agent_governance_toolkit 3.0.2__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- agent_governance_toolkit-3.0.2/LICENSE +21 -0
- agent_governance_toolkit-3.0.2/PKG-INFO +318 -0
- agent_governance_toolkit-3.0.2/README.md +275 -0
- agent_governance_toolkit-3.0.2/pyproject.toml +64 -0
- agent_governance_toolkit-3.0.2/setup.cfg +4 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/__init__.py +34 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/cli/__init__.py +2 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/cli/main.py +198 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/governance/__init__.py +27 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/governance/attestation_validator.py +141 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/integrity.py +386 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/lint_policy.py +346 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/promotion.py +410 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/security/__init__.py +31 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/security/scanner.py +906 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/supply_chain.py +464 -0
- agent_governance_toolkit-3.0.2/src/agent_compliance/verify.py +328 -0
- agent_governance_toolkit-3.0.2/src/agent_governance_toolkit.egg-info/PKG-INFO +318 -0
- agent_governance_toolkit-3.0.2/src/agent_governance_toolkit.egg-info/SOURCES.txt +28 -0
- agent_governance_toolkit-3.0.2/src/agent_governance_toolkit.egg-info/dependency_links.txt +1 -0
- agent_governance_toolkit-3.0.2/src/agent_governance_toolkit.egg-info/entry_points.txt +4 -0
- agent_governance_toolkit-3.0.2/src/agent_governance_toolkit.egg-info/requires.txt +18 -0
- agent_governance_toolkit-3.0.2/src/agent_governance_toolkit.egg-info/top_level.txt +1 -0
- agent_governance_toolkit-3.0.2/tests/test_cli_edge_cases.py +124 -0
- agent_governance_toolkit-3.0.2/tests/test_governance_attestation.py +527 -0
- agent_governance_toolkit-3.0.2/tests/test_integrity_and_verify.py +293 -0
- agent_governance_toolkit-3.0.2/tests/test_lint_policy.py +479 -0
- agent_governance_toolkit-3.0.2/tests/test_promotion.py +359 -0
- agent_governance_toolkit-3.0.2/tests/test_security_scanner.py +673 -0
- agent_governance_toolkit-3.0.2/tests/test_supply_chain.py +341 -0
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
MIT License
|
|
2
|
+
|
|
3
|
+
Copyright (c) Microsoft Corporation.
|
|
4
|
+
|
|
5
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
6
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
7
|
+
in the Software without restriction, including without limitation the rights
|
|
8
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
9
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
10
|
+
furnished to do so, subject to the following conditions:
|
|
11
|
+
|
|
12
|
+
The above copyright notice and this permission notice shall be included in all
|
|
13
|
+
copies or substantial portions of the Software.
|
|
14
|
+
|
|
15
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
16
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
17
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
18
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
19
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
20
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
21
|
+
SOFTWARE.
|
|
@@ -0,0 +1,318 @@
|
|
|
1
|
+
Metadata-Version: 2.1
|
|
2
|
+
Name: agent_governance_toolkit
|
|
3
|
+
Version: 3.0.2
|
|
4
|
+
Summary: Public Preview — Unified installer and runtime policy enforcement for the Agent Governance Toolkit
|
|
5
|
+
Author-email: Microsoft Corporation <agentgovtoolkit@microsoft.com>
|
|
6
|
+
Maintainer-email: Agent Governance Toolkit Team <agentgovtoolkit@microsoft.com>
|
|
7
|
+
License: MIT
|
|
8
|
+
Project-URL: Homepage, https://github.com/microsoft/agent-governance-toolkit
|
|
9
|
+
Project-URL: Documentation, https://github.com/microsoft/agent-governance-toolkit/tree/main/docs
|
|
10
|
+
Project-URL: Repository, https://github.com/microsoft/agent-governance-toolkit
|
|
11
|
+
Project-URL: Bug Tracker, https://github.com/microsoft/agent-governance-toolkit/issues
|
|
12
|
+
Project-URL: Agent OS, https://github.com/microsoft/agent-governance-toolkit
|
|
13
|
+
Project-URL: AgentMesh, https://github.com/microsoft/agent-governance-toolkit
|
|
14
|
+
Project-URL: Agent Runtime, https://github.com/microsoft/agent-governance-toolkit
|
|
15
|
+
Project-URL: Agent SRE, https://github.com/microsoft/agent-governance-toolkit
|
|
16
|
+
Keywords: ai-agents,governance,multi-agent,trust,security,compliance,audit,policy-enforcement,agent-os,agentmesh,runtime,sre,llm,autonomous-agents,enterprise-ai
