From angreal
Guides writing effective ToolDescriptions for Angreal tasks to instruct AI agents, covering prose structure, risk levels, usage scenarios, and examples.
npx claudepluginhub angreal/angreal --plugin angrealThis skill uses the workspace's default tool permissions.
Write effective ToolDescription prose to guide AI agents using your tasks.
Creates isolated Git worktrees for feature branches with prioritized directory selection, gitignore safety checks, auto project setup for Node/Python/Rust/Go, and baseline verification.
Executes implementation plans in current session by dispatching fresh subagents per independent task, with two-stage reviews: spec compliance then code quality.
Dispatches parallel agents to independently tackle 2+ tasks like separate test failures or subsystems without shared state or dependencies.
Write effective ToolDescription prose to guide AI agents using your tasks.
When AI agents discover tasks via angreal tree --long, they see:
The ToolDescription is the primary guidance for AI agents deciding whether and how to use your task. Think of it as a mini-prompt that teaches agents when and how to use your tool.
import angreal
@angreal.command(
name="deploy",
about="Deploy to environment",
tool=angreal.ToolDescription(
"""
Deploy the application to a specified environment.
## When to use
- After successful build and tests pass
- When user explicitly requests deployment
## When NOT to use
- Directly from feature branches
- Without running tests first
## Examples
```
angreal deploy --env staging
angreal deploy --env production --version v1.2.3
```
""",
risk_level="destructive"
)
)
def deploy(env, version=None):
pass
angreal.ToolDescription(
description, # Prose description (required)
risk_level="safe" # "safe", "read_only", or "destructive"
)
| Level | Meaning | Use When |
|---|---|---|
safe | No destructive effects | Default. Build, test, lint tasks |
read_only | Only reads/reports | Status checks, info gathering |
destructive | May modify or delete | Deploy, clean, database migrations |
Use consistent markdown sections:
tool=angreal.ToolDescription(
"""
One-line summary of what the tool does.
## When to use
- Scenario 1
- Scenario 2
## When NOT to use
- Anti-pattern 1
- Anti-pattern 2
## Examples
Concrete invocation examples with real arguments
## Preconditions
What must be true before running
## Output
What to expect from the tool
""",
risk_level="safe"
)
Bad:
tool=angreal.ToolDescription("Runs tests")
Good:
tool=angreal.ToolDescription(
"""
Run the complete test suite including unit and integration tests.
## When to use
- Before committing changes
- After pulling changes to verify environment
- As part of CI/CD validation
## Output
Returns exit code 0 if all tests pass, non-zero otherwise.
"""
)
tool=angreal.ToolDescription(
"""
Build documentation from source.
## Examples
```
# Build to default output
angreal docs build
# Build with specific output path
angreal docs build --output ./site
# Build in watch mode
angreal docs build --watch
```
"""
)
tool=angreal.ToolDescription(
"""
Install project dependencies.
## Common Failures
- **Network error**: Check internet connectivity
- **Permission denied**: May need sudo or virtual environment
- **Version conflict**: Check requirements.txt
## Recovery
If installation fails:
1. Clear cache: `angreal dev clean-cache`
2. Retry installation
"""
)
tool=angreal.ToolDescription(
"""
Run unit tests only (fast).
## Related Tasks
- `test.integration` - Integration tests (slower)
- `test.all` - Complete test suite
- `lint.check` - Check style (run before tests)
## Typical Workflow
1. `angreal lint check`
2. `angreal test unit` - Fast feedback
3. `angreal test all` - Full validation
"""
)
tool=angreal.ToolDescription(
"""
Build the project for release distribution.
## When to use
- Creating a release build
- Before running deployment tasks
## Preconditions
- Dependencies installed (`angreal dev check-deps`)
- Source compiles without errors
## Output
- Artifacts written to `dist/`
- Returns 0 on success
""",
risk_level="safe"
)
tool=angreal.ToolDescription(
"""
Apply pending database migrations.
## When to use
- After pulling changes with new migrations
- During deployment to update schema
## When NOT to use
- On production without backup
- Without reviewing migration contents
## Recovery
If migration fails:
1. Check migration logs
2. Run `angreal db rollback`
3. Fix and retry
""",
risk_level="destructive"
)
tool=angreal.ToolDescription(
"""
Check project status and health.
## When to use
- Starting work to understand state
- Debugging to gather context
- Verifying environment setup
## Output
Reports on git branch, dependencies, build state, test summary.
""",
risk_level="read_only"
)