itaywol/adeptability / .adeptability/staging/.claude/skills/authoring-adept-skills
authoring-adept-skills
VerifiedWrite a good, portable adept skill: pick the right activation, craft a triggering description, keep it scan-safe and within harness byte budgets. Apply when creating or editing a SKILL.md or running `adept skill add`.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 3.5 KB |
SKILL.md
View on GitHub---
name: authoring-adept-skills
description: 'Write a good, portable adept skill: pick the right activation, craft a triggering description, keep it scan-safe and within harness byte budgets. Apply when creating or editing a SKILL.md or running `adept skill add`.'
allowed-tools: [Bash, Read, Edit, Write]
---
# Authoring a good adept skill
A skill is a directory `<root>/skills/<id>/` with one `SKILL.md` (YAML frontmatter +
markdown body) and optional sidecars. `adept` renders it into every enabled harness, so
write it **once, well, and harness-neutral**.
## Canonical SKILL.md format
```markdown
---
id: pr-review # ^[a-z0-9](?:[a-z0-9-]{0,48}[a-z0-9])?$ — matches the directory name
description: Use before opening a PR. Tests, security, performance. # <= 280 chars
activation: agent # always | globs | agent | manual
globs: [] # REQUIRED when activation: globs (e.g. ["**/*.go"])
allowed-tools: [Read, Grep] # carried into Claude Code
targets: [] # empty = all enabled harnesses
tags: [review, quality]
---
# PR Review Checklist
- [ ] Tests added or updated
- [ ] No secrets in the diff
```
Validated against `pkg/adeptschema/skill.schema.json` on every load — `adept sync` fails
loudly on a malformed skill, so let it catch your mistakes.
## Pick activation deliberately — this is the whole game
| activation | Fires when | Use for |
|---|---|---|
| `agent` | The model decides from the **description** | Most skills. Procedures, checklists, how-tos. |
| `globs` | A matching file is in context | Language/path-specific rules (`**/*.go`, `**/*.tsx`). |
| `always` | Every request | Rare. Project-wide invariants only — costs budget on every turn. |
| `manual` | Only when explicitly invoked | Heavy or destructive procedures you don't want auto-firing. |
## The description is the trigger — write it like one
For `activation: agent`, the description is the **only** thing the model sees when deciding
whether to load the skill. Make it earn the load:
- Lead with **when to apply**, not what it is: *"Use before opening a PR…"* beats *"A PR
helper."*
- Name concrete nouns the user will say (file types, commands, tools, error strings).
- One line, ≤ 280 chars. If you can't say when it applies in one line, the skill is too broad
— split it.
## Keep it portable and lean
- **Harness-neutral body.** Don't say "in Cursor, click…"; describe the behavior, not the UI.
- **Budget-aware.** Aggregator harnesses (Codex/Copilot) cap total bytes; a bloated skill
gets dropped first. Tighten prose; move long reference material into `references/` sidecars.
- **Scan-safe.** Installs run a safety scan. Avoid anything that reads like prompt injection,
exfiltrates secrets, or runs opaque remote code — it will flag (`critical` hard-blocks).
- **Sidecars** (`scripts/`, `references/`, `assets/`) ride along to harnesses that support
them; keep the SKILL.md body self-sufficient for those that don't.
## Loop
```bash
adept skill add my-skill --edit # scaffold + open $EDITOR
adept sync # render to every enabled harness
adept status && adept diff # confirm it landed clean
```
Run `adept skill --help` (and `adept skill add --help`) for the current flags and
subcommands — prefer it over memory; it never drifts. Edit canonical, `adept sync`, never
hand-edit rendered files. See [[using-adept]] for the
full CLI and [[adept-self-improve]] for turning a one-off lesson into a durable skill.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| 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: 472k | Updated | Claude CodeCodex | Verified | Install |
| 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: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| 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 available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |