Skills

AI skills for building software factories. My personal library of domain-agnostic agent skills, reused across every project. Small, composable, and hackable — works with any harness that supports skills: Claude Code, Codex, opencode, Cursor, duet, and 70+ others.

npx skills add dzhng/skills

Add --list to pick individual skills, or copy any skills/<category>/<name>/ folder into your harness's skills directory (e.g. .claude/skills/).

From a clone, npm run install-skills does the same without the registry:

npm run install-skills              # into ~/.agents/skills, linked from ~/.claude/skills
npm run install-skills -- ../my-app # into a repo instead of the home directory
npm run install-skills -- --only write-spec,review ../my-app
npm run list-skills                 # names and categories

.agents/skills/<name>/ holds the real files (flat, category-free, with cross-category links rewritten to match); .claude/skills/<name> is a relative symlink into it, so both harnesses read one copy. Re-running overwrites the installed copies — a .claude/skills/<name> you keep as a real directory is left alone, and a .claude/skills that is already a symlink is left as is. Add --dry-run to see the plan first.

Why

Software is moving from tasks to factories: agents that pursue a goal autonomously until the output can be trusted. The hard part isn't breaking the goal into tasks — it's breaking it into independently verifiable pieces, and knowing where the pieces even are.

These skills run that loop. Treat the unknown as fog of war: map the terrain, carve it into territories that build and verify in isolation, and recursively re-slice whatever hides more map. And re-planning doesn't stop when planning ends — the spec is a living document, updated and re-sliced mid-implementation whenever the work teaches the agent that the plan is stale. Every piece must prove itself — architecture review, code review, and visual review against a baseline — before the loop moves on. Each iteration gets less wrong, until the goal is done.

Proof: one unattended Codex run pursuing a single goal for 3d 18m on top of these skills, slicing and iterating until done.

How to use

Use a chained pipeline to build a feature, a research loop to discover what works, or individual skills as needed. Every skill stands alone.

The full loop — a big feature, start to finish

1. Map the fog. /explore-unknowns on the idea. It interviews you quadrant by quadrant and hands you rendered options, mocks, and decision tables to react to instead of asking you to imagine. By the end you know what the feature does.

2. Codify. /write-spec on that map. Most decisions were already made upstream, so this pass is transcription — I don't read the spec. Anything genuinely new it hits, it asks about instead of deciding.

3. Build. Kick off the loop:

   /goal /implement-spec specs/<feature>
   

   /goal is what puts the harness in loop mode — same move in Claude Code or Codex — and the spec drives it from there. A couple of hours for a small feature, two or three days for a large one. Add whatever framing fits: on the xyz branch, or using /codex as the implementer while you stay the parent orchestrator and reviewer.

4. Review the choices, not the diff. The run ends by consolidating specs/<feature>/choices.md — every decision the agent made where the spec was silent, ranked least-confident first. That's the review surface. Send changes back and the next pass re-audits: every time the AI writes code, you audit what it chose.

   The rest fires on its own: a /review pass at the end of every slice, /screenshot-critique and /compare-screenshots on anything visual, /close-spec when the last slice lands, and a re-slice of the plan whenever implementation proves it stale.

Budget: 30 minutes to a few hours on steps 1–2, 30 minutes to a few hours on step 4. A run that goes two days is more like 2–3 hours on each end. Your time is in the bookends; the middle is unattended.

Research — learn through fast experiments

Use Auto Research when the next decision needs experimental evidence. It starts with one fast, revealing task, tests a short batch of hypotheses, checks combinations, and expands coverage as the approach improves. New failures become the focus; earlier tasks become regression checks.

/auto-research Reduce cost per task by at least 15% relative to the saved
baseline, without reducing task success. Start with one fast development task.

If the evaluator, metric, baseline, or required improvement is unclear, the skill asks before experimenting. Passing an evaluation and meeting an improvement target are separate requirements. The output includes the best verified artifact and a parameter-effect map: what was tested, where it helps or hurts, and how changes interact. Use that evidence to inform a spec when the research is ready for implementation.

À la carte — the spontaneous path

• A brainstorm turns out to be a feature. /explore-unknowns works at the end of a discussion as well as at the start — run it to sweep for the angles neither of you thought of, then pick the loop up at step 2.

