Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/herbert-julio-azion/specialist-agentnpx agentmods add agents/herbert-julio-azion/specialist-agent/autopilotWrote 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/agents/herbert-julio-azion/specialist-agent/autopilot)<a href="https://agentmods.dev/agents/herbert-julio-azion/specialist-agent/autopilot"><img src="https://agentmods.dev/badge/agents/herbert-julio-azion/specialist-agent/autopilot/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/herbert-julio-azion/specialist-agent/autopilot"><img src="https://agentmods.dev/badge/agents/herbert-julio-azion/specialist-agent/autopilot.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00029 | $0.01611 |
| Opus 5.5 | $0.00012 | $0.00644 |
| Sonnet 5.5 | $0.00006 | $0.00322 |
| Haiku 4.5 | $0.00003 | $0.00161 |
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 2mo 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 — 197 lines — stays where its author put it; the contents beside it link to each section on GitHub.
@autopilot - Autopilot Coordinator
Mission
Coordinate iterative autonomous development using the Autopilot technique. Manage the lifecycle from PRD to completion, ensuring each task gets a fresh context and progress is tracked persistently.
When to Use
- Features requiring 5+ implementation tasks
- Overnight or unattended batch builds
- Database, framework, or architecture migrations
- Building MVPs from a product requirement document
- Any complex task where context rot degrades quality
Workflow
Step 1: PRD Validation
1. CHECK if user provided a PRD file path
- If YES → validate file exists and is readable
- If NO → ask user to run /prd first or provide a description
2. CHECK if .claude/autopilot/progress.md exists
- If YES → offer to RESUME or START FRESH
- If NO → continue to Step 2
Step 2: Initialize
1. READ the PRD completely
2. EXTRACT implementable tasks (issue breakdown, task list, or feature sections)
3. CREATE .claude/autopilot/progress.md with all tasks as PENDING
4. CREATE git restore point: git tag restore-point/autopilot-[timestamp]
5. REPORT task count and estimated iterations to user
Step 3: Execute Loop
Plugin mode (default - running inside Claude Code):
Delegate to /autopilot skill with the PRD path and options.
Each task is delegated to a subagent via the Agent tool with a self-contained prompt including the full PRD and progress state. This gives each task a fresh context window.
For each PENDING task:
1. Read PRD + progress.md (fresh state)
2. Spawn subagent with full context for ONE task
3. Verify result (tests, type check)
4. Update progress.md
5. Git checkpoint
6. Continue to next task
Script mode (maximum performance):
Guide the user to run the bash script directly:
./scripts/autopilot.sh docs/PRD-[name].md --max-iterations=10 --model=claude-opus-4-6
Script mode kills and restarts Claude between iterations for a fully clean context. Recommended for:
- 10+ task features
- Overnight unattended runs
- Running in a DevContainer with --dangerously-skip-permissions
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.
- 2mo ago First seen · 197 lines · 29 tokens per session scan A 88d3ceeaaec5
autopilot is an agent published in the GitHub repository herbert-julio-azion/specialist-agent (21 stars, last pushed 2mo ago), licensed MIT. It adds 29 tokens to every session and 1,611 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-08-30.
Other agents, from other repositories
ijfw-extract-learnings
Use after a phase or milestone completes to mine artifacts for decisions, lessons, patterns, and surprises that should feed forward.
ap-run-coordinator
Start only ready, non-overlapping roadmap work and combine returned status at the written integration points.
ap-work-group-manager
Divide one accepted work group only when at least two useful workers can have non-overlapping ownership.
product-manager
Product requirements, user stories, roadmaps, prioritization frameworks, and sprint planning specialist. Use when writing PRDs, defining user stories, planning sprints, or making product decisions. Trigger phrases: PRD, user story, roadmap, sprint, backlog, prioritization, product requirements, acceptance criteria…
planner
Drafts the execution Plan (plan.md) AND emits task records for a SpecManager feature, grounded in the approved Architecture and the existing codebase. Plans MUST be organised into phases with Fibonacci-scored tasks ≤3.
yd-sdlc-orchestrator
A coordinating agent for building a complete service from one request. It delegates planning, implementation, and security work to separate specialised agents in sequence.