cperler

batch-plan

Turn a pile of ALREADY-FILED, independently-authored GitHub issues into a validated, dependency-ordered batch plan and apply it to a run. You run the analysis — which issue depends on which, which lane fits each — over the real issue bodies; deterministic `orchestrator batch-plan` code validates and applies. The producer for existing issues (spec-intake is the producer for new work).

brainstorm

Turn a fuzzy area or goal into a ranked shortlist of concrete, scored ideas and hand the human's picks off — small ones filed as standalone enhancement issues, larger ones fed to spec-intake. You run the divergent exploration and the idea authoring over the real codebase, backlog, and run history; deterministic `orchestrator brainstorm` code ranks and files. The front door ABOVE intake — ideas out, not issues-from-a-known-idea and not code.

new-project

Stand up a brand-new project for the harness end to end and leave it with work queued — interview, create the repo skeleton, create the GitHub repo, generate and verify the project adapter, then carry the idea through brainstorm (if fuzzy) and spec-intake into a filed, dependency-ordered set of issues. Phases 0-2 in one session, for a project that does not exist yet; use adapter_bootstrap_skill.md instead when the repo already exists.

orchestrate-batch-headless

Drive a multi-task batch on the headless×claude lane — the default. One `run-headless` driver owns the run and the engine's Scheduler spawns `claude -p` per stage, so no stage prompt or output passes through the session. Use for ordinary batches; the interactive lane is only for watching stages live.

orchestrate-batch-interactive

Drive a multi-task batch through the engine scheduler on the interactive×claude lane. Thin supervisor — the DAG/retry/cascade/capacity logic is all in the `orchestrator` engine; this skill only sequences the fan-out.

orchestrate-task-interactive

Drive one task through the engine on the interactive×claude lane. Thin supervisor — all deterministic logic lives in the `orchestrator` CLI; this skill only sequences calls.

spec-intake

Turn a raw idea into a validated, dependency-ordered spec and file it as a batch of GitHub issues that feed the batch lane. You run the conversation and the decomposition; deterministic `orchestrator spec` code validates, plans, and files. The front door above intake — issues out, not code.

triage-followups

After a run completes, walk the GitHub issues that run auto-filed (review non-blocking findings + improvement ideas from the evidence-out seam, AND any issue a task filed for work it deliberately cut) ONE AT A TIME with the human — explaining each from its source finding/rationale and the code it points at — and decide keep / close / promote / edit. The human triage gate the auto-filing seams deliberately do not have. Runs after any completed run (batch or single-task); re-runnable later.