Skip to content

Commit f1b9c5e

Browse files
authored
chore(guardrails): update AGENTS.md with commands and agents documentation (#91)
Contributes to #51. Documentation update for all profile commands and agents.
2 parents 952d382 + e21b3fb commit f1b9c5e

1 file changed

Lines changed: 31 additions & 0 deletions

File tree

‎packages/guardrails/profile/AGENTS.md‎

Lines changed: 31 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -11,3 +11,34 @@
1111
- Use `implement` as the guarded default primary agent. Route review, ship, and handoff work through the packaged `/review`, `/ship`, and `/handoff` commands instead of freeform release flows.
1212
- Keep review paths read-only. If a workflow needs edits, return to `implement` or a project-local implementation agent instead of widening the review agent.
1313
- All configured providers are available for standard work. The `provider-eval` agent and `/provider-eval` command remain available for dedicated evaluation workflows.
14+
15+
## Commands
16+
17+
| Command | Description |
18+
|---------|-------------|
19+
| `/review` | Run a read-only code review on the current diff or PR. |
20+
| `/ship` | Merge-ready workflow: CI check, review gate, and push. |
21+
| `/handoff` | Generate a handoff document for cross-session continuity. |
22+
| `/plan` | Analyze requirements, assess risks, and produce a phased implementation plan. |
23+
| `/investigate` | Systematic debugging with root cause analysis — spawns an exploration subagent. |
24+
| `/test` | Run the TDD workflow: RED, GREEN, IMPROVE, then verify coverage. |
25+
| `/delegate` | Route work to parallel subagents or Codex CLI based on task shape. |
26+
| `/provider-eval` | Dedicated evaluation workflow for comparing configured providers. |
27+
28+
## Agents
29+
30+
### Primary agents
31+
32+
| Agent | Description |
33+
|-------|-------------|
34+
| `implement` | Default guarded agent for all implementation work. Edits, tests, and commits code within permission boundaries. |
35+
| `provider-eval` | Evaluation-only agent for benchmarking and comparing LLM providers. |
36+
37+
### Subagents
38+
39+
| Agent | Trigger | Description |
40+
|-------|---------|-------------|
41+
| `planner` | `/plan`, complex feature requests | Read-only planning agent. Produces phased plans with risk assessment without modifying code. |
42+
| `investigate` | `/investigate`, debugging tasks | Deep exploration subagent. Reads code, traces data flow, and identifies root causes without edits. |
43+
| `security` | `/review` (security scope), OWASP checks | Security-focused review subagent. Scans for OWASP Top 10 vulnerabilities, credential leaks, and injection risks. |
44+
| `code-reviewer` | `/review`, PR review pipeline | Read-only review agent. Analyzes diffs for quality, correctness, and style issues. |

0 commit comments

Comments
 (0)