Setup Workflow

Scaffold the per-repo configuration the CONTEXT-driven spec workflow assumes. Local markdown under docs/specs/ is the only home of planning artifacts — there is no external tracker.

• Spec artifacts — docs/specs/<feature-slug>/, read and written by write-idea, write-prd, write-techspec, write-tasks, implement-task, implement-spec, qa-gate, and archive-spec
• Spec routing — how an agent picks the pipeline entry point for a given change (large initiative / feature / refactor-bugfix / trivial) and what marks a spec done
• Domain docs — CONTEXT.md (glossary) and docs/adr/, and the consumer rules for reading them
• Triage labels (conditional) — only when the repo receives external/incoming issues on its forge (e.g. a public GitHub repo) that the triage skill will process
• Autonomous work model (conditional) — only when the repo delegates implementation to agent runtimes (e.g. through Roundfix): the Supervisor/implementer split (the Supervisor orchestrates and authors Specs; implementation goes to an ACP Runtime), runtime routing, and the hard rule that makes the split binding

These usage rules are seeded into the repo as docs/agents/*.md — the canonical, always-current explanation of how agents work inside the CONTEXT-driven workflow. This skill owns those files: a re-run regenerates them.

This is a prompt-driven skill, not a deterministic script. Explore, present what you found, confirm with the user, then write.

Process

1. Explore

Look at the current repo to understand its starting state. Read whatever exists; don't assume: