Overview
VibeGov is a repo-local governance framework for AI-assisted software delivery.
It helps teams ship fast without losing control of intent, quality, evidence, and traceability.
The model is simple:
- use Development to change reality with proof,
- use Exploration to discover reality and hydrate backlog,
- keep release verification inside Development as part of the delivery gate,
- keep
.governance/as the canonical source of delivery guidance.
This site is published incrementally so guidance can evolve in the open with real team feedback.
Who VibeGov is for
VibeGov is for people using AI coding agents on real repositories where delivery intent, evidence, and traceability matter.
It is a good fit for:
- teams using AI coding agents on production-bound or maintained software
- solo developers who want agent speed without losing traceability
- technical founders and consultants managing multiple AI-assisted delivery threads
- maintainers who want issue, spec, and evidence discipline without heavyweight process tooling
- teams building agent harnesses that need repo-local operating rules
It is probably overkill for:
- throwaway scripts with no maintenance horizon
- teams unwilling to require evidence for completion claims
- projects that do not need traceability, backlog quality, or long-term change safety
- users looking for a hosted SaaS or magical autonomous runtime
If terms like evidence, traceability, bootstrap, or blocker are new to you, use the Glossary as a companion reference while reading the deeper docs.
Content status
VibeGov separates canonical contracts from evolving guidance.
Use the Content Status Model when you need to know whether a surface is:
- a canonical contract,
- a published governance rule,
- an operational guide,
- a role pack,
- a runtime-specific profile, or
- blog/commentary.
Canonical bootstrap surfaces such as agent.txt, bootstrap.json, and Bootstrap stay authoritative when other pages summarize or explain them.
Why VibeGov exists
AI-assisted delivery makes it easy to generate code quickly.
The failure mode is not just bad code. It is delivery drift:
- why something is being built gets blurred
- constraints get skipped
- "done" collapses to output instead of evidence
- partial review gets reported like full review
- blockers stall work or disappear into chat
- future change gets harder because intent and proof were not preserved
VibeGov exists to make delivery judgement more explicit and portable.
It captures SDLC wisdom as reusable guidance so humans and agents inherit stronger defaults without heavy process tooling.
Why .governance/ exists
.governance/ isolates delivery guidance from product code so it is easy to copy, version, and evolve.
Folder roles:
.governance/rules/: cross-project governance rules (gov-*.mdc).governance/project/: project-level intent and constraints.governance/specs/: feature/change specs for scoped work
Canonical-source model:
.governance/is the single source of truth for governance.- Different agents/providers can use the same rules and link them via their native mechanism (for example
AGENTS.mdor an existing provider-native rules directory detected during onboarding).
Install
- First decision path: Start Here
- Short install path: Quick Start
- Canonical bootstrap contract: Bootstrap
What changes when VibeGov is present
- agents ask better scoping questions
- Development and Exploration stop being blurred together, and release verification becomes an explicit part of Development rather than an implicit afterthought
- evidence quality improves because proof is mode-aware
- exploratory review becomes structured backlog hydration
- blockers become visible routing events instead of hidden stalls
- releases leave clearer artifacts
- long-term evolution gets safer
Best next reads
- Start Here
- Quick Start
- Glossary
- VibeGov vs Agent Rules, Prompts, and Frameworks
- The VibeGov SDLC
- VibeGov Quick Decisions
- Mode Selection and Evidence Closing
- Simplicity-First Guidance
- Output Quality and Anti-Slop Guidance
- Execution Modes
- Evaluation Pattern
- Exploratory Review Mode
- Checkpoint Reporting
- Blocker Escalation
- Workflow Quality Rubric
Give feedback
- Open suggestions: Create an issue
- View current discussions: Issue log
Published governance pages
- GOV-01: Instructions
- GOV-02: Workflow
- GOV-03: Communication
- GOV-04: Quality
- GOV-05: Testing
- GOV-06: Issues
- GOV-07: Tasks
- GOV-08: Exploratory Review
- GOV-09: Agent Continuity Bootstrap
- GOV-10: Agent State Closure and Git Hygiene
- GOV-11: Agent Legibility and In-Repo Truth
- GOV-12: Drift Control and Garbage Collection
- GOV-13: Review Loops and Completion Discipline