AgentIndex icon
AgentIndex
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Submit Tool
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompare
Home/
Security & Safety/
agent-governance-toolkit
agent-governance-toolkit logo

agent-governance-toolkit

Active·★ 6.5k·MIT·Updated 2026-10-22
★ Trending

Agent Governance Toolkit (AGT) is Microsoft's answer to a specific problem: once agents call tools and delegate to each other, prompt instructions are not a reliable control. AGT intercepts every tool call, message and delegation in deterministic application code before it reaches the wire, so an action the policy denies cannot run.

Around that policy engine it adds per-agent identity, execution sandboxing, tamper-evident audit records and reliability tooling. The `agt` CLI checks installations, runs OWASP compliance checks, audits prompts for injection risk and lints policy files, and can fail CI when evidence is weak. The README maps its coverage to the OWASP Agentic Top 10.

Python has the full stack (`pip install "agent-governance-toolkit[full]"`, Python 3.11+), with SDKs for TypeScript and other languages and plugins for Claude Code, Copilot CLI, Codex CLI and OpenCode. It is labeled public preview, and the README notes breaking changes, including a recent policy model migration.

© 2026 AgentIndex.app|Built by a 10-year iOS Developer.
Best by use caseAboutContactAdvertisePrivacy PolicyTerms of UseQYSGitHubBuy me a coffee ☕

Browse by Category

MCP ServersCode AssistantWorkflow AutomationRAG / Knowledge BaseMulti-AgentBrowser AutomationLLM InfraDev ToolingObservability

Not affiliated with Anthropic, OpenAI or Microsoft.

Guides

This write-up is based on the project's README and public repository data. We have not hand-tested the tool. Data as of 2026-10-19.

License
MIT
Latest release
v4.1.0 · 2026-06-09
Last commit
2026-10-19
Contributors
168
Forks
1,141
Open issues and PRs
99
First published
2026-03-02
npm package
@microsoft/agent-governance-sdk 5.0.0 · 1,676 downloads last week

Source: GitHub, npm and PyPI public APIs.

#ai-safety#governance#policy-engine#zero-trust#owasp#python
$ Install
$ pip install "agent-governance-toolkit[full]"
↗ Visit site★ GitHub
01

Features

01Policy enforcement on tool calls, messages and delegations before they execute
02Per-agent identity and tamper-evident audit records
03Execution sandboxing and reliability (SRE) tooling
04agt CLI for OWASP checks, prompt-injection audits and policy linting
05Plugins for Claude Code, Copilot CLI, Codex CLI and OpenCode
02

Why choose it

+Denied actions are blocked in code, not by asking the model nicely.
+Gives each agent its own identity and keeps tamper-evident records for audits.
+Works across frameworks, with plugins for Claude Code, Codex and Copilot CLI.
03

Trade-offs

!Still a public preview with breaking changes before GA.
!Package layout is in flux: the base wheel installs only the compliance CLI, and older packages are deprecated.
!Only Python has the full feature set; other language SDKs cover the core.
04

Maintenance check

ActiveChecked: 2026-10-18
Last commit
2026-10-18
Commits, last 90 days
100+
Releases, last 90 days
0
Latest release
v4.1.0 (2026-06-09)
Open issues
96
Recent issues with at least one comment
8/9
Commits per week, last 12 weeks
Commits per week, last 12 weeks: 67, 22, 24, 3, 4, 37, 141, 99, 55, 37, 42, 7

Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

05

Compatibility

Claude Code
Plugin
Verified via docs
Codex
Plugin
Verified via docs
OpenCode
Plugin
Verified via docs
06

Quick start

1
$ pip install "agent-governance-toolkit[full]"
07

Use cases

↳Stop an agent with database access from running destructive queries
↳Attribute actions to specific agents in a multi-agent system
↳Produce audit evidence for compliance reviews and fail CI on weak evidence
08

Alternatives

⚡Top VS:
VSida-pro-mcp★13.1k→VSsleepless-agent★834→VSsupabase-mcp-server★833→

Comments

Log in to leave a comment

No comments yet. Be the first!

On this page
01Features02Why choose it03Trade-offs04Maintenance check05Compatibility06Quick start

Most discussed open issues

  • #3111 Implementation follow-up: EvidenceAnchor / mycelium-agt — ecosystem developments since #2244 merged19 comments · 2026-06-18
  • #3661 [Bug]: lint-policy accepts a rule whose condition can never match, and the evaluator silently applies the default7 comments · 2026-08-10
  • #3485 RFC: durable artifacts carry labels across sessions — extending accumulated-context governance past workflow lifetime6 comments · 2026-07-29
  • #3349 RFC: Port the evidence-led agent red-team benchmark into AGT's testing suite6 comments · 2026-07-16
  • #3084 feat: language parity for accumulated context governance (TypeScript, .NET, Rust, Go)6 comments · 2026-06-16

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

VS
martin-loop
★398
→
ida-pro-mcp logo
ida-pro-mcp★ 13.1k
AI-powered reverse engineering assistant that bridges IDA Pro with language models through MCP.
VSCompare→
sleepless-agent logo
sleepless-agent★ 834
🤖 24/7 AI agent that maximizes Claude Code Pro usage via Slack. Auto-processes tasks, manages isolated workspaces, creates Git commits/PRs, and optimizes day/night usage thresholds.
VSCompare→
supabase-mcp-server logo
supabase-mcp-server★ 833
Query MCP enables end-to-end management of Supabase via chat interface: read & write query executions, management API support, automatic migration versioning, access to logs and much more.
VSCompare→
martin-loop logo
martin-loop★ 398
The control plane for autonomous work and coding agent teams.
VSCompare→
node9-proxy logo
node9-proxy★ 219
The Execution Security Layer for the Agentic Era. Providing deterministic "Sudo" governance and audit logs for autonomous AI agents.
VSCompare→
c2detect logo
c2detect★ 43
C2 server fingerprinter — Cobalt Strike, Sliver, Mythic, Havoc, Brute Ratel
VSCompare→
arifOS logo
arifOS★ 54
ArifOS — AAA MCP-governed constitutional kernel for AI agents.
VSCompare→
mcp-server-for-oscal logo
mcp-server-for-oscal★ 52
OSCAL tools for AI agents
VSCompare→
See all alternatives →
07
Use cases
08Alternatives
microsoft /
agent-governance-toolkit
↗

Microsoft's toolkit that blocks disallowed agent actions in code, with identity, sandboxing and audit trails

6.5k1.1kMIT
Last commit1d ago
StatusActive
CategorySecurity & Safety
Contributors168
Latest releasev4.1.0
Links
Documentation↗Discussion↗Issues↗Releases↗
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit