The Problem We’re Solving
Most agent governance systems rely on:- Probabilistic classifiers
- Post-hoc analysis
- Human review after damage is done
- Explain probabilistic decisions to auditors
- Replay historical decisions reliably
- Enforce guarantees with systems that “usually” work
What Rules Can See
Rules are evaluated at well-defined points (primarily before tool calls). They can inspect:- Agent identity
- Run metadata
- Tool identity and metadata
- Tool call parameters
- Organisation configuration
- Historical signals (explicitly versioned)
Rule Evaluation Lifecycle
At runtime:- An agent attempts a tool call
- Relevant rules are evaluated in a defined order
- Actions are produced
- Actions are enforced
Ordering and Conflicts
Rules are evaluated in a predictable order. When multiple rules apply:- Actions are composed deterministically
- Conflicts are resolved explicitly
- A single block action wins over multiple allows
- A require approval can short-circuit execution
Versioning and Replay
Every rule has a version. When a decision is made, we record:- Rule IDs
- Rule versions
- Inputs
- Outputs
- Exact replay of historical decisions
- Diffing behaviour across rule versions
- Confident rule changes