|
|
17
|
+
Classifier: Development Status :: 4 - Beta
|
|
18
|
+
Classifier: Intended Audience :: Developers
|
|
19
|
+
Classifier: Programming Language :: Python :: 3
|
|
20
|
+
Classifier: Programming Language :: Python :: 3.9
|
|
21
|
+
Classifier: Programming Language :: Python :: 3.10
|
|
22
|
+
Classifier: Programming Language :: Python :: 3.11
|
|
23
|
+
Classifier: Programming Language :: Python :: 3.12
|
|
24
|
+
Classifier: Topic :: Security
|
|
25
|
+
Classifier: Topic :: Software Development :: Libraries :: Application Frameworks
|
|
26
|
+
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
|
|
27
|
+
Requires-Python: >=3.9
|
|
28
|
+
Description-Content-Type: text/markdown
|
|
29
|
+
License-File: LICENSE
|
|
30
|
+
Requires-Dist: agent-os-kernel<2.0,>=1.0.0
|
|
31
|
+
Requires-Dist: agentmesh-platform<2.0,>=1.0.0
|
|
32
|
+
Requires-Dist: pydantic<3.0,>=2.4.0
|
|
33
|
+
Provides-Extra: runtime
|
|
34
|
+
Requires-Dist: agentmesh-runtime<3.0,>=2.0.0; extra == "runtime"
|
|
35
|
+
Provides-Extra: sre
|
|
36
|
+
Requires-Dist: agent-sre<2.0,>=1.0.0; extra == "sre"
|
|
37
|
+
Provides-Extra: opa
|
|
38
|
+
Provides-Extra: cedar
|
|
39
|
+
Requires-Dist: cedarpy<5.0,>=4.0.0; extra == "cedar"
|
|
40
|
+
Provides-Extra: full
|
|
41
|
+
Requires-Dist: agentmesh-runtime<3.0,>=2.0.0; extra == "full"
|
|
42
|
+
Requires-Dist: agent-sre<2.0,>=1.0.0; extra == "full"
|
|
43
|
+
|
|
44
|
+
<div align="center">
|
|
45
|
+
|
|
46
|
+
# Agent Governance
|
|
47
|
+
|
|
48
|
+
**Unified installer and runtime policy enforcement for the Agent Governance Toolkit**
|
|
49
|
+
|
|
50
|
+
*One install for the complete governance stack — kernel · trust mesh · runtime supervisor · reliability engineering*
|
|
51
|
+
|
|
52
|
+
[](https://github.com/microsoft/agent-governance-toolkit/actions/workflows/ci.yml)
|
|
53
|
+
[](../../LICENSE)
|
|
54
|
+
[](https://python.org)
|
|
55
|
+
[](https://pypi.org/project/agent-governance-toolkit/)
|
|
56
|
+
|
|
57
|
+
> [!IMPORTANT]
|
|
58
|
+
> **Public Preview** — The `agent-governance-toolkit` package on PyPI is a Microsoft-signed
|
|
59
|
+
> public preview release. APIs may change before GA.
|
|
60
|
+
|
|
61
|
+
```
|
|
62
|
+
pip install agent-governance-toolkit[full]
|
|
63
|
+
```
|
|
64
|
+
|
|
65
|
+
[Architecture](#architecture) • [Quick Start](#quick-start) • [Components](#components) • [Why Unified?](#why-a-unified-governance-stack) • [Ecosystem](#the-agent-governance-ecosystem) • [OWASP Compliance](docs/OWASP-COMPLIANCE.md) • [Traction](docs/TRACTION.md)
|
|
66
|
+
|
|
67
|
+
</div>
|
|
68
|
+
|
|
69
|
+
> ⭐ **If this project helps you, please star it!** It helps others discover the agent governance stack.
|
|
70
|
+
|
|
71
|
+
> 🔗 **Part of the Agent Governance Ecosystem** — Installs [Agent OS](https://github.com/microsoft/agent-governance-toolkit) · [AgentMesh](https://github.com/microsoft/agent-governance-toolkit) · [Agent Runtime](https://github.com/microsoft/agent-governance-toolkit) · [Agent SRE](https://github.com/microsoft/agent-governance-toolkit)
|
|
72
|
+
|
|
73
|
+
> **Migrating from `ai-agent-compliance`?** The package has been renamed to `agent-governance-toolkit`.
|
|
74
|
+
> Run `pip install agent-governance` — the old name is deprecated and will redirect here for 6 months.
|
|
75
|
+
|
|
76
|
+
---
|
|
77
|
+
|
|
78
|
+
## Architecture
|
|
79
|
+
|
|
80
|
+
```
|
|
81
|
+
┌─────────────────────────────────────────────────────────────────┐
|
|
82
|
+
│ agent-governance │
|
|
83
|
+
│ pip install agent-governance-toolkit[full] │
|
|
84
|
+
├─────────────────────────────────────────────────────────────────┤
|
|
85
|
+
│ │
|
|
86
|
+
│ ┌───────────────────┐ ┌───────────────────────────┐ │
|
|
87
|
+
│ │ Agent OS Kernel │◄────►│ AgentMesh Platform │ │
|
|
88
|
+
│ │ │ │ │ │
|
|
89
|
+
│ │ Policy Engine │ │ Zero-Trust Identity │ │
|
|
90
|
+
│ │ Capability Model │ │ Mutual TLS for Agents │ │
|
|
91
|
+
│ │ Audit Logging │ │ Encrypted Channels │ │
|
|
92
|
+
│ │ Syscall Layer │ │ Trust Scoring │ │
|
|
93
|
+
│ └────────┬──────────┘ └─────────────┬─────────────┘ │
|
|
94
|
+
│ │ │ │
|
|
95
|
+
│ ▼ ▼ │
|
|
96
|
+
│ ┌───────────────────┐ ┌───────────────────────────┐ │
|
|
97
|
+
│ │ Agent Runtime │ │ Agent SRE │ │
|
|
98
|
+
│ │ │ │ │ │
|
|
99
|
+
│ │ Execution Rings │ │ Health Monitoring │ │
|
|
100
|
+
│ │ Resource Limits │ │ SLO Enforcement │ │
|
|
101
|
+
│ │ Runtime Sandboxing│ │ Incident Response │ │
|
|
102
|
+
│ │ Kill Switch │ │ Chaos Engineering │ │
|
|
103
|
+
│ └───────────────────┘ └───────────────────────────┘ │
|
|
104
|
+
│ │
|
|
