leighstillard

Shepherd an open PR to merge. Watches CI, fixes failures, addresses review comments, resolves conflicts, and enables/executes merge when green. Use after opening a PR, when the user says babysit/watch/get this merged, or as the final step of /ship.

Visual and UX review for user-visible changes: frontend components, HTML/CSS, templates, dashboards, rendered docs, emails, and CLI output formatting. Use as part of /ship when a diff touches UI/human-facing output. If nothing user-visible changed, explicitly mark not applicable and stop.

Commit, push, and open a pull request for the current change. Use when the user says open a PR, ship this, commit and push, or as the PR creation step of /ship. Requires verification evidence in the PR body.

Take a finished local change to a merged PR through non-negotiable release gates. Use when the user says ship it, /ship, get this merged, release this, or when /coach has a READY local diff. Enforces simplify, verify, design applicability, PR evidence, independent review, CI, and merge babysitting while preserving Anthropic budget where safe.

Simplify a completed diff before it ships. Use before /ship, after a feature works, or when a change feels over-engineered. Removes unnecessary abstraction, scope creep, dead code introduced by the diff, and confusing structure without changing behaviour.

Prove a change actually works before it ships. Use whenever a change is done, after implementing a feature/fix, before opening a PR, or when the user says verify/check this works/is this actually working. Tests passing is not sufficient; this skill requires real entry-point evidence.

Shepherd an open PR to merge. Watches CI, fixes failures, addresses review comments, resolves merge conflicts, and enables auto-merge when green. Use after opening a PR, when the user says "babysit", "watch the PR", "get this merged", or as the final step of /ship.

Visual and UX review pass for any change that touched UI — frontend components, HTML, CSS, templates, dashboards, CLI output formatting, or anything a human looks at. Use whenever a diff includes UI files, when the user asks for a design review, or as part of /ship. If the diff touched nothing user-visible, state that and exit immediately.

Commit, push, and open a pull request for the current change. Use when the user says "open a PR", "ship this", "commit and push", or as the PR step of /ship. Handles branch hygiene, commit message quality, and PR description.

Take a finished change from local diff to merged PR. Use whenever the user says "ship it", "/ship", "get this merged", or has a completed change ready to go out. Drives the change to a goal state through non-negotiable quality gates.

Prove a change actually works before it ships. Use whenever a change is "done" and needs validation — after implementing a feature, fixing a bug, or before opening a PR. Also trigger when the user says "verify", "check this works", "is this actually working", or invokes /ship. Tests passing is not sufficient evidence; this skill drives the change end-to-end.

Use when stuck after 2+ failed attempts, facing architectural decisions, debugging loops, cross-repo changes, or needing a second opinion. Auto-routes: Sonnet consults Fable, Fable consults Codex. Invoke manually or let it trigger automatically when things get hard.

Two-model orchestration workflow where Claude acts as planner/reviewer and Codex GPT-5.5 (xhigh reasoning, fast tier) acts as executor, plus an advisor mode implementing Anthropic's advisor strategy (executor drives end-to-end, escalates to Claude only when stuck; Claude returns a plan, correction, or stop signal and never writes code). Use this skill whenever the user invokes /smart-orchestrate, asks to delegate implementation to Codex, wants Claude to plan/review while another model writes the code, mentions the codex-rescue agent, or asks to run a task in advisor mode / advisor strategy where the executor drives and Claude only advises.

Pre-implementation review skill that catches over-engineered specs, plans, and architecture proposals before code is written. Produces "the boring version" — the simplest implementation that satisfies stated requirements. Use this skill whenever reviewing a technical spec, implementation plan, architecture proposal, ADR, RFC, design doc, or PR description for unnecessary complexity. Also trigger when the user asks to simplify a plan, find the minimal viable approach, check for over-engineering, produce a boring baseline, or review whether a design is more complex than the problem requires. Trigger even if the user just pastes a spec and asks "what do you think?" — the default lens should be simplification review.

Recall context from a Slack thread — either the current thread (if in one) or the most recent thread in the channel. Acts as a memory layer by pulling conversation history from Slack. Only works when Slack message context is present (channel_id available). Use when you need to review what was discussed, recall decisions, or catch up on a prior conversation thread.

Persistent verification discipline for Claude Code sessions. Activates on explicit toggle ("duncemode on/off/status/all"), on the "bullshit" family of rejection phrases ("bullshit", "/bullshit", "call bullshit", "that's bullshit"), on disbelief signals ("really?", "are you sure", "prove it", "did you actually"), on accusations of wrongness or fabrication ("you're wrong", "you made that up", "you're hallucinating"), on demands to think harder ("think harder", "dig deeper", "trace it end to end", "rubber duck"), and on any expression of user frustration ("wtf", "what the fuck", "ffs", "jesus christ", "come on", "stupid", "useless", "broken", "garbage", calling the agent stupid or similar). Runs a verification triage loop (catches narrated tool use, fabricated citations, silent patch failures, fabricated MCP results, scope drift) and an end-to-end trace protocol (catches lifecycle bugs, missed state transitions, wrong mental models of the system). Escalates to full mode with a mandatory rubber-duck restatement w