rexchao1
UserA light/dark factory workflow: agent skills talking to an adapted local factory server, with planning, a roadmap, and a dark factory for unattended pull requests.
Categories
Indexed Skills (19)
lavish
Turn complex or visual agent responses into rich, reviewable HTML artifacts the user can annotate and send feedback on, using the lavish-axi CLI. Use when about to give a plan, comparison, diagram, table, code diff, report, or anything easier to grasp visually than as prose.
teach
Teach the user a new skill or concept, within this workspace. Use when the user asks to learn or understand something rather than have it done for them.
handoff
Compact the current conversation into a handoff document for another agent to pick up. Use when a session is ending, context is running low, or the user asks for a handoff.
ask-user-authority
Decision procedure for findings that say "ask the user". Use before deciding any such finding, regardless of how much autonomy the project grants, to distinguish corrections within accepted intent from product or engineering contract expansion that requires the user.
boulder
Break an idea too big to build in one pass into a route of checkpoints, then draft, revise, or freeze the next checkpoint's PRD with every decision cited or marked Not yet specified. Use when the user says "boulder", asks to break a big idea into checkpoints, or asks for a checkpoint PRD to be drafted, revised after a critique, or frozen after review.
checkpoint-critic
Critique a checkpoint PRD with fresh eyes and return findings as JSON: decisions dressed as decided, missing failure modes, uncited claims, and scope past the split rule. Use when the user says "critique this checkpoint", or a checkpoint loop runs a critique pass on a PRD.
codebase-design
Shared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable or AI-navigable, or when another skill needs the deep-module vocabulary.
dark-factory
Send a decided spec to the factory, read what the factory is doing, and answer its questions. Work builds unattended on your factory host, one pull request per task. Use when triage says dark, when the light factory hands over pebbles, and for any question about a run.
diagnostic-reasoning
Procedure for diagnosing reported bugs. Use before scoping a reported bug and before acting on a diagnostic report. Owns end-user-aligned reproduction, causal separation, divergent-path and history inspection, counterfactual testing, and disconfirming evidence.
improve-codebase-architecture
Scan a codebase for deepening opportunities, present them as a reviewable artifact, then grill through whichever one you pick. Use when the user asks for an architecture review, wants to find refactors that make code more testable or navigable, or asks what to restructure.
light-factory
Plan an idea too big for one spec, with the user in the loop: research it in subagents, route it into checkpoints, write one checkpoint's PRD with every decision cited, run a fresh critic on it, review it in the browser, freeze it, cut it into pebbles for the dark factory. Use when triage says light, when the user says "light factory" or "boulder", or to move a checkpoint that sits in review, fog, or frozen.
pebble
Turn a frozen checkpoint PRD and its settled decisions into ordered factory tasks in blueprint's shape, one pull request each, ending in a closure task. Use when the user says "pebble", or a frozen checkpoint PRD needs factory tasks.
stow
Sweep the current conversation for durable knowledge - user preferences, project facts, operational gotchas, standing decisions, and unfinished next steps - and file each through explicit instructions, existing local conventions, or the private `.stow-notes.md` fallback, curating tiered, decaying destination files as it writes. Use when the user invokes /stow, asks to save or write down what was learned this session, or before a context reset or long break.
triage
Decide what a request for software work gets: done here in the conversation, a spec sent to the dark factory, or a planning session in the light factory. Use at the start of every request for work, before reading code for it.
unslop
Cut AI tells from writing and put human voice back in. Use when drafting or editing any prose a human will read - docs, comments, commit messages, PR descriptions, issues, posts, chat replies - and when asked to unslop, de-AI, or humanize a piece of text.
wayfinder
Plan a huge chunk of work - more than one agent session can hold - as a shared map of decision tickets on your issue tracker, and resolve them one at a time until the way to the destination is clear.
wizard
Generate an interactive bash wizard that walks a human through steps only they can perform. Use when provisioning infrastructure, setting up credentials or CI secrets, walking an unfamiliar third-party dashboard, or running a one-off migration or cutover. Don't invoke this for steps the agent can perform itself.
writing-for-agents
Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.
grilling
Grill the user relentlessly about a plan, decision, or idea. Use when the user wants to stress-test their thinking, or uses any 'grill' trigger phrases.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.