agent-launcher-skills

Plugin agent-launcher-skills from marketplace claude-code-skills in alirezarezvani/claude-skills. Commands read from the repo's manifest — today.

Skills (6)

• agent-launcher-orchestrator

  Use when someone wants to build, launch, grade or schedule a Claude Managed Agent in their own Anthropic account ('build me an agent', 'run this nightly'). Reads the session goal, routes with goal_router.py and compiles the loop with loop_compiler.py.

• grade-iterate

  Use when a launched managed agent needs grading against a rubric and improving ('grade my agent', 'is it good enough'). Builds the outcome with outcome_builder.py, reads verdicts with verdict_reader.py and runs held-back evals from eval_scaffold.py.

• interview

  Use when someone has an agent idea to scope before launch ('help me scope an agent'). Interviews them about the one job and writes a build sheet with build_sheet_builder.py, checked by primitives_validator.py; no API key needed.

• run-without-you

  Use when a graded managed agent should run on a schedule, on an event trigger or on demand ('run it every morning'). Validates cron with cron_validator.py and builds the deployment with deployment_builder.py.

• stage-launch

  Use when a managed-agent build sheet is ready to launch ('launch it', 'deploy the agent'). Generates and validates API payloads with payload_generator.py and payload_validator.py and writes a resumable curl script with launch_script_writer.py that uses the user's own key.

• wrap-up

  Use when closing out a launched managed agent ('wrap up', 'what do I own now'). Inventories primitives with primitives_inventory.py, regenerates the overview with overview_page.py and suggests next upgrades with upgrade_suggester.py.

Slash commands (8)

• /agent-launcher-skills:cs-goal

  Set, show, or advance the per-session agent-launcher goal (./my-agent/goal.json) — the through-line of a CMA launch. Backs the opt-in SessionStart hook. Subcommands map to goal_state.py init/set/status/advance.

• /agent-launcher-skills:cs-grade

  Phase 3 — the bounded grade→iterate loop. Define a CMA outcome (required rubric, max_iterations 1..20), read each grader verdict, decide the next move, and run held-back eval cases once a version passes, via the grade-iterate skill. Never unbounded.

• /agent-launcher-skills:cs-grill-agent-launcher

  Matt Pocock docs-anchored grill for an agent-launcher goal — walks the phase's forcing questions ONE at a time, each with a recommended answer and a citation to a reference doc, refusing to advance on a fuzzy input. Use to pressure-test a CMA plan before launching.

• /agent-launcher-skills:cs-interview

  Phase 1 — interview the founder into a validated CMA build sheet (primitives table + v1/v2 deferrals + eval plan) via the interview skill. No API key needed. Walks the six intake slots, maps to primitives, validates against CMA limits.

• /agent-launcher-skills:cs-launch

  Main entry / resume for building a Claude Managed Agent. Runs the agent-launcher-orchestrator skill from the current session goal — reads ./my-agent/goal.json, routes to the right phase (interview → stage-launch → grade-iterate → run-without-you → wrap-up), compiles the loop/workflow, and forks to t

• /agent-launcher-skills:cs-run-without-you

  Phase 4 — make the agent run without you. Turn a graded agent into a recurring POSIX-cron scheduled deployment (optionally self-grading each firing), an event-driven curl trigger, or on-demand use, then finalize NEXT-DIRECTIONS, via the run-without-you skill. Tests with a manual run before trusting

• /agent-launcher-skills:cs-stage-launch

  Phase 2 — turn a build sheet into exact API payloads and a resumable BYOK curl launch script, then launch (environment → agent → session → kickoff) with the founder's own key via the stage-launch skill. No tool makes API calls; the key never enters chat.

• /agent-launcher-skills:cs-wrap-up

  Close out a launched Claude Managed Agent — recap every primitive owned, regenerate the single-file overview page, and suggest the next 1–2 upgrades, via the wrap-up skill. The last stop before phase=done.

Subagents (4)

• cs-agent-deployer

  Phase-4 specialist for making a Claude Managed Agent run without you. Turns a graded agent into a recurring POSIX-cron scheduled deployment (optionally self-grading each firing), an event-driven curl trigger, or confirmed on-demand use, then finalizes NEXT-DIRECTIONS. Invoke for phase=run-without-yo

• cs-agent-grader

  Phase-3 specialist for the bounded grade→iterate loop when building a Claude Managed Agent. Defines a CMA outcome (required rubric, max_iterations clamped 1..20), reads each grader verdict, decides the next move (sharpen / re-run / escalate / promote), and runs held-back eval cases in parallel once

• cs-agent-interviewer

  Phase-1 specialist for building a Claude Managed Agent — interviews the founder through the six intake slots (job, trigger, inputs, actions, definition-of-done, recurrence) and produces a validated build sheet (primitives table + v1/v2 deferrals + eval plan) without needing an API key. Invoke for ph

• cs-agent-launcher-orchestrator

  Session-goal router for building Claude Managed Agents. Reads ./my-agent/goal.json, routes deterministically to a phase skill (interview → stage-launch → grade-iterate → run-without-you → wrap-up) via goal_router.py, and compiles the goal+phase into a single-pass workflow, a bounded grade→iterate lo

Hooks (2)

• SessionStart

  python3 ${CLAUDE_PLUGIN_ROOT}/hooks/session_start.py

• SessionEnd

  python3 ${CLAUDE_PLUGIN_ROOT}/hooks/session_end.py

Similar plugins