Skip to content

umararshad4/honey-stack / skills/write-skill

write-skill

Verified

Write a new honey-stack skill or patch an existing one to the quality bar, with evals.

FileKindSize
SKILL.mdskill md7.8 KB
agents/openai.yamlconfig154 B
references/evals.mdmarkdown4.0 KB
references/writing-for-agents.mdmarkdown1.4 KB

references/writing-for-agents.md

View on GitHub
# Write documents agents can follow

Distilled from Matt Pocock's `writing-for-agents` at 4ae2317ddb22b046e26455e4f9c9b3aaaf62212d (MIT). The house rules in `docs/skill-style.md` still own packaging and invocation.

- A reference pointer says **when to read** as well as what the target contains. Repair a weak pointer before copying the whole reference into an always-loaded skill.
- Put steps in execution order. Keep each definition beside its rule, exception and example; scattered fragments force readers to reconstruct the procedure.
- State observable, exhaustive completion criteria: every caller updated, every changed flow exercised. An attractive output is not a completion criterion for the work behind it.
- Inline material needed by every branch. Move branch-specific detail to a linked reference. Split an entry point only for a genuinely distinct invocation, not just to shorten a file.
- Keep global instructions as navigation and decisions. Mechanical standards belong in checks; judgment rubrics belong with review. Don't make the implementer reread a whole standards catalog every turn.
- Distinguish context cost from user choice. More commands and more descriptions both cost attention; extend an existing owner before creating a competing skill.

Done when: every pointer has a trigger, each meaning has one owner, and a cold reader can identify both the next step and its stopping condition.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 470kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall