almogbaku/aos/capability-lifecycle

Claude Code and agent skill by almogbaku/aos

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

The aos capability lifecycle's front door. Use when the user asks to install, update, upgrade, remove, or customize an aos capability, mentions or shares a CAPABILITY.md or a capability directory, asks what capabilities are installed or what aos is, or asks about the manifest, the lockfile, or a harness cheat-sheet. Routes to the narrower skill that owns the job — capability-install, capability-upgrade, capability-remove, capability-onboard, capability-evolve, capability-import, capability-build, capability-contribute, capability-review — and carries the household layout, the install contract and the naming rules they all share. Do NOT use when one of those already matches the ask; this is the map, not the work.

Install

npx vskill@latest install almogbaku/aos/capability-lifecycle

SKILL.md

capability-lifecycle — the map

Capabilities are self-describing prompts; you are the installer. Judgment is yours; bookkeeping is aos-cap's (aos-cap --help — the lockfile is its file, never edit the YAML). Everything lives in the household <aos_home> — the aos home: --home/AOS_HOME, else ~/aos (not your home directory): upstream/ the pristine kit clone, personal/ the user's private repo (their MOD files, the pinned renders harnesses symlink to, their private capabilities), .aos/ machine state (the lockfile; harness.yml, the household's harness — aos-cap harness show). The diff gate is never optional. MOD.md states the user's personalization — what they changed from the shipped defaults, current state, edited in place: you re-apply it, you never overwrite it — and you never write anything to upstream (PR, issue, +1, push, fork, remote branch) without the user's explicit approval. The full rules: reference/contract.md. The overlay doctrine (interviews, transform, capture, promote/retire, persist): reference/overlay.md. Naming, installed skill names, and the uniqueness gate: reference/naming.md — read it before authoring or installing anything that ships a skill. Names are computed, never authored: write \{{skill: <id>}} and \{{agent: <id>}} in prose and aos-cap render substitutes the installed name. A backslash-escaped slot stays literal, which is how this sentence survives its own render.

AskWhere it goes
"install <capability>" · a capability offered for install{{skill: install}}
"update" / "upgrade" (kit-wide or one capability){{skill: upgrade}}
"remove/uninstall <capability>"{{skill: remove}}
"run/redo my interview" · bootstrap a new user{{skill: onboard}}
"wrap what I already built into a capability"{{skill: import}}
something recurring/systemic to build{{skill: build}}
"fix/change <capability> for everyone" · contribute upstream{{skill: contribute}}
"change how <capability> behaves for me"{{skill: evolve}}
"promote my tweak" / "this should be for everyone"{{skill: evolve}} classifies, then hands to {{skill: contribute}}
"review/audit <capability>" · before contributing · "something looks wrong here"{{skill: review}} — read-only, architecture first
"what's installed?"aos-cap list
your harness's mappingreference/harness-<harness-runtime>.md — frontmatter mapping (one row per aos concept and depends.host word) plus the harness's judgment; an aid, not a gate: loaded per operation, never standing context; none → reference/no-cheatsheet.md

Manifest quick facts: CAPABILITY.md frontmatter is strict-typed (aos-cap manifest <dir> parses and validates it — its errors name the field and rule); the fields are id · version · tags · summary · depends · cron · skills · skill_prefix · kb · uses (a cron entry: id · expr · agent · prompt_ref · exec · degraded), x-* reserved for extensions; the prose below the frontmatter is the installer's briefing, never runtime context. A skill's id is capability-local — the name it installs under is aos-cap skills <dir>'s answer, and it is single-owner across the whole harness; an agent's works the same way, from the same prefix, via aos-cap agents <dir> (reference/naming.md).

Experience

Every lifecycle interaction is warm + expert: concept before mechanics, explain-then-act. Before doing anything visible, say what is about to happen and what it buys the user — then proceed; the diff gate is the safety net, not repeated consent prompts. Batch questions; if you can infer it, don't ask. Set expectations before slow steps ("installing the tool takes a minute"). Show the gate as a payoff: each artifact → what it does for you. Close with a specific, celebratory summary — what was installed, where, which cron jobs, any degraded modes — never a vague "done." User-facing lines are warm prose; your internal steps stay silent.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
3
Repository forks
2
7-day trend score
0

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/almogbaku/aos/capability-lifecycle/badge)](https://verified-skill.com/skills/almogbaku/aos/capability-lifecycle)

Something wrong with this skill or its page? Tell us.