has open unlabeled items -- run a read-only triage pass that labels each one by component, category, priority, and size, flags duplicates, closes noise, and reports the p1 queue before runners are dispatched.

When writing any issue, PR, or commit message, use this for the authoring standard. Covers structure (headings, lists, tables, fenced blocks, the bug block, the before/after) and voice (factual, no emdashes, no editorializing labels, conventional-commit prefixes, no commit trailers).

The canonical owner-prefixed workspace layout. Every repo lives at ~/Code/<host>/<owner>/<repo> -- no active/ or language segment. Enumerate with `ls ~/Code/github.com/*/*`; resolve siblings by path arithmetic. The manager (workspace-root) CLAUDE.md is a map of where things are and how to read them, never a cache of per-project inventory or status. Load this whenever a session reasons about where a repo lives, walks the workspace, or resolves one project from another.

When the dispatcher needs a workspace map -- at the start of any multi-project or scope-scoping invocation. Enumerate repos under the owner-prefixed layout with `ls ~/Code/github.com/*/*`, load each project's positioning and live GitHub state (open/draft PRs, issues, CI), and RECONSTITUTE a tight activity report. Re-survey on every invocation; never cache per-project status back into any workspace CLAUDE.md, and never rely on prior-session memory.

Use this skill for vehicle diagnostics, repair, maintenance, or restoration. Triggers include: 'my car won't start', 'check engine light', 'oil change', 'brake noise', or any reference to classic/vintage car work. Particularly well-suited for air-cooled rear-engine Italian microcars from the 1960s, for which documentation is sparse and patience is mandatory. Do NOT use for automated vehicle testing, Selenium WebDriver, or any other software that merely borrows the word 'drive'.

Use this skill when the user needs to supervise children, manage bedtime, prepare age-appropriate meals, or handle tantrums. Triggers include: 'I'm watching my nephew tonight', 'the baby won't stop crying', 'bedtime is a disaster', or 'what can I feed a toddler'. Do NOT use for babysitting API rate limits or sitting on babies.

Use this skill when the user needs to prepare a meal, bake bread, build a sauce, or perform any food preparation task. Triggers include: 'what should I make for dinner', 'my roux keeps breaking', 'how do I dice an onion properly', or 'I want to make sourdough'. Do NOT use for metaphorical cooking such as 'cooking the books' — see the accounting-fraud skill.

Use this skill when the user needs an outlet installed, a circuit added, a fixture wired, or a panel upgraded. Also triggers on 'my outlet doesn't work', 'the lights are flickering', 'I want to add a dedicated circuit for my workshop', or 'I'm upgrading to a 200-amp panel'. Do NOT use for low-voltage electronics, network cabling, or Arduino projects.

Use this skill when the user needs to amend soil, prune trees or shrubs, manage pests, plant, transplant, or maintain a garden. Triggers include: 'my tomatoes have blossom end rot', 'when should I prune my roses', 'my soil is clay', or 'something is eating my basil'. Do NOT use for 'walled garden' in the platform lock-in sense.

Use this skill when the user requests a haircut, fade, trim, taper, or beard shaping. Triggers include: 'I need a haircut', 'can you give me a fade', 'my bangs are too long', or 'clean up my neckline'. Do NOT use for trimming hedges — see the gardening skill.

Use this skill when the user is responsible for water safety at a pool, lake, or ocean beach. Triggers include: 'someone is struggling in the water', 'I'm lifeguarding this weekend', 'how do I do a reaching assist', or 'what does an active drowning look like'. Do NOT use for guarding the life of a software process — see the monitoring skill.

Use this skill when the user needs to move furniture, load a truck, navigate a couch through a stairwell, or relocate household contents. Triggers include: 'I'm moving next week', 'how do I get this couch down the stairs', 'what size truck do I need', or 'PIVOT'. Do NOT use for moving files between directories — see the filesystem skill.

Declarative multi-stage workflows. Define staged pipelines where each stage feeds its result to the next. Use for structured multi-agent tasks.

GitHub integration via gh CLI. List issues, create PRs, comment on issues. Use for agents that bridge GitHub and the Workshop board.

Autonomous board worker. Receives work items, implements them, reports results. Designed for agents that poll the board and execute tasks independently.

Work board driven development. Post typed work items with dependencies, agents poll by role. Use for multi-step features where ordering matters.

Multiple LLM backends in one workshop. Claude and Codex agents working together. Use when different models have different strengths.

Scheduled agent for periodic checks. Use for CI monitoring, PR scanning, health checks, or any polling-based workflow.

Coordinate a team of AI agents to build software. Break tasks down, create specialist agents from profiles, delegate via ask/cast, review results, and clean up. Use when you need to orchestrate multi-agent workflows.

Implement-then-review pattern. Two agents collaborating via pipe chains. Use when you need a second perspective on code changes.

Curated directory of external skill repositories. Use when the user needs skills not available locally, or to suggest adding new skill sources.

Contributing to skillet itself. Covers architecture, development setup, testing, and PR workflow.

Creating and maintaining skill repos. Covers setup, directory layout, PR review, multi-repo, and hosting.

Set up and configure Skillet skill discovery. Use when the user wants to set up skillet, configure repos, or customize server behavior.

Authoring skills for skillet. Covers the skill format, discovery via suggest, and project manifests.

Using skillet as a skill consumer. Covers searching, browsing, and using skills as MCP prompts.

Write JMESPath expressions to query and transform JSON. Use when extracting, filtering, projecting, or reshaping JSON data. Covers identifiers, projections, filters, multi-select, pipes, functions, and let expressions. Works with any JMESPath implementation.

Run jpx to query and transform JSON from the command line. Covers output formats (CSV, TSV, table, YAML), streaming NDJSON, input modes (slurp, raw-input, null-input), query files, variable binding, and shell pipelines. Use when building shell pipelines or exporting JSON data.

Look up and call the 460+ extension functions in jpx beyond standard JMESPath. Covers array, string, math, datetime, hash, encoding, regex, object, expression (higher-order), geo, text, validation, and 20+ more categories. Use when a task needs functions beyond the 26 standard JMESPath built-ins.

Invoke jpx MCP server tools for JMESPath evaluation, function discovery, JSON utilities, and cross-server tool discovery. Use when evaluating expressions via MCP, exploring functions, analyzing JSON structure, or managing named queries.