Agents Overview
Agents are complete, self-contained jobs. Each agent knows exactly what to do — you don't describe the task, you just pick the agent and run it.
Agents vs Skills
| Agents | Skills | |
|---|---|---|
| You provide | Just the agent name | A task description |
| Agent knows | What to do and how | Only how to approach what you told it |
| Example | nimbus run --agent test-coverage | nimbus run --skill add-tests "write tests for auth.py" |
Use an agent when you want a complete, well-defined job done on your codebase. Use a skill when you have a specific task in mind and want Nimbus to approach it a particular way.
List agents
nimbus agents
Filter by category:
nimbus agents --category security
Get agent details
nimbus agents --info security-audit
Run an agent
nimbus run --agent security-audit
You can also run agents against the hosted backend:
nimbus run --agent test-coverage --backend https://api.get-nimbus.com
Agent categories
| Category | Count | What they do |
|---|---|---|
security | 3 | Find and fix vulnerabilities, leaked secrets, CVEs |
quality | 5 | Type safety, error handling, dead code, complexity |
testing | 3 | Coverage, integration tests, mutation testing |
docs | 3 | OpenAPI docs, README, inline docstrings |
performance | 2 | N+1 queries, async conversion |
infra | 2 | CI/CD pipelines, Docker hardening |
architecture | 2 | Feature flags, observability |
Total: 20 built-in agents