autopilot

Orchestrates the idea-to-ship pipeline autonomously by composing ideate, deep-dive, prompt-pack, and build-loop skills.

83|8|Updated Jun 9, 2026
One-click install
npx skills add https://github.com/nelsonwerd/idea-to-ship-skills --skill autopilot-nelsonwerd
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: autopilot
Source: https://github.com/nelsonwerd/idea-to-ship-skills/tree/main/skills/autopilot
Command: npx skills add https://github.com/nelsonwerd/idea-to-ship-skills --skill autopilot-nelsonwerd

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Running a full idea-to-product workflow by hand means manually chaining ideation, validation, build planning, and iterative building across many sessions, with gates easily skipped or faked along the way. This Skill flies that entire pipeline autonomously in character as a grounded founder-persona, producing a near-finish-line first-draft product plus an honest ledger of what only a human or the market can finish. ## Core Features & Use Cases - End-to-end orchestration: Invokes the ideate, deep-dive, prompt-pack, and build-loop skills in order, carrying each phase's file output into the next without reimplementing any sub-skill. - Autonomy contract with honest gates: Answers founder-decidable gates in-character without stopping, but emits — never fakes — any gate requiring human or real-world signal, enforced by execute-discipline and a kill-ledger for parked concepts. - Durable file-based memory and handoff: Every phase writes files (CONCEPT_BRIEF, prompt pack, build ledgers) so long runs survive context limits and resume cleanly in fresh chats. - Use Case: You have a grounded niche backed by real lived pain or real data. Kick off an autonomous run and receive a validated concept brief, a sequenced build pack, a machine-validated first-draft product, and an explicit handoff listing the last-mile tail, design taste spot-check, and market validation left for humans. ## Quick Start Run autopilot on my grounded niche of freelance designers struggling to track client revisions and build a first-draft product end to end.

Frequently Asked Questions about autopilot

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run the idea-to-ship pipeline autonomously?▼

Invoke autopilot with a grounded niche and it emits a kickoff locking the persona, success metric, and kill criterion, then runs ideate, deep-dive, prompt-pack, and build-loop in order. Each phase writes a file the next phase reads, so the run is legible and resumable.

What does autopilot produce at the end of a run?▼

It produces a grounded CONCEPT_BRIEF, a validated and sequenced build pack, a near-finish-line-aimed first-draft product, and an honest ledger of what is machine-checked versus what only a human or the market can verify. It explicitly does not claim a finished or market-validated product.

When should I not use autopilot?▼

Do not use it to drive each pipeline phase by hand, when the concept has no real grounding, or when you need market validation. For manual phase control use ideate, deep-dive, prompt-pack, or build-loop directly; autopilot will not invent demand to fill a grounding gap.

Can autopilot validate that people actually want the product?▼

No, market validation is explicitly out of scope and is the human handoff. A synthetic persona's reaction to a proposed solution never counts as validation under its grounding firewall; real data may discover and seed the problem but cannot manufacture demand.

What happens if an autonomous run hits a context limit?▼

Files are the durable memory, not chat history: the brief, pack, build-loop ledgers, and proof notes persist across sessions. If context runs low, autopilot emits a prompt-pack Mode C handoff so a fresh chat resumes with no amnesia.

Why does autopilot refuse to auto-pass certain gates?▼

Gates needing human or real-world signal, such as real use, taste sign-off, or market tests, are emitted honestly rather than faked in-character. This execute-discipline rule exists because the sharpest observed failure was a run that narrated a gate it had actually abandoned.