105
|
+
└─────────────────────────────────────────────────────────────────┘
|
|
106
|
+
```
|
|
107
|
+
|
|
108
|
+
---
|
|
109
|
+
|
|
110
|
+
## Quick Start
|
|
111
|
+
|
|
112
|
+
```python
|
|
113
|
+
import asyncio
|
|
114
|
+
from agent_os import StatelessKernel, ExecutionContext
|
|
115
|
+
from agentmesh import AgentIdentity
|
|
116
|
+
|
|
117
|
+
# Boot the governance kernel
|
|
118
|
+
kernel = StatelessKernel()
|
|
119
|
+
ctx = ExecutionContext(agent_id="my-agent", policies=["read_only"])
|
|
120
|
+
|
|
121
|
+
# Establish zero-trust agent identity
|
|
122
|
+
identity = AgentIdentity.create(
|
|
123
|
+
name="my-agent",
|
|
124
|
+
sponsor="alice@company.com",
|
|
125
|
+
capabilities=["read:data", "write:reports"],
|
|
126
|
+
)
|
|
127
|
+
|
|
128
|
+
# Execute a governed action
|
|
129
|
+
async def main():
|
|
130
|
+
result = await kernel.execute(
|
|
131
|
+
action="database_query",
|
|
132
|
+
params={"query": "SELECT * FROM users"},
|
|
133
|
+
context=ctx,
|
|
134
|
+
)
|
|
135
|
+
print(f"Success: {result.success}, Data: {result.data}")
|
|
136
|
+
|
|
137
|
+
asyncio.run(main())
|
|
138
|
+
```
|
|
139
|
+
|
|
140
|
+
### Compliance Grading
|
|
141
|
+
|
|
142
|
+
Check your governance coverage with a compliance grade:
|
|
143
|
+
|
|
144
|
+
```python
|
|
145
|
+
from agent_compliance.verify import GovernanceVerifier
|
|
146
|
+
|
|
147
|
+
verifier = GovernanceVerifier()
|
|
148
|
+
attestation = verifier.verify()
|
|
149
|
+
print(f"Grade: {attestation.compliance_grade()}") # A, B, C, D, or F
|
|
150
|
+
print(f"Coverage: {attestation.coverage_pct()}%")
|
|
151
|
+
print(attestation.badge_markdown())
|
|
152
|
+
```
|
|
153
|
+
|
|
154
|
+
Install only what you need:
|
|
155
|
+
|
|
156
|
+
```bash
|
|
157
|
+
# Core: kernel + trust mesh
|
|
158
|
+
pip install agent-governance-toolkit
|
|
159
|
+
|
|
160
|
+
# Full stack: adds runtime + SRE
|
|
161
|
+
pip install agent-governance-toolkit[full]
|
|
162
|
+
|
|
163
|
+
# À la carte
|
|
164
|
+
pip install agent-governance-toolkit[runtime]
|
|
165
|
+
pip install agent-governance-toolkit[sre]
|
|
166
|
+
```
|
|
167
|
+
|
|
168
|
+
---
|
|
169
|
+
|
|
170
|
+
## Components
|
|
171
|
+
|
|
172
|
+
| Package | Role |
|
|
173
|
+
|---------|------|
|
|
174
|
+
| **Agent OS** | Policy engine — deterministic action evaluation |
|
|
175
|
+
| **AgentMesh** | Trust infrastructure — identity, credentials, protocol bridges |
|
|
176
|
+
| **Agent Runtime** | Execution supervisor — rings, sessions, sagas |
|
|
177
|
+
| **Agent SRE** | Reliability — SLOs, circuit breakers, chaos testing |
|
|
178
|
+
| **Agent Compliance** | Regulatory compliance — GDPR, HIPAA, SOX frameworks *(this package)* |
|
|
179
|
+
| **Agent Marketplace** | Plugin lifecycle — discover, install, verify, sign |
|
|
180
|
+
| **Agent Lightning** | RL training governance — governed runners, policy rewards |
|
|
181
|
+
|
|
182
|
+
### Star the ecosystem
|
|
183
|
+
|
|
184
|
+
<p align="center">
|
|
185
|
+
|
|
186
|
+
[](https://github.com/microsoft/agent-governance-toolkit)
|
|
187
|
+
[](https://github.com/microsoft/agent-governance-toolkit)
|
|
188
|
+
[](https://github.com/microsoft/agent-governance-toolkit)
|
|
189
|
+
[](https://github.com/microsoft/agent-governance-toolkit)
|
|
190
|
+
|
|
191
|
+
</p>
|
|
192
|
+
|
|
193
|
+
---
|
|
194
|
+
|
|
195
|
+
## Why a Unified Governance Stack?
|
|
196
|
+
|
|
197
|
+
Running AI agents in production without governance is like deploying microservices without TLS, RBAC, or monitoring. Each layer solves a different problem:
|
|
198
|
+
|
|
199
|
+
| Concern | Without Governance | With Agent Governance |
|
|
200
|
+
|---------|-------------------|----------------------|
|
|
201
|
+
| **Security** | Agents call any tool, access any resource | Capability-based permissions, policy enforcement |
|
|
202
|
+
| **Trust** | No identity verification between agents | Mutual TLS, trust scores, encrypted channels |
|
|
203
|
+
| **Control** | Runaway agents consume unbounded resources | Execution rings, resource limits, kill switches |
|
|
204
|
+
| **Reliability** | Silent failures, no observability | SLO enforcement, health checks, incident automation |
|
|
205
|
+
| **Compliance** | No audit trail for agent decisions | Immutable audit logs, decision lineage tracking |
|
|
206
|
+
|
|
207
|
+
**One install. Four layers of protection.**
|
|
208
|
+
|
|
209
|
+
The meta-package ensures all components are version-compatible and properly integrated. No dependency conflicts, no version mismatches — just a single `pip install` to go from zero to production-grade agent governance.