• Any code change that didn't come from a spec. An ad hoc fix that touched more than expected: /review first (refactor-clean → code-review → write-docs), then /audit-choices. When the diff is too big to read, the choices ledger is how you still understand what is now in your codebase.

Skills

Engineering — slice, build, verify, repeat

Skill	What it does
explore-unknowns	Walk the user through mapping a task's unknowns quadrant by quadrant — known knowns first, then interviews, reactable artifacts, and blindspot passes — ending with a complete four-quadrant map.
auto-research	Optimize through fast, progressive experiments, producing a verified candidate and a map of parameter effects and tradeoffs.
write-spec	Break a large feature into independently verifiable, human-reviewable slices with API seams and playable checkpoints.
implement-spec	Build an existing spec to completion, one reviewable pass at a time, delegating independent slices in parallel.
implement-spec-with-codex	Run implement-spec with Codex writing the code — you orchestrate, integrate, and review every pass.
close-spec	Archive a shipped spec and rewrite it from a build plan into a durable rationale record that points back at the code.
handoff-spec	Hand a spec that is mid-implementation to another machine or agent: capture what the session knows in the spec, then push. User-invoked.
refactor-clean	Refactor by moving ownership to one clean concept instead of layering compatibility sediment beside the problem.
write-tests	Write tests one tracer bullet at a time that pin real behavior — not implementation details, config values, or lucky samples.
audit-tests	Map contracts to independent test proof, consolidate redundant coverage, and remove test-only machinery without losing regression protection.
audit-performance	Find hot paths that amplify or repeat without progress, rank them by real failure risk, and prefer the smallest bounded fix that preserves healing.
write-docs	Write docs as a glossary of principles and pointers, never a mirror of the code that will rot.
code-review	Audit a diff for stale names, dead references, needless complexity, and comments that narrate instead of explain — ending on a clean/not-clean verdict.
audit-choices	Audit the choices an implementer made, not its diff — a pure, never-blocking audit whose ledger discloses the architecture and decisions made on the user's behalf, reviewed instead of the code.
eli5	Explain a spec or change in plain language without losing precision — the ELI5 register other skills borrow for standalone, walked-scenario explanations.
review	Closeout a finished change as one pass — refactor-clean, then code-review, then write-docs — sequenced into a single verdict.
codex	Use the local Codex CLI as an independent second agent for review and (on explicit ask) delegated implementation.
claude	Use Claude Code (claude -p) as an independent second agent for consultation and (on explicit ask) delegated implementation.
marketing-pages	Rulebook for writing, updating, and auditing marketing pages by page class — campaign landers stay noindexed and unlinked with one CTA; everything else earns its sitemap entry, crawl-rail link, and canonical copy source.

Visual review — never accept visuals on vibes

Skill	What it does
design-with-images	Explore generated visual options, preserve the selected target, and iterate real screenshots against it until the implemented design matches.
compare-screenshots	Compare captures against an approved design or the intended result; use telemetry to locate differences rather than treating a historical baseline as correct. Ships a reusable diff script that also measures a lone capture for flat, empty, or misframed content.
screenshot-critique	Use an unprimed subagent as a second set of eyes on visual work before accepting it; mandatory before declaring a reported visual bug fixed.
preview-shots	Open a curated set of image shots in one macOS Preview window for the user to eyeball.

Authoring — keep the skills themselves sharp

Skill	What it does
write-skills	Create or revise agent skills: triggers, leading words, progressive disclosure, and the failure modes to prune.
audit-agents	Audit or rewrite AGENTS.md so it holds only lasting principles, with iteration speed at the core. Ships an example AGENTS.md that pairs with this pack.
eval-skills	Eval a skill against golden cases — blind runs in fresh subagents, a separate judge, and gap-driven edits.

Graphics

Skill	What it does
renderer	Architecture rules for GPU rendering with TypeGPU/WebGPU — one frame orchestrator, single owners, a resource registry, depth and compositing contracts, and verification that looks at pixels.
launch-video	Direct a showreel-grade launch video in code — one visual concept grown from the project, picture and a synthesized score locked to one timing source, and a preview frame that survives social feeds.

License

MIT