Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add momentmaker/kaijutsu --skill autopilotgit clone --depth 1 https://github.com/momentmaker/kaijutsuWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/momentmaker/kaijutsu/autopilot)<a href="https://agentmods.dev/skills/momentmaker/kaijutsu/autopilot"><img src="https://agentmods.dev/badge/skills/momentmaker/kaijutsu/autopilot.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00101 | $0.02376 |
| Opus 5 | $0.00051 | $0.01188 |
| Sonnet 5 | $0.00020 | $0.00475 |
| Haiku 4.5 | $0.00010 | $0.00238 |
Grade A, and why
autopilot scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 184 lines — stays where its author put it; the contents beside it link to each section on GitHub.
autopilot v2 — Intent → PR pipeline
Take a natural-language intent and run the full development pipeline to a polished PR. Two gates total: brainstorm approval (Gate 1) + PR review on GitHub (Gate 2 — the existing PR semantics). Multi-agent adversarial review replaces the single-reviewer model from v1.
NOTE: kaijutsu-managed. Re-running
jutsu install autopilotoverwrites this file. The previous v1 SKILL.md (if any) is preserved at~/.claude/skills/autopilot/.archived/SKILL.md.
Command Detection
/autopilot {intent}→ Run mode (start new pipeline)/autopilot resume→ Resume mode (continue interrupted run)/autopilot abort→ Abort mode (clean up + stop)/autopilot status→ Status mode (show current phase)
CLI equivalents: jutsu autopilot {init,status,abort,resume,run}. Use the slash command for interactive runs; the cobra group is for scripted use + CI test mode.
Phase Map
/autopilot {intent}
→ Phase 1: BRAINSTORM (interactive — anti-sycophancy via jutsu swarm dream)
→ ★ GATE 1 (configurable: skip | review) ★
→ Phase 2: SPEC
→ write spec to docs/specs/<date>-<slug>.md
→ swarm doc-review on spec [report-only]
→ record approved-spec path in state file
→ Phase 3: PLAN
→ write plan to IMPLEMENTATION_PLAN.md
→ swarm doc-review on plan [report-only]
→ Phase 4: BUILD (stage 1..N — each followed by polish loop)
→ Phase 5: SHIP
→ final swarm pr-review (full lens)
→ reverse-drift gate (informational, tagged on PR)
→ gh pr create (or test-mode: write planned PR JSON)
→ ★ GATE 2 = PR review on GitHub ★
→ Phase 6: LEARN (state file, findings.db, MEMORY.md)
Cost Cap
Three layers (per spec Decision #3):
- Soft cap — config default $20 (overridable via
.kaijutsu/autopilot.yaml::cost.max_total_usd); at 50% of soft cap, autopilot pauses + asks user to confirm continue. - Hard ceiling —
MaxAutopilotCostUSD = $100constant incli/internal/cli/autopilot.go. Cannot be raised by editing yaml. Per-run override via--max-cost Nworks only up to ceiling. - Per-shell ceiling override — set
KAIJUTSU_AUTOPILOT_HARD_CAP_OVERRIDE=Nin your shell environment to raise the ceiling for THIS shell. Not stored in repo. Defense against malicious yaml.
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 184 lines · 101 tokens per session scan A 17a3d9de5db2
autopilot is a skill published in the GitHub repository momentmaker/kaijutsu (3 stars, last pushed 2mo ago), licensed MIT. It adds 101 tokens to every session and 2,376 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
release-expert
Multi-repo release coordination: version alignment, RC lifecycle, release waves, rollback planning. Use when saying "release", "version alignment", or "cut an RC".
release
Prepare code for release (version bumps, changelog, README updates) and create an annotated tag to trigger the GoReleaser workflow.
agentplane-release-and-packaging-operator
Use when preparing, validating, publishing, auditing, or recovering an Agentplane release, especially package build ordering, version parity, npm/GitHub/GHCR/external distribution publication, public install smoke tests, hosted publish evidence, or release CI failures.
release
Release Aigon — push to origin, cut a version+tag, or publish to npm. Wraps scripts/ship.js with CHANGELOG draft + dry-run preview. Triggers when the user types "/release", "/push", "/ship", or says "ship a release", "publish a beta", "cut a version", "push to origin", "release X to npm".
pypi-release
This skill should be used when releasing tunacode-cli to PyPI. It keeps the existing local release checks, then hands the actual PyPI upload to a GitHub Actions workflow that uses the repository's PYPIAPITOKEN secret.
ship
Ship phase. Runs isolated integration test in a fresh worktree, creates a PR with full spec + audit report in the body, watches CI, and auto-fixes failures.