aida

What this skill does

<!-- SPDX-FileCopyrightText: 2026 The AIDA Core Authors -->
<!-- SPDX-License-Identifier: MPL-2.0 -->

# AIDA Dispatch

Routes `/aida` commands to appropriate action handlers, managing AIDA's configuration,
diagnostics, feedback systems, and extension management (agents, skills, plugins).

## Activation

This skill activates when:

- User invokes `/aida` command with any action
- AIDA functionality is needed (configuration, status checks, diagnostics)
- Command routing and execution orchestration is required

## Command Routing

When this skill activates, check the `<command-args>` tag to determine which action to route:

### Diagnostic Commands

For `status`, `doctor`, or `upgrade` commands:

- Read `references/diagnostics.md` for execution workflow
- These are non-interactive commands that execute Python scripts directly

### Configuration Commands

For `config` command (alone):

- Read `references/config.md` for YAML-based configuration workflow
- This is an interactive command with:
  - Dynamic menu generation based on installation state
  - Auto-detection of project facts (saved to YAML)
  - Minimal questions (0-3) for unknown preferences only
  - Automatic skill generation from YAML config

For `config validate` command:

- Read `references/validate.md` for the workflow
- This is a **non-interactive** CI-friendly health check
- Run `~/.aida/venv/bin/python3 {base_directory}/scripts/validate.py`
- Exits 0 on success, 1 on failure — suitable for `make` / CI gates
- Pass `--json` for machine-consumable output

### Permissions Commands

For `config permissions` or `permissions` commands:

- **Invoke the `permissions` skill** to handle these operations
- The skill handles interactive permission setup and audit mode

**Process:**

1. Parse the command to detect:
   - `--audit` flag for audit mode
   - No flags for interactive setup

2. Invoke `permissions` skill with the parsed context

**Examples:**

```text
/aida config permissions           → permissions skill
/aida config permissions --audit   → permissions skill (audit)
```

### Feedback Commands

For `feedback`, `bug`, or `feature-request` commands:

- Read `references/feedback.md` for feedback collection workflow
- These are interactive commands that collect and submit user input

### Help Command

For `help` or no arguments:

- Display the help text inline (see Help Text section below)
- Read the plugin version from `{base_directory}/../../.claude-plugin/plugin.json`
  (the `"version"` field) and include it in the footer
- No additional reference files need to be loaded

### About Command

For `about`:

- Read `{base_directory}/../../.claude-plugin/plugin.json`
- Resolve the installed path from `{base_directory}/../..`
- Display plugin metadata:

```markdown
**AIDA Core Plugin**

- **Version:** {version from plugin.json}
- **Author:** {author.name from plugin.json}
- **Repository:** {repository from plugin.json}
- **Installed at:** {resolved path to plugin root}
```

### Agent Management Commands

For `agent` commands:

- **Invoke the `agent-manager` skill** to handle these operations
- Pass the full command arguments to the skill
- The skill handles create, validate, version, and list operations

**Process:**

1. Parse the command to extract:
   - Operation: `create`, `validate`, `version`, `list`
   - Arguments: name, description, options

2. Invoke `agent-manager` skill with the parsed context

**Examples:**

```text
/aida agent create "description"     → agent-manager skill
/aida agent validate --all           → agent-manager skill
/aida agent version my-agent patch   → agent-manager skill
/aida agent list                     → agent-manager skill
```

### Skill Management Commands

For `skill` commands:

- **Invoke the `skill-manager` skill** to handle these operations
- Pass the full command arguments to the skill
- The skill handles create, validate, version, and list operations

**Process:**

1. Parse the command to extract:
   - Operation: `create`, `validate`, `version`, `list`
   - Arguments: name, description, options

2. Invoke `skill-manager` skill with the parsed context

**Examples:**

```text
/aida skill create "description"     → skill-manager skill
/aida skill validate --all           → skill-manager skill
/aida skill version my-skill patch   → skill-manager skill
/aida skill list                     → skill-manager skill
```

### Plugin Management Commands

For `plugin` commands (including `plugin scaffold`):

- **Invoke the `plugin-manager` skill** to handle these operations
- Pass the full command arguments to the skill
- The skill handles create, validate, version, list, scaffold, update,
  deps, and agents operations
- Scaffold creates a NEW plugin project (not an extension inside an existing
  project)
- Update scans an existing plugin and patches it to current standards
- Deps reports declared plugin dependencies + their satisfied / missing /
  wrong-version status (#20)

**Process:**

1. Parse the command to extract:
   - Operation: `create`, `validate`, `version`, `list`, `scaffold`,
     `update`, `deps`
   - Arguments: name, description, options

2. Invoke `plugin-manager` skill with the parsed context

**Examples:**

```text
/aida plugin create "description"        → plugin-manager skill
/aida plugin validate --all              → plugin-manager skill
/aida plugin list                        → plugin-manager skill
/aida plugin scaffold "my-new-plugin"    → plugin-manager skill
/aida plugin scaffold                    → plugin-manager skill (will ask)
/aida plugin update "/path/to/plugin"   → plugin-manager skill
/aida plugin deps                        → plugin-manager skill (cwd)
/aida plugin deps "/path/to/plugin"     → plugin-manager skill
/aida plugin agents                      → plugin-manager skill (cwd)
/aida plugin agents "/path/to/project"  → plugin-manager skill
```

### Hook Management Commands

For `hook` commands:

- **Invoke the `hook-manager` skill** to handle these operations
- Pass the full command arguments to the skill
- Hooks are settings.json config, not files
- The skill handles list, add, remove, and validate operations

**Process:**

1. Parse the command to extract:
   - Operation: `list`, `add`, `remove`, `validate`
   - Arguments: event, matcher, command, scope

2. Invoke `hook-manager` skill with the parsed context

**Examples:**

```text
/aida hook list                      → hook-manager skill
/aida hook add "auto-format"         → hook-manager skill
/aida hook remove my-hook            → hook-manager skill
/aida hook validate                  → hook-manager skill
```

### Expert Registry Commands

For `expert` commands:

- **Invoke the `expert-registry` skill** to handle operations
- Pass the full command arguments to the skill
- The skill handles list, configure, panels, and panel operations

**Process:**

1. Parse the command to extract:
   - Operation: `list`, `list configure`, `panel list`,
     `panel create`, `panel remove`
   - Arguments: panel name (for create/remove)

2. Invoke `expert-registry` skill with the parsed context

**Examples:**

```text
/aida expert list                   → expert-registry skill
/aida expert list configure        → expert-registry skill
/aida expert panel list            → expert-registry skill
/aida expert panel create review   → expert-registry skill
/aida expert panel remove review   → expert-registry skill
```

### Knowledge Commands

`/aida knowledge` commands split across two skills:

- **`knowledge-sync` skill** owns the mechanic — `sync`, `status`,
  `discover`. Deterministic: reads source declarations, fetches
  content, walks spider roots.
- **`knowledge-curator` skill** owns the policy workflow — `curate`,
  `review`. LLM-orchestrated: reasons about which discovered URLs are
  worth using and persists decisions.

**Process:**

1. Parse the command to extract:
   - Operation: `sync`, `status`, `discover`, `curate`, `review`
   - Argument: agent name (and any flags)

2. Route to the