CLI guide, executor, and automator. Use when asking about agentic-agent commands, workflow selection, task management, SDD methodology, or how to start/continue/complete work.
Guides you through agentic-agent CLI commands, helps select the right workflow (Tiny/Small/OpenSpec+/Full SDD) based on risk, and automates context generation and validation.
Ask: "What are you building and how risky is it?"
| Signal | Workflow | Setup | Key Commands |
|---|---|---|---|
| Bug fix, < 1 day, 1 file |
| Tiny |
| 5 min |
task create, claim, complete |
| Feature, 1-2 weeks, 1 service | Small | 10 min | openspec init, tasks |
| Multi-package, monorepo | OpenSpec+ | 20 min | openspec init per package |
| Payment/auth/PII/contract | Full SDD | 1+ hour | sdd start --risk critical |
Risk escalators (each one = increase tier):
See resources/workflow-commands.md for full command sequences for each tier (TINY, SMALL, OpenSpec+, Full SDD) with examples.
Before editing any directory:
AGENTS.md (Read-Before-Write rule)agentic-agent context generate <DIR>Before completing any task:
agentic-agent validate
After architectural changes:
AGENTS.md files immediatelyblocked_by is non-emptytask claim (loses traceability).agentic/ YAML files directly| Symptom | Cause | Fix |
|---|---|---|
| Gate fails | Missing Source or empty section | Fix spec, re-run gate-check |
blocked_by non-empty | Unresolved ADR | sdd adr resolve <id> |
| Validation fails | Context stale or scope violated | Run context generate, validate again |
| "Which workflow?" | Risk not assessed | Walk the decision tree above |
→ resources/workflow-commands.md — Full command sequences for each tier with examples, error recovery details, and workflow selection flow