# Maetra

> Maetra is an AI governance and compliance control plane. It gives engineering, security, compliance, and governance teams one surface to discover AI systems, map applicable requirements, reuse and refresh control evidence, keep agent work aligned with approved tasks, protect employee use of browser AI, route consequential actions through exact policies or decision intelligence, block unsafe runtime behavior, and preserve an immutable audit trail. Built for regulated organizations.

Maetra is an AI governance and compliance control plane with five interoperating modules and two runtime guards. Discover finds the agent, Comply maps applicable requirements and keeps reusable evidence current, Govern applies policy and approval when required, Secure checks the action in real time, and Audit proves what happened. Task Guard keeps connected agents aligned with the active user-authorized task. Interaction Guard protects covered prompts and files before they enter supported browser AI applications. Maetra is API-first: agents can use REST APIs or the MCP server at decision points.

## Product modules

- [Discover](https://maetra.io/#discover): Finds and inventories every AI agent before it reaches production. Scans GitHub and connected repositories to detect official, shadow, and unreviewed agents (LangChain, CrewAI, AutoGen, Vercel AI SDK, raw OpenAI/Anthropic). Turns every pull request into a governance checkpoint. Classifies agents by autonomy level (L1–L5), risk, data access, and environment.
- [Comply](https://maetra.io/#comply): Multi-framework AI compliance. Maetra scans each agent's code (tools, data access, actions, model, environment) and maps what it finds across every applicable framework — EU AI Act, GDPR, NIST AI RMF, ISO/IEC 42001, Colorado AI Act, SOC 2, and more. It auto-detects the controls the code already satisfies, surfaces the remaining gaps, and reuses each piece of evidence across frameworks (classify an agent once, prove it everywhere). Generates technical documentation, transparency notices, DPIAs, conformity assessments, and evidence packs; tracks gaps, evidence freshness, and regulatory deadlines.
- [Govern](https://maetra.io/#govern): Human approval processes and decision intelligence for consequential AI agent actions. Teams can set exact policies for known actions or use AI agent decision intelligence when risk depends on runtime context. High-impact actions (transfers, external emails, data deletion, offers) can trigger role-based approvals with quorum policies, autonomy thresholds, timeouts, escalation paths, and cryptographically signed decision tokens.
- [Interaction Guard](https://maetra.io/download): Browser AI data protection for ChatGPT, Claude, Gemini, and Microsoft Copilot. It checks covered prompts and supported files before submission, then warns, requests a business reason, or blocks the interaction according to workspace Guards.
- [Task Guard](https://maetra.io/docs/task-guard-api): AI agent task alignment and task drift prevention. It maintains an active task contract, checks proposed actions for scope drift, confirms material task changes, refocuses off-task work, and verifies reported effects. Task Guard is available through REST and MCP.
- [Secure](https://maetra.io/#secure): Runtime protection. Scans prompts, tool calls, and outputs in real time for prompt injection, data exfiltration, privilege escalation, cross-tenant leakage, and policy violations before they reach production tools. Built-in and custom rules (regex, keyword, semantic, policy DSL); incidents are tracked and reviewable.
- [Audit](https://maetra.io/#audit): Immutable, hash-chained evidence layer. Records every discovery, classification, approval, rejection, policy change, and security incident. Full-text searchable; exportable as regulator-ready PDF, JSON, CSV, or to a SIEM. Included on every plan. Chain integrity is cryptographically verifiable (SHA-256).

## Compliance coverage

- [Supported frameworks](https://maetra.io/frameworks): 41 AI-obligation frameworks across the US (NIST AI RMF, Colorado AI Act, Texas TRAIGA, California TFAIA/ADMT, Utah, Illinois, NYC LL144, HIPAA, FDA AI/ML, NAIC), the EU (EU AI Act, GDPR, GPAI Code of Practice), national laws (UK, Canada AIDA, Brazil PL 2338, China GenAI, Japan, South Korea, India DPDPA, Singapore, Australia), global frameworks (OECD, UNESCO, OWASP LLM Top 10, MITRE ATLAS, CSA AICM, Google SAIF, Microsoft RAI), and international standards (ISO/IEC 42001, 42005, 23894, and IEEE 7000 — licensed, activated on demand).
- Each framework has its own page at `https://maetra.io/frameworks/{slug}` — what it is, who it applies to, its key obligations, and how Maetra maps agents to it. Examples: [EU AI Act](https://maetra.io/frameworks/eu-ai-act), [GDPR](https://maetra.io/frameworks/gdpr), [NIST AI RMF](https://maetra.io/frameworks/nist-ai-rmf), [ISO/IEC 42001](https://maetra.io/frameworks/iso-iec-42001), [Colorado AI Act](https://maetra.io/frameworks/colorado-ai-act), [Utah AI Policy Act](https://maetra.io/frameworks/utah-ai-policy-act), [NYC Local Law 144](https://maetra.io/frameworks/nyc-local-law-144).

