Context
AgentAPI++ extends a coding-agent API with upstream attribution and explicit agent lifecycle boundaries.
Problem
Agent APIs need clear scope boundaries to prevent runaway execution while maintaining useful autonomy.
Architecture
Extension layer wrapping agent API calls with scope tracking, attribution recording, and execution boundary enforcement.
Key decisions
Agents operate inside declared boundaries, and upstream attribution keeps provenance intact.
Constraints
Agent autonomy must be balanced against safety constraints without breaking real workflows.
Verification
Scope is defined by repository evidence and upstream attribution to coder/agentapi.
Retrospective
Agent tooling needs explicit scope boundaries and provenance tracking.
Fork of coder/agentapi.
Provenance & Attribution
- Evidence source
- Repository documentation and project history
- Provenance
- Fork of coder/agentapi.
- Available in lens
- engineering