| name | aa-resume |
| description | Use when the user says resume, continue, or pick up where we left off, starts a new session on an existing Agent-Agile project, or is recovering after a crash or context reset, for example "/aa-resume", "let's continue", "where were we". |
| license | MIT |
| metadata | {"system":"agent-agile"} |
aa-resume
Resolve the Agent-Agile playbook root: use the first of these that exists — (1) ${CLAUDE_PLUGIN_ROOT}/playbooks, (2) ./.claude/agent-agile/playbooks, (3) ./.agents/agent-agile/playbooks, (4) ~/.claude/agent-agile/playbooks, (5) ~/.agents/agent-agile/playbooks, (6) ./playbooks.
Read playbooks/execution.md's "Resume protocol" section from that root, and follow it exactly:
- If
.planning/STATE.md does not exist, no project has been started here yet. Say so plainly, and point the user at the three front doors instead of guessing at a state that isn't there: a vague spark → /aa-grill, a clear idea → /aa-new-project, an existing plan or spec → /aa-import.
- Otherwise, read
.planning/STATE.md first and only. Don't re-derive status by re-reading the whole epic history — STATE.md is the baton, kept small on purpose so this step stays cheap every time.
- Follow its pointers — at most three files — and read exactly those, nothing wider.
- Continue the work from exactly where
STATE.md says the run is: the phase it names (grill, intake, panel, prereqs, an epic's wave, review, or done), the "Now" sentence, and the "Next" sentence.
- If
STATE.md names a blocker, resolve or surface that blocker before doing anything else — never work around it silently.
Close the session, without exception, with one STATE.md update: what's in progress, what's next, the pointers the next session needs, and any blocker. A session that stops without this update leaves the next session to reconstruct state from scratch — the entire point of the baton is that this never has to happen.
