eda-chief
Legacy compatibility shim for historical eda-chief calls. Use when old prompts/scripts still invoke eda-chief, then dispatch to the repo-level agent policy and execution skills.
EDA Chief
Role
eda-chief is no longer the primary orchestrator.
It is a compatibility layer for older workflows that still call this skill name.
Top-level governance is defined in AGENTS.md at repo root.
What to do when invoked
- Load
AGENTS.mdand follow its global routing/governance policy. - Convert legacy "chief-style" request into a scoped execution brief.
- Dispatch execution to
eda-loop. - Add control skills only when needed:
eda-theory-vetofor high-cost/high-risk proposals.eda-retroafter experiment batches that need next-step decision.
- Keep outputs short and artifact-driven.
Hard boundaries
- Do not redefine global entry policy inside this skill.
- Do not duplicate global recursion/self-update policy from
AGENTS.md. - Prefer domain skills for domain reasoning and implementation.
- Mention compatibility mode explicitly when this skill is used.
References
Load only when needed:
references/delegation-matrix.mdreferences/autonomy-guardrails.md