skills.rest
entropy-cloud
entropy-cloudCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

mission-driver

Create roadmaps and mission configs, then drive an autonomous AI development loop engine.

693|100|Updated Aug 18, 2022
Search Tags:#audit#plan execution#autonomous development#ROADMAP#dev loop#Opencode#mission-driver
One-click install
npx skills add https://github.com/entropy-cloud/nop-entropy --skill mission-driver
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mission-driver
Source: https://github.com/entropy-cloud/nop-entropy/tree/main/.opencode/skills/mission-driver
Command: npx skills add https://github.com/entropy-cloud/nop-entropy --skill mission-driver

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Coordinating long-running, multi-step development work (refactors, tech-debt cleanup, feature implementation) with AI agents is error-prone: plans drift, verification steps get skipped, and progress state gets lost. This Skill turns a development goal into a structured roadmap plus a mission config, then drives the mission-driver engine through a CHECK → REVIEW_PLANS → EXEC_PLANS → DRAFT_PLANS → DEEP_AUDIT loop until the roadmap is done.

Core Features & Use Cases

  • Roadmap authoring: Generates a structured roadmap (docs/backlog/{slug}-roadmap.md) with work items sized to single execution plans, dependency graphs, and a single dynamic state block.
  • Mission config generation: Creates and validates missions/<name>.json (via the draft command or by hand) with stack-agnostic test/build/lint commands for Node, Python, Java, Go, or Rust projects.
  • Autonomous loop execution and monitoring: Launches, resumes, and monitors missions via tools/mission-driver.sh, with a browser dashboard, run-state inspection, graceful interrupt/resume, orphan-process cleanup, and postmortem analysis.
  • Use Case: You need to refactor an auth module across 15 files with tests kept green. Describe the goal, let the skill draft the roadmap and mission config, validate with mission-check.mjs, then run the mission and monitor it at localhost:9300 while the engine drafts plans, executes them, and audits the results.

Quick Start

Ask the AI to create a roadmap and mission for your development goal, then run ./tools/mission-driver.sh run <mission-name> after validating the generated mission config.

Frequently Asked Questions about mission-driver

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

FAQPage Schema
How do I create a roadmap and mission config for an AI dev loop?▼

Run ./tools/mission-driver.sh draft with a goal description and a target requirements file. The draft command generates a brief, a roadmap in docs/backlog/, and a missions/<name>.json config, which you should review and validate with mission-check.mjs before running.

How do I run and monitor a mission-driver mission?▼

Run ./tools/mission-driver.sh run <mission-name> to start the loop with a monitor dashboard on port 9300. You can inspect _tmp/<runDir>/run-state.json for authoritative state, tail the per-step logs, and resume after interruption by re-running the same command.

Does mission-driver work with Java Maven or Python projects?▼

Yes, the engine is stack-agnostic: verification commands are plain shell strings in the mission config. The schema reference includes examples for Maven (mvn -pl <module> -am test), pytest, Go, Rust, and pnpm/npm projects.

When should I not use mission-driver for a development task?▼

Avoid it for quick fixes under 30 minutes, single-file renames, one-off scripts, or exploratory prototyping without measurable acceptance criteria. It is designed for multi-step work over one hour where docs, code, and tests must stay in sync.

Why is my mission stuck and how do I fix it?▼

Check run-state.json and the step logs: a frozen log with 0% CPU indicates a stalled sub-agent stream, while stream errors usually auto-retry. Use kill -TERM on the step process group to advance the flow, or consult TROUBLESHOOTING.md for ping-pong loops and spawn failures.

Related Skills

Explore More Software Engineering Skills→
ThinkInAIXYZThinkInAIXYZ
6.3k

deepchat-sdd

Creates RFC-style specs and implementation plans for substantial DeepChat changes.

Official
Intermediate
nesnilnehcnesnilnehc
7

plan-next

Analyze governance documents to generate an executable next-action plan.

Community
Advanced
ssrxsssrxs

laravel-verification

Verifies Laravel projects' environment, code quality, security, and deployment readiness.

Community
Advanced
Chorus-AIDLCChorus-AIDLC
1.2k

quick-dev

Creates and executes standalone tasks with acceptance criteria, self-checks, and permission-aware verification.

Official
Intermediate