## How compliance works (agent → compliance)

Maetra turns a code scan into compliance evidence, not a questionnaire:

1. Scan & fingerprint — Discover reads each agent's code (tools, data access and sensitivity, actions, model, environment) into an evidence-backed profile, tied to the exact file and commit.
2. Decide what applies — the profile determines which frameworks are in scope and which of their requirements apply to that agent.
3. Auto-detect controls & surface gaps — controls the code already satisfies are detected automatically from the scan; the rest become a clear list of gaps, each tied to a requirement and the evidence it still needs.
4. Prove & keep current — close gaps with linked evidence or generated documents, reused across every framework the same control supports; Maetra re-checks on each rescan and flags evidence that has gone stale.

A scan alone is not compliance — it captures the technical signals, auto-fills what the code can prove, and shows the gaps a human still needs to close with evidence or documentation.

## For developers

- API-first integration: agents use `/v1/task-guard/*` for active task alignment, `POST /v1/checkpoints` for exact policies or AI agent decision intelligence, and `POST /v1/secure/scan` for runtime scanning.
- `agent_id` remains optional for REST `/v1` and MCP. Requests can send only `agent_name`; unregistered or API-only agents are evaluated against organization-wide policies, while names or IDs that match registered agents can use agent-scoped policies.
- Official Node.js SDKs (`@maetra/govern-sdk`, `@maetra/secure-sdk`) and a Model Context Protocol (MCP) server with Task Guard, Govern, and Secure tools for embedding governance directly into agents.
- Integrations: GitHub (PR checks), Slack (approvals), Email (fallback), Webhooks (events), SIEM (audit stream); GitLab and Microsoft Teams planned.
- [Developer documentation](https://maetra.io/docs): Quickstarts, authentication, Task Guard, Govern and Secure API guides, decision-token verification, and MCP integration.
- [Machine-readable documentation index](https://maetra.io/docs/llms.txt): Direct Markdown links for every documentation page.

## Key pages

- [Homepage](https://maetra.io/): Product overview and positioning ("Control AI agents. Keep compliance evidence current.").
- [Press and reviewer resources](https://maetra.io/press): Verified product facts, direct evaluation paths, current limitations, reviewer contact, and official downloadable brand assets.
- [Interaction Guard](https://maetra.io/download): Browser extension installation, supported browsers, setup, and privacy boundaries.
- [Task Guard API](https://maetra.io/docs/task-guard-api): REST and MCP documentation for task contracts, alignment checks, scope-change confirmation, progress, effect verification, and task completion.
- [Pricing](https://maetra.io/pricing): Trial (14 days, free), Go ($20/mo), Starter ($39/mo), Team ($129/mo), Business ($599/mo), and Enterprise (SSO/SAML, SCIM, private cloud or on-premise). Task Guard, Govern, and Secure requests are metered; Interaction Guard is available on Starter and above with protected users tied to workspace seats; audit log is included on every plan.
- [About](https://maetra.io/about): Who Maetra is for and how the team thinks about AI governance and compliance.

## Guides

- [AI governance and compliance guides](https://maetra.io/guides): Plain-language explainers on governance, compliance, and runtime security for teams running AI agents.
- [What is AI governance?](https://maetra.io/guides/what-is-ai-governance): Definition, why it matters now, the five core components (discover, comply, govern, secure, audit), key frameworks, and how to implement it for autonomous AI agents.

## Legal

- [Privacy Policy](https://maetra.io/privacy)
- [Terms of Use](https://maetra.io/terms)

## Notes

- Maetra does not provide legal advice; generated compliance documents are drafts for review by qualified professionals.
- Primary positioning terms: AI governance, AI agent governance, AI governance platform, AI agent task alignment, AI agent task drift prevention, AI agent scope control, browser AI governance, ChatGPT data loss prevention, AI compliance software, EU AI Act compliance, AI agent inventory, shadow AI discovery, human-in-the-loop approvals, AI runtime security, prompt injection protection, AI audit trail.