Skip to content

Latest commit

 

History

History

README.md

Mutiny Documentation

All project documentation lives here. The root README is the public front door; this hub is the source of truth for product intent, architecture, and contributor onboarding.

Mutiny is a behavioral fuzz-testing engine for AI agents. Install it into your agent project via an adapter:

mutiny init → connect adapter → mutiny run → search for policy breaks → prove on traces → minimize → save permanent regression tests.

Shipped today: Adapter #1 — OpenAI Agents SDK. Install with pip install mutiny-ai (CLI stays mutiny; see root README / PUBLISHING.md). Sample/demo agents are reference harnesses, not the primary product.

Safety: Authorized testing only. Targets are local / in-process / localhost with sandboxed mock tools for demos.


Start here

If you want… Read
Install + try in 5 minutes Root README
CLI flags (init / run / test) CLI.md
Packaging / PyPI PUBLISHING.md
Clean-machine bootstrap COLD_START.md
Contribute / first PR CONTRIBUTING.md · GOOD_FIRST_ISSUES.md
Architecture boundaries ARCHITECTURE.md
How the system runs SYSTEM_DESIGN.md
What’s next ROADMAP.md
Why we chose X DECISION_LOG.md (esp. ADR-017, ADR-018)
Support / Windows / security SUPPORT.md · SECURITY.md

Document map

Document Owns Does not own
PRD.md Product intent, users, requirements, non-goals Algorithms, package layout, day-by-day tasks
ARCHITECTURE.md Principles, layering, package ownership, constraints Product roadmap, competitor detail
SYSTEM_DESIGN.md Lifecycles, data contracts, flows, diagrams Milestone scheduling, demo narration
IMPLEMENTATION_PLAN.md Milestones, DoD, sequencing Product vision, long-term roadmap
DEMO_SCRIPT.md Live demo scripts, backup, Q&A System internals
COLD_START.md Clean-machine bootstrap + smoke gate Product narrative
GOOD_FIRST_ISSUES.md Catalog of small contributor tasks Execution ownership of Core redesigns
CLI.md mutiny commands and flags from --help Hosted UI walkthrough
../CHANGELOG.md Unreleased + release notes Live demo narration
DEVPOST.md Short public pitch copy Live demo choreography
COMPETITOR_ANALYSIS.md Competitive landscape (claims freeze) Mutiny feature specs
DECISION_LOG.md Architecture Decision Records Ongoing task lists
ROADMAP.md Phased future work Near-term execution detail
DESIGN.md Pointer to Hosted visual canon Product requirements

Community files at repo root: CONTRIBUTING, CODE_OF_CONDUCT, SECURITY, SUPPORT, LICENSE.


Conflict resolution

  1. Product intent → PRD
  2. Hard engineering constraints / boundaries → ARCHITECTURE
  3. Behavioral contracts (how the system works) → SYSTEM_DESIGN
  4. What to build next / DoD → IMPLEMENTATION_PLAN
  5. Why we chose X → DECISION_LOG

If docs disagree, fix the docs in that order—do not silently pick a convenient interpretation in code.

Install claims: Prefer the root README. Primary path is pip install mutiny-ai (CLI stays mutiny). Never recommend bare pip install mutiny / mutiny-sdk (those names are taken by other projects). Git/source install is optional — see PUBLISHING.md.


Reading order (new engineer)

  1. Root README (What / Why / Install / Try)
  2. PRD.md (thesis + non-goals)
  3. ARCHITECTURE.md (boundaries)
  4. SYSTEM_DESIGN.md (how it runs)
  5. DECISION_LOG.md (ADR-017 customer-project primary; ADR-018 adapter-first)
  6. ROADMAP.md / GOOD_FIRST_ISSUES.md as needed

Product shape (canonical)

Customer agent project
        ↑
OpenAI Agents SDK Adapter   ← Adapter #1 (shipped)
        ↑
   Adapter Layer            ← future: LangGraph, CrewAI, PydanticAI, AutoGen, HTTP, …
        ↑
   Mutiny Core              ← behavioral fuzz-testing engine
        ↑
   CLI (mutiny init / run / test)  ← primary
   Hosted API + UI                 ← secondary
   Sample / demo agent             ← reference harness

Screenshots / demo assets: docs/assets/ — live Hosted PNGs + optional GIF; SVG storyboard remains as a static diagram. See assets/README.md. Release notes: CHANGELOG.md.