vico-skills

Execute the current Vico plan continuously until completion or a real blocker is reached. Use when the user wants the agent to keep going without stopping after each small step, carry out an active plan under `.vico/plans/active/` item by item, continue from temporary reconcile state when needed, keep going until complete, or follow a Ralph-like persistent execution loop. For Claude Code, this skill can be paired with bundled hooks or a repo-local runner loop; for Codex, use the same execute loop directly.

Turn user feedback about vico-skills into a GitHub issue draft, check for likely duplicates when useful, and create the issue only after explicit user confirmation. Use when the user wants to report a bug, suggest a feature, describe workflow friction, complain about naming or UX, or ask to file an issue against vico-skills.

Build just enough shared ground to choose a safe next route before planning or execution. Use when the user wants to orient in a repo, inspect a design, clarify intent or scope, pressure-test a proposal, or hand current understanding forward into vico-plan.

Repo-local operations surface for tracked Vico maintenance. Use when the user wants to bootstrap tracked docs, sync stale `.vico` state, close or cancel tracked work, extract durable truth, or validate the current Vico workspace.

Default front door for Vico-style repo-local planning. Decide whether work should stay `no-doc`, become `plan_only`, or upgrade to `prd_backed`; reconcile active state when needed; then create or update the active plan under `.vico/plans/active/`. Use when the user wants to start tracked work, make a plan, create a tracked plan, continue tracked work, recover from Vico drift, or turn grounded decisions into an executable plan.