list

What this skill does

# List Claude Code Components

List various Claude Code components based on the requested action.

## Argument Routing

| Action | Description |
|--------|-------------|
| `skills` (default) | List all available skills with descriptions |
| `env-vars` | List environment variables (configured or researched) |
| `sessions` | List recent sessions with sizes, ages, resumability |
| `settings` | List settings.json configuration options |
| `tools` | List core Claude Code tools with parameters |

When invoked without arguments, default to `skills`.

Parse `$ARGUMENTS` to determine the action. The first token is the action keyword. Remaining tokens are passed as sub-arguments to the action handler.

---

## Action: skills

List ALL available Claude Code skills by running the list_skills.py script.

### Step 1: Find the Script

Locate list_skills.py in the claude-ecosystem plugin:

```bash
# Check installed plugins first
find ~/.claude/plugins -name "list_skills.py" -path "*claude-ecosystem*" 2>/dev/null | head -1
```

If not found, check current repo (for development):

```bash
find . -name "list_skills.py" -path "*claude-ecosystem*" 2>/dev/null | head -1
```

### Step 2: Run the Script

Execute with Python using the path found in Step 1:

```bash
python "<found-path>/list_skills.py"
```

### What the Script Scans

The script finds ALL skills from:

- **Personal skills**: `~/.claude/skills/*/SKILL.md`
- **Project skills**: `.claude/skills/*/SKILL.md` (current working directory)
- **Plugin skills**: `~/.claude/plugins/marketplaces/*/*/skills/*/SKILL.md`

### Output Format

The script outputs formatted markdown:

```text
## Personal Skills (~/.claude/skills/)
### **skill-name**
Description from SKILL.md frontmatter.
---

## Project Skills (.claude/skills/)
### **skill-name**
Description from SKILL.md frontmatter.
---

## Plugin Skills
### **plugin-name:skill-name**
Description from SKILL.md frontmatter.
---

**Total: X skills** (Y personal, Z project, W plugin)
```

Output the script results directly - they are already formatted.

---

## Action: env-vars

Research and list ALL known Claude Code environment variables from multiple sources, or show currently configured variables across all settings scopes.

### Sub-arguments

| Argument | Description |
|----------|-------------|
| `--configured` | Show currently configured env vars from all settings files |
| `--official-only` | Only show variables documented in official docs |
| `--changelog-only` | Only extract from CHANGELOG.md |
| `--full-research` | Run comprehensive MCP research (slower, more complete) |
| (none) | Default: official docs + changelog extraction |

### Behavior Matrix

| Arguments | Behavior |
|-----------|----------|
| (none) | Research: official docs + changelog |
| `--configured` | Read current settings files only (no research) |
| `--official-only` | Research: official docs only |
| `--changelog-only` | Research: changelog extraction only |
| `--full-research` | Research: all sources including MCP agents |

### Mode: --configured

When `--configured` is passed, skip all research and read currently configured environment variables from all settings scopes.

#### Settings File Hierarchy (Precedence: Enterprise < User < Project < Local)

| Scope | Path | Description |
|-------|------|-------------|
| Enterprise | `~/.claude/enterprise/settings.json` | IT-managed defaults |
| User/Global | `~/.claude/settings.json` | User preferences |
| Project | `.claude/settings.json` | Project-specific settings |
| Local | `.claude/settings.local.json` | Gitignored local overrides |

#### Workflow

1. Read each settings file if it exists
2. Extract `env` section from each
3. Calculate effective values (later scopes override earlier)
4. Display merged view with source indication

#### Output Format

```markdown
## Currently Configured Environment Variables

| Variable | Value | Source | Overridden By |
|----------|-------|--------|---------------|
| DISABLE_TELEMETRY | "1" | user | project |
| ANTHROPIC_API_KEY | "sk-***" | user | - |
| CLAUDE_HOOK_LOG_DIR | "./logs" | project | - |
| DISABLE_AUTOUPDATER | "0" | enterprise | - |

### By Scope

**Enterprise** (~/.claude/enterprise/settings.json)
- DISABLE_AUTOUPDATER = "0"

**User** (~/.claude/settings.json)
- DISABLE_TELEMETRY = "1"
- ANTHROPIC_API_KEY = "sk-***" (masked)

**Project** (.claude/settings.json)
- DISABLE_TELEMETRY = "0" (overrides user)
- CLAUDE_HOOK_LOG_DIR = "./logs"

**Local** (.claude/settings.local.json)
(none configured)
```

#### Security Note

Mask sensitive values in output:

- `ANTHROPIC_API_KEY` -> `"sk-***"` (show prefix only)
- `AWS_*` credentials -> masked
- `GOOGLE_*` credentials -> masked

### Mode: Research (Default)

When no `--configured` flag, research environment variables from multiple sources.

#### Step 1: Parallel Research via Agents

Launch research agents in parallel to gather env var information from multiple sources.

##### Agent 1: MCP Research (full-research mode only)

Spawn `mcp-research` agent:

```text
Research Claude Code environment variables using perplexity and firecrawl.

Search for:
- "Claude Code environment variables configuration"
- "anthropic claude cli env vars"
- "ANTHROPIC_API_KEY CLAUDE_CODE environment"

Extract ALL environment variables mentioned, including:
- Variable name
- Default value (if mentioned)
- Purpose/description
- Source URL

Return structured findings with source citations.
```

##### Agent 2: Official Documentation

Spawn `claude-code-guide` agent:

```text
First WebFetch https://code.claude.com/docs/en/claude_code_docs_map.md to find
relevant doc pages about environment variables and settings.

Then WebFetch these specific pages:
- Settings/configuration pages
- Environment variables documentation
- Any pages mentioning ANTHROPIC_*, CLAUDE_*, DISABLE_*, ENABLE_*

Do NOT use Skill tool (not available).
Return complete env var tables with source URLs.
```

##### Skill: Local Documentation Cache

Invoke `docs-management` skill with query:

```text
Search for "environment variables" and "env" in Claude Code documentation.
Include any settings, configuration, or customization topics.
```

#### Step 2: Changelog Extraction

WebFetch the CHANGELOG.md to extract environment variables:

```text
URL: https://raw.githubusercontent.com/anthropics/claude-code/main/CHANGELOG.md
```

Extract all patterns matching:

- `ANTHROPIC_*`
- `CLAUDE_*`
- `DISABLE_*`
- `ENABLE_*`
- `*_TIMEOUT`
- `*_API_KEY`
- `HTTP_PROXY`, `HTTPS_PROXY`, `NO_PROXY`

Note the version where each variable was introduced.

#### Step 3: Consolidate and Categorize

Merge findings from all sources, deduplicate, and categorize:

| Category | Criteria |
|----------|----------|
| **OFFICIAL** | Documented on code.claude.com/docs |
| **CHANGELOG** | In CHANGELOG but not in official docs |
| **DISCOVERED** | Found via MCP research only |

#### Step 4: Output Report

Format the complete list as structured tables:

```markdown
## Claude Code Environment Variables

### Official (Documented)
| Variable | Default | Purpose | Category |
|----------|---------|---------|----------|
| ANTHROPIC_API_KEY | (required) | API key for Anthropic API | Auth |
| CLAUDE_CODE_USE_BEDROCK | "0" | Use AWS Bedrock instead of Anthropic API | Provider |
| CLAUDE_CODE_USE_VERTEX | "0" | Use Google Vertex AI | Provider |
| ... | ... | ... | ... |

### Changelog (Undocumented)
| Variable | Default | Purpose | Version Added |
|----------|---------|---------|---------------|
| CLAUDE_BASH_MAINTAIN_PROJECT_WORKING_DIR | "0" | Keep project working directory | v2.x |
| ... | ... | ... | ... |

### Discovered (MCP Research)
| Variable | Default | Purpose | Source |
|----------|---------|---------|--------|
| ... | ... | ... | [source URL] |

---

## Sources
- Official: https://code.claude.com/docs/en/settings
- Changelog: https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md
- MCP: perplexity, firecrawl (if --full-research)
```

### Known Enviro