Autopilot
- 43 installs
- 40 repo stars
- Updated August 4, 2026
- akillness/oh-my-skills
Autopilot is a Claude Code skill that runs an autonomous build workflow from a request through planning, implementation, QA, and verification.
About
Autopilot is an exact-name entry point for a Codex/OMX-style autonomous build workflow that takes a product or code request through planning, implementation, QA, and final verification. A developer invokes it with $autopilot when they want the agent to build the whole thing end to end rather than a focused one-file edit. It implements in small safe slices and stops on repeated failures or missing decisions.
- Front-door alias for $autopilot: end-to-end autonomous build from request to verification
- Runs planning, implementation, QA, and a final clean verification pass in small safe slices
- Defined stop conditions: same failure three times, missing credentials, or user cancels
Autopilot by the numbers
- 43 all-time installs (skills.sh)
- Ranked #7,972 of 16,546 AI & Agent Building skills by installs in the Skillselion catalog
- Data as of Aug 5, 2026 (Skillselion catalog sync)
autopilot capabilities & compatibility
- Capabilities
- autonomous build · planning · qa · verification
- Use cases
- orchestration · testing · planning
- Runs
- Runs locally
- Pricing
- Free
What autopilot says it does
Run a Codex/OMX-compatible autonomous build workflow from a product or code request through planning, implementation, QA, and final verification.
The same verification failure repeats three times with no new evidence.
npx skills add https://github.com/akillness/oh-my-skills --skill autopilotAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 43 |
|---|---|
| repo stars | ★ 40 |
| Last updated | August 4, 2026 |
| Repository | akillness/oh-my-skills ↗ |
What it does
Run an autonomous build workflow from a request through planning, implementation, QA, and final verification when the user invokes $autopilot.
Who is it for?
Tasks that need end-to-end autonomous execution: plan, implement, QA, and verify the whole thing.
Skip if: Small focused one-file edits; route those to direct implementation.
When should I use this skill?
The user explicitly invokes $autopilot, autopilot, auto pilot, or full-auto.
What you get
The agent produces a working build in small verified slices with a final clean verification pass.
- execution packet (goal, constraints, files, verification commands)
- implemented changes
- verification report
By the numbers
- 4-stage flow (planning, implementation, QA, verification)
- 3-repeat verification-failure stop condition
Files
Autopilot
Use this skill as the exact-name front door for $autopilot in Codex-style skill loaders. It exists because documenting $autopilot only inside omx is too weak for clients that activate skills by folder name.
When to use this skill
- The user explicitly invokes
$autopilot,autopilot,auto pilot, or full-auto. - The task needs planning, implementation, QA, and final verification.
- The user wants the agent to build the whole thing end to end.
Instructions
1. Confirm the request needs end-to-end autonomous execution, not a small focused edit. 2. Inspect the repo enough to identify stack, test commands, and risky boundaries. 3. Freeze a compact execution packet:
- goal
- constraints
- files or areas likely to change
- verification commands
4. Implement in the smallest safe slices. 5. Run QA after each meaningful slice, then a final clean verification pass. 6. Report what changed and the exact checks that passed or failed.
Runtime Notes
- If the local OMX runtime is available and the user clearly wants that runtime,
use the existing OMX workflow state under .omx/.
- If OMX is not available, still apply this skill as an in-session Codex workflow:
plan, edit, verify, and keep going until the acceptance checks are honestly resolved or blocked.
- Do not pretend that shell
omx teamor Claude/autopilotis the same
implementation. Treat them as runtime-specific equivalents of the same intent.
Stop Conditions
- The same verification failure repeats three times with no new evidence.
- Required credentials, services, or product decisions are unavailable.
- The user cancels or narrows the request to a smaller task.
Examples
$autopilot build a task API with auth and testsautopilot migrate this CLI to use SQLite persistence
Best practices
- Keep the execution packet small enough to verify.
- Prefer repo-native test/build commands over invented commands.
- Preserve failure evidence before changing code.
References
omxfor broader Codex runtime setup and command parity.ooofor immutable spec-first execution.
{
"skill": "autopilot",
"version": "1.0.0",
"cases": [
{
"prompt": "$autopilot build a task API with auth and tests",
"expected_output": "The autopilot skill activates as the exact-name front door and runs an end-to-end plan, implementation, QA, and verification workflow.",
"checks": [
"Treats `$autopilot` as a direct trigger",
"Does not route first to the broad `omx` router",
"Includes final verification before done"
]
},
{
"prompt": "Fix this one typo in README",
"expected_output": "The autopilot skill should not activate; the task should be handled as a small direct edit.",
"checks": [
"Rejects small focused edits",
"Does not inflate the task into a multi-phase autonomous build"
]
}
]
}
N:autopilot
D:Exact-name Codex/OMX-compatible `$autopilot` front door for autonomous idea-to-verified-code work. Use when the user invokes `$autopilot`, autopilot, auto pilot, full-auto, or asks for an end-to-end build; route small focused edits to direct implementation and stricter spec-first loops to `ooo` or `ralph`.
T:Read|Write|Bash|Grep|Glob
G:autopilot|$autopilot|omx|codex|autonomous-build|full-auto|implementation|qa|verification
F:Codex CLI|Claude Code|Gemini CLI|OpenCode
S:
1:Inspect stack, likely files, test commands, and constraints before editing
2:Freeze a compact execution packet with goal, constraints, work areas, and verification commands
3:Implement in safe slices and verify after meaningful changes
4:Finish only after a fresh final verification pass or a concrete blocker
R:
ooo:Use for immutable spec-first execution or drift-aware completion loops
ralph:Use for high-assurance fix/verify loops over a known plan
omx:Use for broader Codex runtime setup, team mode, and command parity guidance
Related skills
Setup Matt Pocock SkillsScaffold the per-repo configuration that Matt Pocock’s engineering agent skills rely on so they understand the issue tracker, triage labels, and domain documentation la495k196k
Lark AppsConnect Claude, Cursor or custom agents directly to Lark (Feishu) for messaging, document automation, approval workflows and enterprise data access.402k
CavemanSlash token usage by roughly 75% while keeping every technical detail intact when working with Claude Code, Cursor or similar agents.391k94.7k
Lark Skill MakerQuickly turn any Lark/Feishu OpenAPI call or multi-step workflow into a reusable agent skill with its own SKILL.md.386k16k
Running Claude Code Via Litellm CopilotRun Claude Code at a fraction of the cost by routing requests through LiteLLM to the GitHub Copilot Chat API.270k72
Codex PetGenerate a complete Codex Pet spritesheet and metadata from one reference image without needing an OpenAI key or Codex Pro.246k10FAQ
When should I use autopilot versus a direct edit?
Use it for end-to-end autonomous builds; route small one-file fixes to direct implementation.
When does it stop?
When the same verification failure repeats three times, required credentials or decisions are missing, or the user cancels.