|
|
210
|
+
|
|
211
|
+
---
|
|
212
|
+
|
|
213
|
+
## The Agent Governance Ecosystem
|
|
214
|
+
|
|
215
|
+
```
|
|
216
|
+
agent-governance ─── The meta-package (you are here)
|
|
217
|
+
├── agent-os-kernel ─── Governance kernel
|
|
218
|
+
├── agentmesh-platform ─── Zero-trust mesh
|
|
219
|
+
├── agentmesh-runtime ─── Runtime supervisor (optional)
|
|
220
|
+
└── agent-sre ─── Reliability engineering (optional)
|
|
221
|
+
```
|
|
222
|
+
|
|
223
|
+
Each component works standalone, but they're designed to work together. The kernel enforces policy, the mesh secures communication, the runtime controls execution, and SRE keeps everything running.
|
|
224
|
+
|
|
225
|
+
---
|
|
226
|
+
|
|
227
|
+
## Examples
|
|
228
|
+
|
|
229
|
+
See the [`examples/`](examples/) directory for runnable demos:
|
|
230
|
+
|
|
231
|
+
```bash
|
|
232
|
+
# Quick start — boot the governance stack in 30 lines
|
|
233
|
+
python examples/quickstart.py
|
|
234
|
+
|
|
235
|
+
# Full stack — all 4 layers working together
|
|
236
|
+
python examples/governed_agent.py
|
|
237
|
+
```
|
|
238
|
+
|
|
239
|
+
---
|
|
240
|
+
|
|
241
|
+
## Framework Integration
|
|
242
|
+
|
|
243
|
+
```bash
|
|
244
|
+
# LangChain
|
|
245
|
+
pip install langchain agent-governance
|
|
246
|
+
|
|
247
|
+
# CrewAI
|
|
248
|
+
pip install crewai agent-governance
|
|
249
|
+
|
|
250
|
+
# AutoGen
|
|
251
|
+
pip install pyautogen agent-governance
|
|
252
|
+
```
|
|
253
|
+
|
|
254
|
+
---
|
|
255
|
+
|
|
256
|
+
## 🗺️ Roadmap
|
|
257
|
+
|
|
258
|
+
| Quarter | Milestone |
|
|
259
|
+
|---------|-----------|
|
|
260
|
+
| **Q1 2026** | ✅ Unified meta-package, 4 components integrated, PyPI published |
|
|
261
|
+
| **Q2 2026** | Cross-component integration tests, unified CLI, dashboard UI |
|
|
262
|
+
| **Q3 2026** | Helm chart for Kubernetes, managed cloud preview |
|
|
263
|
+
| **Q4 2026** | SOC2 Type II certification, enterprise support tier |
|
|
264
|
+
|
|
265
|
+
---
|
|
266
|
+
|
|
267
|
+
## 🛡️ OWASP Agentic Top 10 Coverage
|
|
268
|
+
|
|
269
|
+
The agent governance stack covers **10 of 10** risks from the [OWASP Top 10 for Agentic Applications (2026)](https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/):
|
|
270
|
+
|
|
271
|
+
| OWASP Risk | Coverage | Component |
|
|
272
|
+
|-----------|----------|-----------|
|
|
273
|
+
| Agent Goal Hijack | ✅ | Agent OS — Policy Engine |
|
|
274
|
+
| Tool Misuse | ✅ | Agent OS — Capability Sandboxing |
|
|
275
|
+
| Identity & Privilege Abuse | ✅ | AgentMesh — DID Identity |
|
|
276
|
+
| Supply Chain Vulnerabilities | ✅ | AgentMesh — AI-BOM v2.0 |
|
|
277
|
+
| Unexpected Code Execution | ✅ | Agent Runtime — Execution Rings |
|
|
278
|
+
| Memory & Context Poisoning | ✅ | Agent OS — VFS + CMVK |
|
|
279
|
+
| Insecure Inter-Agent Communication | ✅ | AgentMesh — IATP Protocol |
|
|
280
|
+
| Cascading Failures | ✅ | Agent SRE — Circuit Breakers |
|
|
281
|
+
| Human-Agent Trust Exploitation | ✅ | Agent OS — Approval Workflows |
|
|
282
|
+
| Rogue Agents | ✅ | Agent Runtime — Kill Switch |
|
|
283
|
+
|
|
284
|
+
**[→ Full OWASP compliance mapping with code examples](docs/OWASP-COMPLIANCE.md)**
|
|
285
|
+
|
|
286
|
+
---
|
|
287
|
+
|
|
288
|
+
## 📈 Traction
|
|
289
|
+
|
|
290
|
+
The ecosystem is growing — **3,000+ views, 9,400+ clones, and 1,278 unique developers** in the last 14 days alone. Traffic from Medium, Reddit, LinkedIn, Google, and even ChatGPT.
|
|
291
|
+
|
|
292
|
+
**[→ See full traction report](docs/TRACTION.md)**
|
|
293
|
+
|
|
294
|
+
---
|
|
295
|
+
|
|
296
|
+
## Contributing
|
|
297
|
+
|
|
298
|
+
We welcome contributions! See our [Contributing Guide](CONTRIBUTING.md) for details.
