dev-loop

Plugin dev-loop from marketplace karlorz-agent-skills in karlorz/agent-skills. Commands read from the repo's manifest — 16 days ago.

Skills (7)

• dev-loop

  Run a project dev cycle: implement, investigate, prep, status, office-hours, setup, or dashboard. Use for /dev-loop or config-lint.

• investigate

  Proactive project-health scan that queues schema-valid vault findings. Use for /dev-loop investigate or finding work.

• office-hours

  Use when running an attended dev-loop office-hours checkpoint before prep, promotion, or /goal to choose one current-project or cross-project topic and capture requirements.

• ranked-audit

  Use when /dev-loop ranked-audit performs an unattended, read-only audit of ranked project work and prepares evidence for later office-hours reconciliation.

• research

  Sonnet research agent for dev-loop code and vault health. Use when /dev-loop research should scan and rank work items.

• setup-dev-loop

  Scaffold per-repo dev-loop config and project glossary. Use once per repo before /dev-loop, or for setup-dev-loop.

• status

  Read-only preview of the next dev-loop cycle with no vault or git writes. Use for /dev-loop status or doctor.

Subagents (7)

• ci-health-worker

  Use this agent when you need a CI health check — querying GitHub Actions API for recent workflow runs, verifying required status checks, and reporting CI pipeline health. Typical triggers include dev-loop IDLE DISCOVERY CI monitoring, checking whether required checks are failing on the release branc

• codex-review-worker

  Independent Codex review of a working-tree diff. Prefers native `codex review --uncommitted`; Claude companion is fallback only. Complements simplify:simplify.

• doctor-worker

  Use this agent when you need a dev-loop dependency drift check and auto-compact firing count probe — reading skills/dev-loop/dependencies.yaml plus the current session's ~/.claude/projects/<slug>/<uuid>.jsonl, then emitting a structured JSON health classification. Typical triggers include dev-loop R

• research-worker

  Use this agent when you need a code and vault health research scan — scanning for coverage gaps, skills drift, spec drift, unpushed commits, raw coverage, cross-link density, page quality, and type coverage. Typical triggers include dev-loop IDLE DISCOVERY research cycles, pre-sprint health assessme

• sdd-execute-worker

  Use this agent when dev-loop wants an isolated adapter for superpowers:subagent-driven-development during EXECUTE. The worker reads and follows the upstream skill as source of truth, while preserving dev-loop's sonnet-pinned execution-subagent policy. Falls back to inline execution guidance only whe

• simplify-worker

  Use this agent when dev-loop needs an isolated adapter for the required simplify:simplify code-review pass. The worker reads and follows simplify:simplify as source of truth, using its local minimum contract only when the skill file cannot be resolved. Falls back to inline simplify:simplify only whe

• status-worker

  Use this agent when dev-loop MODE=status needs an isolated read-only probe — running scripts/dev-loop-status.js and returning dev-loop-status.v1 JSON without vault/git/PR/release writes. Typical triggers include /dev-loop status, /dev-loop doctor alias, or Codex spawn_agent parity for the status pip

Similar plugins