pilot

Generates a deterministic, seedable first-person drone flight as a single self-contained HTML file.

465|41|Updated Aug 4, 2026
One-click install
npx skills add https://github.com/autonomous-ai/openharness --skill pilot-autonomous-ai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pilot
Source: https://github.com/autonomous-ai/openharness/tree/main/store/agents/drone-pilot/skills/pilot
Command: npx skills add https://github.com/autonomous-ai/openharness --skill pilot-autonomous-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Turning a plain-English description into a playable first-person drone flight usually requires wiring together physics, procedural course generation, and rendering by hand. This Skill produces a complete, reproducible FPV flight experience in one HTML file that can be previewed and re-seeded live in a web viewer. ## Core Features & Use Cases - Seeded procedural generation: A seeded PRNG with named sub-streams generates courses (gates, rings, obstacles, spawn) so every seed produces an identical, reproducible flight. - Fixed-timestep physics: A stable 1/60 simulation loop drives position, velocity, yaw/pitch, and throttle, with autopilot/manual handoff, brake/boost, HUD, and minimap. - Live seed browsing: A ?seed= query parameter plus a re-seed UI lets users browse seeds in the viewer, with a watch mode that flies the course autonomously. - Use Case: Ask for a slalom course through twelve gates; the Skill outputs flight/index.html, verifies seeds 0-99 via node tools/check.mjs --seeds 100, and records a verdict in .harness/verdict.json. ## Quick Start Ask the agent to build a seeded FPV drone flight, for example: create a canyon weave drone run with gates and a minimap that I can re-seed in the browser.

Frequently Asked Questions about pilot

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

FAQPage Schema
How do I generate a drone flight simulation from a text description?

Describe the flight you want, such as a slalom through gates or a canyon weave, and the Skill produces a single self-contained flight/index.html with seeded course generation, fixed-timestep physics, and a first-person view you can open in a browser.

How do I make a procedural course reproducible with a seed?

The Skill derives a seeded PRNG from the hash of the seed string and uses named sub-streams per concern (course layout, obstacles, decor). The seed is read from a ?seed= query parameter, and a re-seed button lets you browse seeds live.

Can I verify that the same seed produces the same flight?

Yes. Run node tools/check.mjs --seeds 100 in the workspace to check domain invariants and repeat each seed against the pure model. Same-machine reproducibility is verified, though frame-timing physics cannot be proven bit-identical.

What physics does the drone flight simulation use?

It uses a small fixed-timestep loop at 1/60 simulation ticks tracking position, velocity, yaw/pitch, and throttle. The template includes autopilot/manual handoff, brake/boost controls, a HUD, and a minimap.

What are the limitations of seed-based flight reproducibility?

Reproducibility is guaranteed only on the same machine. Because rendering depends on frame timing, the Skill cannot prove bit-identical results across environments, and it states this explicitly in the verdict findings.