[Epic]: Autonomy Foundations: Multi-Agent, Simulation, Checkpoints & Budgets (v1.0.0)

Thematic sub-Epic under v1.0.0 Roadmap #654.

Foundational autonomy capabilities that power safe, high-autonomy operation and the endless feed experience.

Scope (initial)

• Multi-agent orchestration + delegation (#644)
• Simulation, dry-run, shadow mode (#645)
• Consistent checkpoint/approval primitive (#648)
• User-configurable budgets and observability (#634, #653)
• Full autonomous loops (e.g. bug resolution #638 with detailed stories; feature #639 with detailed stories + design validation + media)
• Scheduled release cut/finalize/implement + marketing (#641 with stories)
• Related: Epic #470 Autonomy Engine, cost control, risk tolerance in .plate

Acceptance (to be refined)

• Agents can delegate and coordinate with provenance
• Shadow runs surface rich previews for feed checkpoints
• Unified approval mechanism works across surfaces
• Budgets enforced and visible in feed
• End-to-end autonomous bug/feature resolution with human checkpoints where required, including media and design validation
• Scheduled release and marketing ops with feed visibility

Roadmap: #654
Vision pointer: #627
Status: Stub.

Children assigned: #644, #645, #648, #634, #653, #638, #639, #641

Start with roadmap #654 and vision #627, then read the child issues #644, #645, #648, #634, #653, #638, #639, and #641 to understand the proposed autonomy work. The issue is marked as a stub, so completion requires refining the scope and acceptance criteria for delegation, simulation, checkpoints, budgets, autonomous resolution, and scheduled release operations.