Get running
decapod init creates .decapod/, a local folder your agent uses to remember intent, rules, context, specs, and proof.
Your conversational workflow does not change. You keep talking to your agent normally; behind the scenes, the agent calls Decapod to ensure work happens in isolated environments and adheres to your project's constitution.
How it works
AI coding agents often lose the plot: they forget intent, pull too much context, skip dependencies, and touch protected files. Decapod provides a repo-native checkpoint system to absorb these deficiencies.
The Loop
User
│
▼
Agent ───────┐
│ │
│ ┌────▼────┐
│ │ Decapod │
│ │ (check) │
│ └────┬────┘
│ │
├─────────┤
│ │
Model Agent
│ │
└────┬────┘
▼
User
Decapod is not the agent. It is the governance kernel called before:
- Acting — clarify intent and generate specs
- Inference — resolve surgical context capsules
- Touching Code — enforce boundaries and protected paths
- Completing — produce verification and proof
Capabilities
- Clarifies intent — Converts vague requests into explicit, versioned specifications.
- Bounds context — Resolves only the minimal relevant code/docs for the task.
- Enforces boundaries — Safeguards protected branches and sensitive modules.
- Governs adaptation — Manages feedback-driven instruction changes through explicit review.
- Requires proof — Gates completion on deterministic verification artifacts.
The substrate
Decapod preserves what agent workbenches lose: reusable, repo-native knowledge that survives the session.
.decapod/
generated/
specs/ # Human-visible intent and architecture specs
context/ # Deterministic context capsules
artifacts/ # Verification output and proof provenance
data/ # Durable repo-native state (DBs, events, todos)
config.toml # Project shape and agent-facing configuration
OVERRIDE.md # Local rules that override embedded defaults
Every run leaves operational evidence. The generated files are the human-visible proof surface—inspect them locally, review them in PRs, and use them to re-establish state across different agents (Claude, Codex, Gemini, Cursor).
The constitution
Decapod ships with an embedded engineering constitution: over 100 declarative documents covering architecture, security, performance, and testing.
Everything an engineering org usually keeps in tribal memory or review culture becomes executable guidance. Your agent does not guess; it reads the constitution, cites claim IDs, follows gates, and produces proof.
Guarantees
- Daemonless — Runs on demand like
gitorgrep. - Repo-native — All state lives in your repository.
- Provider-agnostic — Works across all agent workbenches.
- Proof-gated — Completion requires passed verification gates.
- Boundary-aware — Enforces protected paths and branch isolation.
Contributing
&&