|
|
299
|
+
|
|
300
|
+
For component-specific contributions, see:
|
|
301
|
+
- [Agent OS](https://github.com/microsoft/agent-governance-toolkit/blob/master/CONTRIBUTING.md)
|
|
302
|
+
- [AgentMesh](https://github.com/microsoft/agent-governance-toolkit/blob/master/CONTRIBUTING.md)
|
|
303
|
+
- [Agent Runtime](https://github.com/microsoft/agent-governance-toolkit/blob/master/CONTRIBUTING.md)
|
|
304
|
+
- [Agent SRE](https://github.com/microsoft/agent-governance-toolkit/blob/master/CONTRIBUTING.md)
|
|
305
|
+
|
|
306
|
+
## License
|
|
307
|
+
|
|
308
|
+
MIT — see [LICENSE](LICENSE) for details.
|
|
309
|
+
|
|
310
|
+
---
|
|
311
|
+
|
|
312
|
+
<div align="center">
|
|
313
|
+
|
|
314
|
+
**[github.com/microsoft/agent-governance-toolkit](https://github.com/microsoft/agent-governance-toolkit)** · **[Documentation](https://github.com/microsoft/agent-governance-toolkit/tree/main/docs)** · **[GitHub](https://github.com/microsoft/agent-governance-toolkit)**
|
|
315
|
+
|
|
316
|
+
*Building the governance layer for the agentic era*
|
|
317
|
+
|
|
318
|
+
</div>
|
|
@@ -0,0 +1,275 @@
|
|
|
1
|
+
<div align="center">
|
|
2
|
+
|
|
3
|
+
# Agent Governance
|
|
4
|
+
|
|
5
|
+
**Unified installer and runtime policy enforcement for the Agent Governance Toolkit**
|
|
6
|
+
|
|
7
|
+
*One install for the complete governance stack — kernel · trust mesh · runtime supervisor · reliability engineering*
|
|
8
|
+
|
|
9
|
+
[](https://github.com/microsoft/agent-governance-toolkit/actions/workflows/ci.yml)
|
|
10
|
+
[](../../LICENSE)
|
|
11
|
+
[](https://python.org)
|
|
12
|
+
[](https://pypi.org/project/agent-governance-toolkit/)
|
|
13
|
+
|
|
14
|
+
> [!IMPORTANT]
|
|
15
|
+
> **Public Preview** — The `agent-governance-toolkit` package on PyPI is a Microsoft-signed
|
|
16
|
+
> public preview release. APIs may change before GA.
|
|
17
|
+
|
|
18
|
+
```
|
|
19
|
+
pip install agent-governance-toolkit[full]
|
|
20
|
+
```
|
|
21
|
+
|
|
22
|
+
[Architecture](#architecture) • [Quick Start](#quick-start) • [Components](#components) • [Why Unified?](#why-a-unified-governance-stack) • [Ecosystem](#the-agent-governance-ecosystem) • [OWASP Compliance](docs/OWASP-COMPLIANCE.md) • [Traction](docs/TRACTION.md)
|
|
23
|
+
|
|
24
|
+
</div>
|
|
25
|
+
|
|
26
|
+
> ⭐ **If this project helps you, please star it!** It helps others discover the agent governance stack.
|
|
27
|
+
|
|
28
|
+
> 🔗 **Part of the Agent Governance Ecosystem** — Installs [Agent OS](https://github.com/microsoft/agent-governance-toolkit) · [AgentMesh](https://github.com/microsoft/agent-governance-toolkit) · [Agent Runtime](https://github.com/microsoft/agent-governance-toolkit) · [Agent SRE](https://github.com/microsoft/agent-governance-toolkit)
|
|
29
|
+
|
|
30
|
+
> **Migrating from `ai-agent-compliance`?** The package has been renamed to `agent-governance-toolkit`.
|
|
31
|
+
> Run `pip install agent-governance` — the old name is deprecated and will redirect here for 6 months.
|
|
32
|
+
|
|
33
|
+
---
|
|
34
|
+
|
|
35
|
+
## Architecture
|
|
36
|
+
|
|
37
|
+
```
|
|
38
|
+
┌─────────────────────────────────────────────────────────────────┐
|
|
39
|
+
│ agent-governance │
|
|
40
|
+
│ pip install agent-governance-toolkit[full] │
|
|
41
|
+
├─────────────────────────────────────────────────────────────────┤
|
|
42
|
+
│ │
|
|
43
|
+
│ ┌───────────────────┐ ┌───────────────────────────┐ │
|
|
44
|
+
│ │ Agent OS Kernel │◄────►│ AgentMesh Platform │ │
|
|
45
|
+
│ │ │ │ │ │
|
|
46
|
+
│ │ Policy Engine │ │ Zero-Trust Identity │ │
|
|
47
|
+
│ │ Capability Model │ │ Mutual TLS for Agents │ │
|
|
48
|
+
│ │ Audit Logging │ │ Encrypted Channels │ │
|
|
49
|
+
│ │ Syscall Layer │ │ Trust Scoring │ │
|
|
50
|
+
│ └────────┬──────────┘ └─────────────┬─────────────┘ │
|
|
51
|
+
│ │ │ │
|
|
52
|
+
│ ▼ ▼ │
|
|
53
|
+
│ ┌───────────────────┐ ┌───────────────────────────┐ │
|
|
54
|
+
│ │ Agent Runtime │ │ Agent SRE │ │
|
|
55
|
+
│ │ │ │ │ │
|
|
56
|
+
│ │ Execution Rings │ │ Health Monitoring │ │
|
|
57
|
+
│ │ Resource Limits │ │ SLO Enforcement │ │
|
|
58
|
+
│ │ Runtime Sandboxing│ │ Incident Response │ │
|
|
59
|
+
│ │ Kill Switch │ │ Chaos Engineering │ │
|
|
60
|
+
│ └───────────────────┘ └───────────────────────────┘ │
|
|
61
|
+
│ │
|
|
62
|
+
└─────────────────────────────────────────────────────────────────┘
|
|
63
|
+
```
|
|
64
|
+
|
|
65
|
+
---
|
|
66
|
+
|
|
67
|
+
## Quick Start
|
|
68
|
+
