// Recover active change and next action from artifacts. Use on fresh session with existing work.
Product go/no-go on a framed spec. Use after auto-frame, before planning.
Engineering go/no-go on a plan. Use after auto-plan, before execution.
Implement approved plan slices. Use as the execute-stage entry point.
Bound and de-risk a request into SPEC.md. Use when the objective is clear but scope needs constraining.
Sharpen a vague idea into a bounded objective. Use before framing when scope is undefined.
Build project truth from repo evidence. Use when steering is missing or stale.
| name | auto-resume |
| description | Recover active change and next action from artifacts. Use on fresh session with existing work. |
| metadata | {"stage":"resume"} |
Session recovery. Rebuilds context from durable artifacts, not memory or guessing.
First action: run scripts/get-context.mjs → JSON {activeChange, stage, canonicalSpec, canonicalDesign, canonicalPlan, productReview, engineeringReview, diagnostics} (missing state normalizes to "none"/null). If any diagnostic has level "error", stop and report it before proceeding.
auto-resume rebuilds context from durable artifacts, not from the user's description or the agent's training data. It does not modify artifacts, advance the stage, or start new work. It loads canonical artifacts in dependency order (spec first, then design, then plan) and reports what it found, what was blocked, and what comes next.
Loading discipline: start with artifacts needed for the current stage. Read project files when understanding the codebase helps rebuild accurate context for the next action.
Before producing the recovery summary:
references/quality.md (~36 lines: anti-patterns, better shape, prose hygiene scan patterns) when the summary becomes narrative recap.Read .agent/steering/STATUS.md. Read references/ARTIFACT-LIFECYCLE.md for recovery order, stale-pointer handling, and stage handoffs. If the recovered state suggests the active change is complete or no active work exists, read .agent/steering/ROADMAP.md when it exists to surface pending phases as context, not to auto-start them.
If .agent/ does not exist or current.json is missing, recommend auto-onboard and stop.
Check that canonical pointers in current.json resolve to actual files:
canonical_spec → does SPEC.md exist?canonical_design → does DESIGN.md exist?canonical_plan → does PLAN.md exist?If any pointer is stale (file missing or moved), report it plainly. Recommend auto-onboard if steering is missing, or auto-frame / auto-plan if the specific artifact is missing.
Load artifacts in this order. Stop at the current stage; do not load artifacts from future stages.
Stage: frame → Load INTAKE.md (if exists), SPEC.md
Stage: plan → Load SPEC.md, then DESIGN.md (if exists), then PLAN.md
Stage: execute → Load SPEC.md, DESIGN.md (if exists), PLAN.md, current slice
Stage: verify → Change complete; load PLAN.md only if reporting what was verified
Stage: resume → Load SPEC.md, STATUS.md
If current.json and STATUS.md disagree on active change or stage, report the mismatch. Prefer current.json for recovery, but surface the discrepancy.
If current.json contains product_review or engineering_review, read the corresponding ## Review: sections from canonical artifacts and include them in the resume summary.
Produce a concise summary:
**Active change:** [name]
**Stage:** [frame|plan|execute|verify|resume]
**Artifacts loaded:** [list]
**What was done:** [1-2 sentences]
**What was blocked:** [1-2 sentences, or "nothing"]
**What comes next:** [specific next action, or "none - change complete"]
**Review verdicts:** [product: X, engineering: Y, or "none"]
**Missing state:** [list or "none"]
**Roadmap:** [N pending / M total, or "not tracked"]
Keep it under 200 tokens. The goal is orientation, not transcription.
Based on the recovered state:
frame with intake but no spec → auto-frame (intake survives)frame with no spec and no intake → auto-frameframe with spec but no product review → auto-ceo-reviewplan with no plan → auto-planplan with plan but no engineering review → auto-eng-reviewexecute → auto-executeverify → change complete; report completion. If ROADMAP.md has pending items, surface them as optional future work, but do not recommend auto-office-hours unless the user explicitly asks to start the next phase.resume with missing steering → auto-onboard.agent/.automaton/state/current.json is read-only for auto-resume; stale pointers are reported, not silently repairedwarning in the summary; error-level diagnostics block the resumeauto-onboard.auto-office-hours handoff.Read references/recovery-scenarios.md for common recovery situations. (~41 lines: 8 state→action pairs covering fresh session, no active change, stale pointers, current.json/STATUS.md mismatch, review verdict blocks, scaffold-level steering, multiple changes, stale status prose.)
Read references/artifact-order.md for the full artifact dependency graph. (~48 lines: ASCII dependency graph from REPO-MAP through PLAN, loading rules by stage in table form, 3 anti-patterns.)
Read references/CONTEXT-BUDGET.md for progressive loading and degradation tiers. (~76 lines: 4 principles, 6-step loading order, 4 degradation tiers with behavioral rules, no-re-read rule with exceptions.)