Agent governance
Agent governance means deciding who or what an agent can access and which actions need approval. I explain the permission checks, approvals, tests, and records used to enforce those decisions. Related project pages describe their implementations and limits.
What governance means here
I start by listing the data and tools an agent needs, what it may do, and which actions a person must approve. Then I check how the system enforces those choices and what happens when a run fails. These are the checks I document:
- I check proposed actions against explicit rules before a provider call.
- I use acceptance tests to check the requested change and detect regressions.
- I run a fixed evaluation suite in continuous integration and enforce its release threshold.
- I connect requests, tool calls, and results with a shared trace ID.
- I check the remaining budget before issuing a provider call.
- I pause actions that require approval until a named person approves the exact payload.
- I document reversal procedures and identify changes that cannot be reversed.
The patterns
Acceptance tests for agent work →
Executable checks test the task and detect regressions.
Evaluation tests for releases →
A fixed evaluation suite checks release requirements.
MCP tool permissions →
The server checks which tools a caller may use.
Recording agent runs →
A shared trace ID connects the records of a run.
Spending limits for agent runs →
Budget checks run before provider requests.
Pausing for human approval →
The run waits for approval before the action executes.
Replaying and comparing agent traces →
Step-by-step comparison finds where recorded runs differ.
Recovering runs from checkpoints →
Stored checkpoints support recovery after interruption.
Reversing agent-made changes →
Each change needs a documented reversal procedure.
Handing a system over to a team →
A named owner practices running, debugging, and stopping the system.
Related project examples
ClawKeeper →
ClawKeeper checks proposed Claude agent actions against explicit rules before a model call. Autonomous spend defaults to zero, and the database enforces an append-only audit log.
TechTide Harness Kit →
TechTide Harness Kit uses JSON Schema to declare skill permissions, risk tiers, network access, and approval requirements. Automated checks validate declarations and run prompt-injection fixtures.
CipherClaw →
CipherClaw analyzes recorded agent activity without model calls. It builds causal graphs, scores possible causes, and flags failure patterns.
Swarm 357 →
357 Claude agent roles across six business layers, with portable memory, enforced cost caps, a bash policy gate, and durable checkpoints.
FintheFinder →
FintheFinder pauses research runs for human approval and audits citations deterministically. Its repository includes contracts and test files.
TechTideAI2 →
TechTideAI2 separates agent decisions, application data, and audit output. It uses explicit state transitions, execution timeouts, and adversarial evaluation tests.
I explain the engineering here. TechTide AI publishes current service scope, prices, and terms.
Systems Decision Audit at TechTide AI→© 2026 Alex Cinovoj · TechTide AI · Columbus, OH