|
|
69
|
+
```python
|
|
70
|
+
import asyncio
|
|
71
|
+
from agent_os import StatelessKernel, ExecutionContext
|
|
72
|
+
from agentmesh import AgentIdentity
|
|
73
|
+
|
|
74
|
+
# Boot the governance kernel
|
|
75
|
+
kernel = StatelessKernel()
|
|
76
|
+
ctx = ExecutionContext(agent_id="my-agent", policies=["read_only"])
|
|
77
|
+
|
|
78
|
+
# Establish zero-trust agent identity
|
|
79
|
+
identity = AgentIdentity.create(
|
|
80
|
+
name="my-agent",
|
|
81
|
+
sponsor="alice@company.com",
|
|
82
|
+
capabilities=["read:data", "write:reports"],
|
|
83
|
+
)
|
|
84
|
+
|
|
85
|
+
# Execute a governed action
|
|
86
|
+
async def main():
|
|
87
|
+
result = await kernel.execute(
|
|
88
|
+
action="database_query",
|
|
89
|
+
params={"query": "SELECT * FROM users"},
|
|
90
|
+
context=ctx,
|
|
91
|
+
)
|
|
92
|
+
print(f"Success: {result.success}, Data: {result.data}")
|
|
93
|
+
|
|
94
|
+
asyncio.run(main())
|
|
95
|
+
```
|
|
96
|
+
|
|
97
|
+
### Compliance Grading
|
|
98
|
+
|
|
99
|
+
Check your governance coverage with a compliance grade:
|
|
100
|
+
|
|
101
|
+
```python
|
|
102
|
+
from agent_compliance.verify import GovernanceVerifier
|
|
103
|
+
|
|
104
|
+
verifier = GovernanceVerifier()
|
|
105
|
+
attestation = verifier.verify()
|
|
106
|
+
print(f"Grade: {attestation.compliance_grade()}") # A, B, C, D, or F
|
|
107
|
+
print(f"Coverage: {attestation.coverage_pct()}%")
|
|
108
|
+
print(attestation.badge_markdown())
|
|
109
|
+
```
|
|
110
|
+
|
|
111
|
+
Install only what you need:
|
|
112
|
+
|
|
113
|
+
```bash
|
|
114
|
+
# Core: kernel + trust mesh
|
|
115
|
+
pip install agent-governance-toolkit
|
|
116
|
+
|
|
117
|
+
# Full stack: adds runtime + SRE
|
|
118
|
+
pip install agent-governance-toolkit[full]
|
|
119
|
+
|
|
120
|
+
# À la carte
|
|
121
|
+
pip install agent-governance-toolkit[runtime]
|
|
122
|
+
pip install agent-governance-toolkit[sre]
|
|
123
|
+
```
|
|
124
|
+
|
|
125
|
+
---
|
|
126
|
+
|
|
127
|
+
## Components
|
|
128
|
+
|
|
129
|
+
| Package | Role |
|
|
130
|
+
|---------|------|
|
|
131
|
+
| **Agent OS** | Policy engine — deterministic action evaluation |
|
|
132
|
+
| **AgentMesh** | Trust infrastructure — identity, credentials, protocol bridges |
|
|
133
|
+
| **Agent Runtime** | Execution supervisor — rings, sessions, sagas |
|
|
134
|
+
| **Agent SRE** | Reliability — SLOs, circuit breakers, chaos testing |
|
|
135
|
+
| **Agent Compliance** | Regulatory compliance — GDPR, HIPAA, SOX frameworks *(this package)* |
|
|
136
|
+
| **Agent Marketplace** | Plugin lifecycle — discover, install, verify, sign |
|
|
137
|
+
| **Agent Lightning** | RL training governance — governed runners, policy rewards |
|
|
138
|
+
|
|
139
|
+
### Star the ecosystem
|
|
140
|
+
|
|
141
|
+
<p align="center">
|
|
142
|
+
|
|
143
|
+
[](https://github.com/microsoft/agent-governance-toolkit)
|
|
144
|
+
[](https://github.com/microsoft/agent-governance-toolkit)
|
|
145
|
+
[](https://github.com/microsoft/agent-governance-toolkit)
|
|
146
|
+
[](https://github.com/microsoft/agent-governance-toolkit)
|
|
147
|
+
|
|
148
|
+
</p>
|
|
149
|
+
|
|
150
|
+
---
|
|
151
|
+
|
|
152
|
+
## Why a Unified Governance Stack?
|
|
153
|
+
|
|
154
|
+
Running AI agents in production without governance is like deploying microservices without TLS, RBAC, or monitoring. Each layer solves a different problem:
|
|
155
|
+
|
|
156
|
+
| Concern | Without Governance | With Agent Governance |
|
|
157
|
+
|---------|-------------------|----------------------|
|
|
158
|
+
| **Security** | Agents call any tool, access any resource | Capability-based permissions, policy enforcement |
|
|
159
|
+
| **Trust** | No identity verification between agents | Mutual TLS, trust scores, encrypted channels |
|
|
160
|
+
| **Control** | Runaway agents consume unbounded resources | Execution rings, resource limits, kill switches |
|
|
161
|
+
| **Reliability** | Silent failures, no observability | SLO enforcement, health checks, incident automation |
|
|
162
|
+
| **Compliance** | No audit trail for agent decisions | Immutable audit logs, decision lineage tracking |
|
|
163
|
+
|
|
164
|
+
**One install. Four layers of protection.**
|
|
165
|
+
|
|
166
|
+
The meta-package ensures all components are version-compatible and properly integrated. No dependency conflicts, no version mismatches — just a single `pip install` to go from zero to production-grade agent governance.
