# AI agent permissions and approval rules: who or what an agent can access, and which actions need approval

> Agent governance means deciding who or what an agent can access and which actions need approval. I document permission checks, approvals, tests, and recorded actions, with related project examples.

Source page: https://alexcinovoj.dev/agent-governance
Markdown: https://alexcinovoj.dev/llms/agent-governance.md

## Controls, and where each one is documented

- [Acceptance tests for agent work](https://alexcinovoj.dev/llms/proof/ai/acceptance-tests.md): Executable checks test the task and detect regressions. Answers: How do I check whether an agent completed its task?
- [Evaluation tests for releases](https://alexcinovoj.dev/llms/proof/ai/evals.md): A fixed evaluation suite checks release requirements. Answers: How do evaluation tests check a release?
- [MCP tool permissions](https://alexcinovoj.dev/llms/proof/ai/mcp-permissions.md): The server checks which tools a caller may use. Answers: How do I restrict an MCP caller’s tool access?
- [Recording agent runs](https://alexcinovoj.dev/llms/proof/ai/observability.md): A shared trace ID connects the records of a run. Answers: What should I record to investigate an agent failure?
- [Spending limits for agent runs](https://alexcinovoj.dev/llms/proof/ai/cost-controls.md): Budget checks run before provider requests. Answers: How do I check spending limits before an agent call?
- [Pausing for human approval](https://alexcinovoj.dev/llms/proof/ai/approval-gates.md): The run waits for approval before the action executes. Answers: How do I pause an agent for human approval?
- [Replaying and comparing agent traces](https://alexcinovoj.dev/llms/proof/ai/trace-replay.md): Step-by-step comparison finds where recorded runs differ. Answers: How do I compare recorded agent runs?
- [Recovering runs from checkpoints](https://alexcinovoj.dev/llms/proof/ai/recovery.md): Stored checkpoints support recovery after interruption. Answers: How do checkpoints help a run recover?
- [Reversing agent-made changes](https://alexcinovoj.dev/llms/proof/ai/rollback.md): Each change needs a documented reversal procedure. Answers: How do I reverse changes an agent made?
- [Handing a system over to a team](https://alexcinovoj.dev/llms/proof/ai/ownership-handoff.md): A named owner practices running, debugging, and stopping the system. Answers: How do I hand a system over to a team?

## Related projects

- [ClawKeeper](https://alexcinovoj.dev/llms/projects/clawkeeper.md): Open-source project. 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](https://alexcinovoj.dev/llms/projects/techtide-harness-kit.md): Open-source project. 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](https://alexcinovoj.dev/llms/projects/cipherclaw.md): Open-source project. CipherClaw analyzes recorded agent activity without model calls. It builds causal graphs, scores possible causes, and flags failure patterns.
- [Swarm 357](https://alexcinovoj.dev/llms/projects/swarm-357.md): Open-source project. 357 Claude agent roles across six business layers, with portable memory, enforced cost caps, a bash policy gate, and durable checkpoints.
- [FintheFinder](https://alexcinovoj.dev/llms/projects/finthefinder.md): Open-source project. FintheFinder pauses research runs for human approval and audits citations deterministically. Its repository includes contracts and test files.
- [TechTideAI2](https://alexcinovoj.dev/llms/projects/techtideai2.md): Open-source project. TechTideAI2 separates agent decisions, application data, and audit output. It uses explicit state transitions, execution timeouts, and adversarial evaluation tests.
- [BRI](https://alexcinovoj.dev/llms/projects/bri.md): Open-source project. BRI exposes video analysis tools through a FastAPI MCP server, with SQLite job storage and a Streamlit interface.
- [Miro Workflows](https://alexcinovoj.dev/llms/projects/miro-workflows.md): Open-source project. Workflow operations dashboard that moves board-shaped work into explicit SQLite schemas with MCP tooling and optional live sync back to Miro.
- [AgentRewind](https://alexcinovoj.dev/llms/projects/agentrewind.md): Private build. Trace forensics build: zero-trust import parser and a diff engine for replaying agent runs. Private build with no public artifacts.
- [Molten](https://alexcinovoj.dev/llms/projects/molten.md): Private build. Governed agent runtime: kill switch, circuit breakers, four-tier sandbox, dead-man's switch. Private build with no public artifacts.
