> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mobile-starter.amisi.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Skills

> Apply AMiSi workflow and domain procedures with any AI coding tool

Skills define repeatable procedures under `.ai/skills/<name>/SKILL.md`. They use portable `name` and `description` metadata and remain readable when your AI tool does not support native skill discovery.

## Workflow skills

| Skill                  | Use it for                                                         |
| ---------------------- | ------------------------------------------------------------------ |
| `plan-change`          | Research non-trivial or ambiguous work before editing.             |
| `implement-change`     | Apply an approved change through one worktree writer.              |
| `review-change`        | Review a diff, commit range, or pull request independently.        |
| `verify-workspace`     | Run targeted Nx checks and `bun run verify`.                       |
| `update-documentation` | Keep Mintlify pages aligned with shipped behavior.                 |
| `prepare-commit`       | Generate a commit message from the staged diff.                    |
| `prepare-release`      | Prepare release context, migrations, risks, and rollback guidance. |

## Domain skills

| Skill                         | Use it for                                                                         |
| ----------------------------- | ---------------------------------------------------------------------------------- |
| `debug-problem`               | Reproduce failures, gather evidence, and confirm a root cause before fixing.       |
| `write-tests`                 | Add behavior-focused Jest, React Native, Zustand, adapter, or regression tests.    |
| `build-mobile-feature`        | Build routes, screens, state, UI, translations, analytics, and mobile states.      |
| `add-provider-adapter`        | Integrate a provider behind a package contract, registry, runtime, and mock.       |
| `optimize-mobile-performance` | Measure and improve rendering, lists, startup, memory, animations, or bundle size. |

Load only skills that match the request. Broad workflows should not activate every domain skill.

## Invoke a skill

Give your tool the task and skill path:

```text theme={null}
Read AGENTS.md and .ai/README.md. Plan this change by following
.ai/skills/plan-change/SKILL.md. Do not modify files.
```

For a complete feature:

```text theme={null}
Use .ai/tasks/templates/feature.md to structure this request, then follow
.ai/workflows/feature.md. Show me the plan before implementation.
```

## Compose skills by task

A mobile feature normally combines `plan-change`, `build-mobile-feature`, `write-tests`, `implement-change`, `verify-workspace`, and `review-change`.

A provider integration adds `add-provider-adapter`. A reproducible bug starts with `debug-problem`. Use `optimize-mobile-performance` only when the task has a measured performance problem or explicit performance acceptance criterion.

```text theme={null}
Investigate this Android-only notification registration failure. Follow
.ai/skills/debug-problem/SKILL.md before proposing a fix. If the root cause is
inside the provider boundary, also follow
.ai/skills/add-provider-adapter/SKILL.md. Add a regression test, verify the
notifications project, then run bun run verify.
```

## Verification boundary

The active AI runtime may choose how to execute a procedure, but the repository defines completion. Use `.ai/skills/verify-workspace/SKILL.md` and finish with:

```sh theme={null}
bun run verify
```

## Next steps

* [Choose an agent](/ai/agents) for the role that should perform a skill.
* [Structure a task](/ai/task-specs) before broad or multi-stage work.