|
|
167
|
+
|
|
168
|
+
---
|
|
169
|
+
|
|
170
|
+
## The Agent Governance Ecosystem
|
|
171
|
+
|
|
172
|
+
```
|
|
173
|
+
agent-governance ─── The meta-package (you are here)
|
|
174
|
+
├── agent-os-kernel ─── Governance kernel
|
|
175
|
+
├── agentmesh-platform ─── Zero-trust mesh
|
|
176
|
+
├── agentmesh-runtime ─── Runtime supervisor (optional)
|
|
177
|
+
└── agent-sre ─── Reliability engineering (optional)
|
|
178
|
+
```
|
|
179
|
+
|
|
180
|
+
Each component works standalone, but they're designed to work together. The kernel enforces policy, the mesh secures communication, the runtime controls execution, and SRE keeps everything running.
|
|
181
|
+
|
|
182
|
+
---
|
|
183
|
+
|
|
184
|
+
## Examples
|
|
185
|
+
|
|
186
|
+
See the [`examples/`](examples/) directory for runnable demos:
|
|
187
|
+
|
|
188
|
+
```bash
|
|
189
|
+
# Quick start — boot the governance stack in 30 lines
|
|
190
|
+
python examples/quickstart.py
|
|
191
|
+
|
|
192
|
+
# Full stack — all 4 layers working together
|
|
193
|
+
python examples/governed_agent.py
|
|
194
|
+
```
|
|
195
|
+
|
|
196
|
+
---
|
|
197
|
+
|
|
198
|
+
## Framework Integration
|
|
199
|
+
|
|
200
|
+
```bash
|
|
201
|
+
# LangChain
|
|
202
|
+
pip install langchain agent-governance
|
|
203
|
+
|
|
204
|
+
# CrewAI
|
|
205
|
+
pip install crewai agent-governance
|
|
206
|
+
|
|
207
|
+
# AutoGen
|
|
208
|
+
pip install pyautogen agent-governance
|
|
209
|
+
```
|
|
210
|
+
|
|
211
|
+
---
|
|
212
|
+
|
|
213
|
+
## 🗺️ Roadmap
|
|
214
|
+
|
|
215
|
+
| Quarter | Milestone |
|
|
216
|
+
|---------|-----------|
|
|
217
|
+
| **Q1 2026** | ✅ Unified meta-package, 4 components integrated, PyPI published |
|
|
218
|
+
| **Q2 2026** | Cross-component integration tests, unified CLI, dashboard UI |
|
|
219
|
+
| **Q3 2026** | Helm chart for Kubernetes, managed cloud preview |
|
|
220
|
+
| **Q4 2026** | SOC2 Type II certification, enterprise support tier |
|
|
221
|
+
|
|
222
|
+
---
|
|
223
|
+
|
|
224
|
+
## 🛡️ OWASP Agentic Top 10 Coverage
|
|
225
|
+
|
|
226
|
+
The agent governance stack covers **10 of 10** risks from the [OWASP Top 10 for Agentic Applications (2026)](https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/):
|
|
227
|
+
|
|
228
|
+
| OWASP Risk | Coverage | Component |
|
|
229
|
+
|-----------|----------|-----------|
|
|
230
|
+
| Agent Goal Hijack | ✅ | Agent OS — Policy Engine |
|
|
231
|
+
| Tool Misuse | ✅ | Agent OS — Capability Sandboxing |
|
|
232
|
+
| Identity & Privilege Abuse | ✅ | AgentMesh — DID Identity |
|
|
233
|
+
| Supply Chain Vulnerabilities | ✅ | AgentMesh — AI-BOM v2.0 |
|
|
234
|
+
| Unexpected Code Execution | ✅ | Agent Runtime — Execution Rings |
|
|
235
|
+
| Memory & Context Poisoning | ✅ | Agent OS — VFS + CMVK |
|
|
236
|
+
| Insecure Inter-Agent Communication | ✅ | AgentMesh — IATP Protocol |
|
|
237
|
+
| Cascading Failures | ✅ | Agent SRE — Circuit Breakers |
|
|
238
|
+
| Human-Agent Trust Exploitation | ✅ | Agent OS — Approval Workflows |
|
|
239
|
+
| Rogue Agents | ✅ | Agent Runtime — Kill Switch |
|
|
240
|
+
|
|
241
|
+
**[→ Full OWASP compliance mapping with code examples](docs/OWASP-COMPLIANCE.md)**
|
|
242
|
+
|
|
243
|
+
---
|
|
244
|
+
|
|
245
|
+
## 📈 Traction
|
|
246
|
+
|
|
247
|
+
The ecosystem is growing — **3,000+ views, 9,400+ clones, and 1,278 unique developers** in the last 14 days alone. Traffic from Medium, Reddit, LinkedIn, Google, and even ChatGPT.
|
|
248
|
+
|
|
249
|
+
**[→ See full traction report](docs/TRACTION.md)**
|
|
250
|
+
|
|
251
|
+
---
|
|
252
|
+
|
|
253
|
+
## Contributing
|
|
254
|
+
|
|
255
|
+
We welcome contributions! See our [Contributing Guide](CONTRIBUTING.md) for details.
