Using TOOLNAME CLI
Overview
Execute [operations] using the TOOLNAME CLI. Assumes CLI is pre-configured (auth handled). Focus on [core principle — e.g., efficient exploration, safe operation, avoiding context overflow].
Core principle: [The single most important thing agents must remember — e.g., "Always limit results" or "Check existing state before creating resources"]
When to Use
- •[Specific use case 1]
- •[Specific use case 2]
- •[Specific use case 3]
Workflow
1. Context Check
[What should agents verify before doing anything? Current project/account/database? Auth status?]
2. Discovery
[How should agents explore available resources before operating? List/search commands]
3. Execution
Defaults:
- •[Default output format — almost always JSON for agents]
- •[Default safety measure — e.g., LIMIT for queries, --dry-run for mutations]
- •[Other key defaults]
For large results:
- •[How to handle — export to file, paginate, aggregate]
4. Output Handling
| Format | When to Use |
|---|---|
| JSON | [Default — reliable for any data shape] |
| TABLE | [Only when — e.g., narrow results, human review] |
| CSV | [When — e.g., export, downstream processing] |
CLI Reference
| Domain | Reference |
|---|---|
| TOOLNAME | See references/TOOLNAME-cli.md |
Common Mistakes
[Mistake 1 — e.g., "Unbounded query"]: [What goes wrong] Fix: [What to do instead]
[Mistake 2 — e.g., "Wrong output format"]: [What goes wrong] Fix: [What to do instead]
[Mistake 3 — e.g., "Skipping exploration"]: [What goes wrong] Fix: [What to do instead]