bylaws
One policy, enforced in every AI coding agent your team runs.
About
Every AI coding agent has its own way of saying "never read secrets" or "ask before pushing". A team running more than one has no single source of truth, no record of what the agents tried to do, and no way to know when a rule was quietly turned off.
bylaws fixes that. The policy lives in a BYLAWS.md written in English, compiles to a rules file that is committed and reviewed like code, and is installed as a hook into Claude Code, Cursor, Codex and Hermes. Every tool call gets an allow, ask or deny decision, and every decision is logged.
Capabilities
Policy in English
Write rules as prose; compile them to a deterministic rules.yaml you review in a pull request.
Every agent
One install command per agent: Claude Code, Cursor, Codex and Hermes.
Pre-execution hooks
Each tool call is checked before it runs and answered with allow, ask or deny.
Audit log
Every decision appended to a local JSONL log, so you can see what agents attempted.
Drift detection
Catches when the compiled rules no longer match the English, locally and in CI.
Origin
Built by Chito Technologies
This product is part of the Chito Technologies portfolio.
More from the portfolio
06
CogniDesk
A TypeScript runtime SDK for support agents — conversation journeys, typed tools, knowledge retrieval and UI, without vendor lock-in.
07
Nepal MCP
One MCP endpoint that gives AI agents live NEPSE prices, NRB forex and economic indicators, and Nepali news with sentiment.