Spielewoy/autoprompt-skill

AP Implementer builds one approved feature from a roadmap item or frozen plan using strict TDD and real test runs. It is used for direct implementation work and reports PLAN-CONFLICT if the plan no longer fits.

AP Intake reconstructs old intake artifacts when a run is explicitly resuming legacy state. New runs use the roadmap author instead, so this is only for compatibility reads and translating valid legacy facts.

Ap Arbiter renders a binding decision when Autoprompt forks cannot be resolved by the loop. It is used in unattended runs to choose a conservative path, log the ruling, and continue without escalating to the user.

Ap Janitor verifies the required governance state, writes the DONE sentinel atomically, and removes only approved scratch artifacts after validation. Use it at the end of an Autoprompt run when cleanup must preserve governance files and project evidence.

Ap Juror renders a binary PASS/FAIL sign-off on opened evidence for an Autoprompt run. Use it as a final review seat for mission alignment, coverage, tests, and blocker checks.

Ap Manager coordinates multi-lane Autoprompt slices by building compact handoff briefs and dispatching disjoint L3 work without executing it. Use it when a task needs parallel sibling tracks and strict pointer-based coordination.

Ap Planner builds a concise implementation plan only when a roadmap item needs extra detail, such as debug depth-locks or unresolved design forks. It is used for scoped planning, red repro capture, and test-first verification in the Autoprompt workflow.

Autoprompt turns a mission into one executable roadmap, coordinates dependency-safe lanes, and verifies the result with independent reviewers and tests. Use it only when you explicitly invoke autoprompt mode.

Autoprompt turns an explicit mission into one executable roadmap, dispatches dependency-safe work lanes, and verifies the result with independent reviewers and tests. Use it for coordinated agent runs that need scoped planning, implementation, and fresh verification.

Autoprompt turns a mission into one executable roadmap, dispatches dependency-safe work lanes, and verifies the result with independent reviewers and tests. Use it for explicit-run orchestration rather than ad hoc prompt handling.

Autoprompt turns a mission into one executable roadmap, assigns dependency-safe work lanes, and verifies the result with independent reviewers and real tests. Use it for explicit, useful-first multi-agent orchestration runs.

Autoprompt turns a mission into one executable roadmap, dispatches dependency-safe work lanes, and verifies the result with independent reviewers and tests. Use it for explicit, useful-first multi-agent orchestration in Claude Code.

Autoprompt turns a mission into one executable roadmap, routes dependency-safe work lanes, and verifies the result with independent reviewers and tests. Use it for explicit orchestration runs, not ordinary requests or implicit resumes.

Autoprompt runs the Prime Agent orchestration loop through a native RLM dispatcher for explicit /autoprompt missions. Use it to launch, resume, and route bounded subagents with enforced topology and verification gates.

Autoprompt turns a mission into one executable roadmap, splits work into dependency-safe lanes, and verifies the result with independent review and tests. Use it for explicit mission-driven orchestration in Claude Code.

Autoprompt turns a mission into one executable roadmap, assigns dependency-safe lanes, and verifies the result with independent reviewers and real tests. Use it for explicit, useful-first multi-agent orchestration.

Autoprompt Framework Generator builds a one-off custom framework when a selector returns MISS. It classifies task axes, assembles the gate sequence, emits a validated leaf descriptor, and hands it to the validator before any gate runs.

Autoprompt Preflight Probe verifies RUN, READ, and WRITE on an explicit cache miss and reports live model and effort bindings. Use it to diagnose recovery capability before continuing an Autoprompt run.

Depth Prober derives the deepest cause of a bug from issue text and code, then checks whether a proposed fix reaches that layer. It is used for terminal analysis where shallower causes must be rejected.

Execution Harness Resolver materializes a per-task execution harness by detecting the repo's build system and ingesting or deriving fail-to-pass and pass-to-pass cases. It is used when an agent needs a validated test harness for a task or needs to block on an unresolvable environment.

Feature Coordinator drives approved roadmap lanes through build, review, and verification gates for an Autoprompt run. Use it to coordinate worker dispatch, dependency order, and frontier status without doing the implementation work itself.

Framework Validator checks a generated framework descriptor for HRN-5 soundness before it is driven. It verifies gate mapping, terminal DONE structure, the BLOCKED invariant, and a non-empty acceptance set.

Fresh Verifier checks a candidate roadmap or plan against the mission and repository, then returns APPROVE or REJECT. It is used for blind validation with explicit failure by default when requirements, tests, or evidence are missing.

Goal Checker verifies whether a delivered result satisfies the original mission text, checking scope, evidence, coverage, and potential flaws. Use it when you need an adversarial pass/fail review of an agent run.

Re-Anchor confirms mission and roadmap alignment after resume or compaction using the three-file governance state. It checks pointers, gate logs, and frontier evidence before reporting ALIGNED or DRIFT.