|
|
256
|
+
|
|
257
|
+
For component-specific contributions, see:
|
|
258
|
+
- [Agent OS](https://github.com/microsoft/agent-governance-toolkit/blob/master/CONTRIBUTING.md)
|
|
259
|
+
- [AgentMesh](https://github.com/microsoft/agent-governance-toolkit/blob/master/CONTRIBUTING.md)
|
|
260
|
+
- [Agent Runtime](https://github.com/microsoft/agent-governance-toolkit/blob/master/CONTRIBUTING.md)
|
|
261
|
+
- [Agent SRE](https://github.com/microsoft/agent-governance-toolkit/blob/master/CONTRIBUTING.md)
|
|
262
|
+
|
|
263
|
+
## License
|
|
264
|
+
|
|
265
|
+
MIT — see [LICENSE](LICENSE) for details.
|
|
266
|
+
|
|
267
|
+
---
|
|
268
|
+
|
|
269
|
+
<div align="center">
|
|
270
|
+
|
|
271
|
+
**[github.com/microsoft/agent-governance-toolkit](https://github.com/microsoft/agent-governance-toolkit)** · **[Documentation](https://github.com/microsoft/agent-governance-toolkit/tree/main/docs)** · **[GitHub](https://github.com/microsoft/agent-governance-toolkit)**
|
|
272
|
+
|
|
273
|
+
*Building the governance layer for the agentic era*
|
|
274
|
+
|
|
275
|
+
</div>
|
|
@@ -0,0 +1,64 @@
|
|
|
1
|
+
[build-system]
|
|
2
|
+
requires = ["setuptools>=68.0,<69.0", "wheel"]
|
|
3
|
+
build-backend = "setuptools.build_meta"
|
|
4
|
+
|
|
5
|
+
[project]
|
|
6
|
+
name = "agent_governance_toolkit"
|
|
7
|
+
version = "3.0.2"
|
|
8
|
+
description = "Public Preview — Unified installer and runtime policy enforcement for the Agent Governance Toolkit"
|
|
9
|
+
readme = "README.md"
|
|
10
|
+
license = {text = "MIT"}
|
|
11
|
+
requires-python = ">=3.9"
|
|
12
|
+
authors = [
|
|
13
|
+
{name = "Microsoft Corporation", email = "agentgovtoolkit@microsoft.com"},
|
|
14
|
+
]
|
|
15
|
+
maintainers = [
|
|
16
|
+
{name = "Agent Governance Toolkit Team", email = "agentgovtoolkit@microsoft.com"},
|
|
17
|
+
]
|
|
18
|
+
keywords = [
|
|
19
|
+
"ai-agents", "governance", "multi-agent", "trust", "security",
|
|
20
|
+
"compliance", "audit", "policy-enforcement", "agent-os", "agentmesh",
|
|
21
|
+
"runtime", "sre", "llm", "autonomous-agents", "enterprise-ai"
|
|
22
|
+
]
|
|
23
|
+
classifiers = [
|
|
24
|
+
"Development Status :: 4 - Beta",
|
|
25
|
+
"Intended Audience :: Developers",
|
|
26
|
+
"Programming Language :: Python :: 3",
|
|
27
|
+
"Programming Language :: Python :: 3.9",
|
|
28
|
+
"Programming Language :: Python :: 3.10",
|
|
29
|
+
"Programming Language :: Python :: 3.11",
|
|
30
|
+
"Programming Language :: Python :: 3.12",
|
|
31
|
+
"Topic :: Security",
|
|
32
|
+
"Topic :: Software Development :: Libraries :: Application Frameworks",
|
|
33
|
+
"Topic :: Scientific/Engineering :: Artificial Intelligence",
|
|
34
|
+
]
|
|
35
|
+
dependencies = [
|
|
36
|
+
"agent-os-kernel>=1.0.0,<2.0",
|
|
37
|
+
"agentmesh-platform>=1.0.0,<2.0",
|
|
38
|
+
"pydantic>=2.4.0,<3.0",
|
|
39
|
+
]
|
|
40
|
+
|
|
41
|
+
[project.optional-dependencies]
|
|
42
|
+
runtime = ["agentmesh-runtime>=2.0.0,<3.0"]
|
|
43
|
+
sre = ["agent-sre>=1.0.0,<2.0"]
|
|
44
|
+
opa = []
|
|
45
|
+
cedar = ["cedarpy>=4.0.0,<5.0"]
|
|
46
|
+
full = [
|
|
47
|
+
"agentmesh-runtime>=2.0.0,<3.0",
|
|
48
|
+
"agent-sre>=1.0.0,<2.0",
|
|
49
|
+
]
|
|
50
|
+
|
|
51
|
+
[project.urls]
|
|
52
|
+
Homepage = "https://github.com/microsoft/agent-governance-toolkit"
|
|
53
|
+
Documentation = "https://github.com/microsoft/agent-governance-toolkit/tree/main/docs"
|
|
54
|
+
Repository = "https://github.com/microsoft/agent-governance-toolkit"
|
|
55
|
+
"Bug Tracker" = "https://github.com/microsoft/agent-governance-toolkit/issues"
|
|
56
|
+
"Agent OS" = "https://github.com/microsoft/agent-governance-toolkit"
|
|
57
|
+
"AgentMesh" = "https://github.com/microsoft/agent-governance-toolkit"
|
|
58
|
+
"Agent Runtime" = "https://github.com/microsoft/agent-governance-toolkit"
|
|
59
|
+
"Agent SRE" = "https://github.com/microsoft/agent-governance-toolkit"
|
|
60
|
+
|
|
61
|
+
[project.scripts]
|
|
62
|
+
agent-governance-toolkit = "agent_compliance.cli.main:main"
|
|
63
|
+
agent-governance = "agent_compliance.cli.main:main"
|
|
64
|
+
agent-compliance = "agent_compliance.cli.main:main"
|