canonical: https://jentic.com/apis/getmarrow.ai/getmarrow

# Getmarrow Marrow Agent Governance API

The Marrow Agent Governance API evaluates an AI agent's proposed actions before they run and records proof of what happened after. It reads the bound agent's identity, health, and tenant-scoped operating context, evaluates a proposed action and returns governance guidance, then closes the action with its outcome and proof evidence. It issues, verifies, and heartbeats signed permits for individual governed actions, reports action-permit coverage, and surfaces reliability and value metrics for the agent. Public health and discovery endpoints let a caller check status and probe whether governance is relevant before binding an agent.

## For AI agents

Govern an AI agent's actions: read its context, evaluate a proposed action before execution, issue and verify signed permits, close actions with proof evidence, and read reliability and value metrics.

## Scope

Does not handle user login, data storage, or payment authorization. Use for governing and permitting an agent's actions before and after execution only.

## Capabilities

- Verify a bound agent's identity and read its health, proof state, and repair guidance
- Read tenant-scoped operating context for the agent before it acts
- Evaluate a proposed action before execution and return governance guidance
- Issue, verify, close, or heartbeat a signed permit for a single governed action
- Close a governed action with its outcome and proof evidence
- Read action-permit coverage and the agent's reliability and value metrics

## Use cases

### Agent action pre-check through Jentic

An AI agent about to take a consequential step can call the Marrow Agent Governance API through Jentic to evaluate the proposed action and get governance guidance before it runs, then commit the outcome with proof once it completes. Jentic returns the evaluate operation from a natural-language query, so the agent checks governance without reading the Marrow reference first.

Example prompt: Evaluate the proposed action against Marrow, proceed only if the guidance allows it, then close the action with its outcome and proof evidence.

### Signed permit lifecycle

Issue a signed permit for a single governed action, verify it at execution time, heartbeat it while the action is in flight, and close it when done. Platform teams use this to keep a verifiable record that each sensitive action was authorized and completed within its permit.

Example prompt: Issue a permit for the governed action, verify it before execution, and close it with the final outcome once the action finishes.

### Agent reliability reporting

Read tenant-scoped reliability, proof, outcome, and value metrics to show how a governed agent is performing over time. Operators use the report to spot drift in permit coverage and to demonstrate the agent stayed within its bounds.

Example prompt: Read the agent value report and the enforcement coverage, then summarize permit coverage and any configuration drift for the tenant.

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| GET | `/v1/agent/status` | Verify agent identity and read repair guidance |
| GET | `/v1/agent/context` | Read tenant-scoped operating context |
| POST | `/v1/agent/runtime` | Evaluate a proposed action before execution |
| POST | `/v1/agent/commit` | Close a governed action with proof evidence |
| GET | `/v1/agent/enforcement` | Read action-permit coverage |
| POST | `/v1/agent/enforcement` | Issue, verify, close, or heartbeat a permit |
| GET | `/v1/agent/report` | Read reliability and value metrics |
| POST | `/v1/public/discovery/diagnose` | Evaluate signals for governance relevance |

## Key resources

- **Agent** — Read agent identity, context, and reliability metrics for a bound agent
- **Runtime** — Evaluate a proposed action before execution and commit its outcome with proof
- **Enforcement** — Issue, verify, close, and heartbeat signed permits and read coverage
- **Discovery** — Public health, status, and governance-relevance discovery endpoints

## AI readiness

This API is usable in Jentic One now. Its AI-readiness score against Jentic's framework shows where it stands today and where improvements would make it even easier for agents to use.

- **Score:** 62 / 100
- **Maturity:** AI-Aware
- **Dimensions:**
  - Foundational Compliance: 83 / 100
  - Developer Experience & Jentic Compatibility: 60 / 100
  - AI-Readiness & Agent Experience: 42 / 100
  - Agent Usability: 94 / 100
  - Security: 60 / 100
  - AI Discoverability: 70 / 100
- **View full report:** https://jentic.com/apis/getmarrow.ai/getmarrow/scorecard
- **How the score is calculated:** https://docs.jentic.com/reference/api-readiness-framework/overview/
- **More about the dimensions:** https://docs.jentic.com/reference/api-readiness-framework/specification/#dimensional-model-overview

### Score it yourself

Every API in the directory is allowlisted, so you can re-score it with no key required.

- **Score your own API:** https://jentic.com/scorecard.md
- **Scoring CLI agent skill:** https://github.com/jentic/jentic-api-scorecard/blob/main/skills/jentic-api-scorecard/SKILL.md

```sh
npx @jentic/api-scorecard-cli score <openapi-url>
```

## Why Jentic

- **Setup:** Wiring the Marrow Agent Governance API by hand means handling its bearer API key, binding the agent, and coding the evaluate, permit, and commit calls in the right order yourself. Through Jentic you install once, import the API from the API Directory, store the key once, and your agent calls it.
- **Permission scoping:** The governed agent operations carry no resource id in the URL path, so bound your agent to the operations it needs, such as reading context and evaluating actions. You choose which operations it may call, so issuing and closing permits is not included unless you add it.
- **Credential handling:** Your Marrow API key is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent such as 'check if an action is allowed' or 'read agent context', and Jentic returns the matching Marrow operation with its input schema so the agent calls the right endpoint without reading the reference docs.

## Related APIs

- **Oso Cloud HTTP API** — Authorization service for roles, permissions, and policy checks
- **Auth0 Authentication API** — User authentication, SSO, and token issuance
- **Stytch API** — Authentication and session management

## FAQ

### Is there a Marrow Agent Governance API MCP server?

You do not need an MCP server to give your agent the Marrow Agent Governance API. Jentic connects it directly from the API Directory: import the API, store your key once, and your agent calls the context, evaluate, permit, and report operations. Operations are discovered on demand, so no extra server definitions load into the agent's context.

### What authentication does the Marrow Agent Governance API use?

The agent operations require a Marrow API key passed as a bearer token per the OpenAPI spec, while the public health and discovery endpoints need no credential. Through Jentic the key is stored encrypted in your own Jentic One instance and supplied only when an authenticated call runs, so it never reaches the agent's context.

### Can I evaluate an action before the agent runs it with the Marrow Agent Governance API?

Yes. The runtime evaluate operation takes a proposed action and returns governance guidance before execution, and a separate commit operation closes the action with its outcome and proof afterward. Together they bound what the agent does and leave a verifiable record of it.

### What are the rate limits for the Marrow Agent Governance API?

The OpenAPI spec does not specify rate limits. A public status endpoint reports component health you can poll, and you can confirm the request limits for your tenant in the Marrow documentation at https://getmarrow.ai.

### Can I limit what my agent is allowed to do with the Marrow Agent Governance API?

Yes. Write a rule that permits only the read operations, reading context and metrics, when the agent should observe governance, and add the permit and commit operations only for workflows that actually govern actions, with every call logged.

### How do I evaluate an agent action with the Marrow Agent Governance API through Jentic?

Search Jentic for 'check if an agent action is allowed' to surface the evaluate operation, load its input schema, pass the proposed action, and execute, then commit the outcome once the action runs. To run it on your own infrastructure, install Jentic One from its GitHub repo.