chainlisted

# Chain — YAML-defined multi-step skill orchestration When invoked, **the active primary IS the chain runtime**: parse the chain YAML, execute each step in declared order (delegating to the skill named in `runner: skill` steps or to the active primary's own reasoning for `runner: <primary>` steps), validate each step's output against its declared schema, evaluate gates, write the JSONL audit log, present the consolidated result. This is a single-session, active-primary runtime. Chain YAML is supplied by the operator or current project and uses package-neutral step references of the form `agent-collab:<skill>@<version>`. Reject unsupported fields rather than consulting an external private schema. ## When to use - **The user explicitly asks for it** — "run the X chain," "execute the X chain on Y," "chain these skills," "run this skill sequence," "run chain," "execute validate-<topic>." - **A multi-step verification workflow is already documented as a chain YAML** in `chains/` (production) or `drafts/sample-chains/` (in-progress prototypes). - **High-stakes repeating work** where ad-hoc skill invocation risks inconsistency — release validation, change-management approval gating, compliance sweeps, regulated-workflow execution. - **the active primary is about to manually invoke 3+ skills in sequence** on the same artifact, where the same sequence is likely to repeat. Codify it as a chain up-front; future runs get reproducibility + audit trail by construction. ## When to skip