πŸ“ˆ GH Trending Explorer 🎲 Surprise me

← all repos

securetron-gh / aifw

git clone https://github.com/securetron-gh/aifw.git Alberta Β· 2d ago

AI Agent Identity and Access Management and Inference Firewall

+0
stars, 24h Β· #29,934
+~175
stars, 7 days Β· #1,244
+~750
stars, 30 days Β· #1,105
272
β˜… total Β· +175% this week
3
forks Β· 0 open issues

Star history (as recorded here)

2026-10-20 Β· 197 β˜… 2026-10-23 Β· 272 β˜…
LanguagePython
Licenseβ€”
Created2026-08-22 (2mo ago)
Last push2d ago
Watchers2
OwnerOrganization

GitHub topics

Links

README excerpt
# AI-FW
#### The AI Firewall & Governance Gateway
**Every AI agent: identified, authorized, inspected, governed.**
Documentation Β· Website Β· Get started Β· Enterprise
AI-FW is the **AI firewall and governance gateway** between your AI agents and the models they call. Every prompt and response is inspected, governed, and routed, with provable agent identity and regulator-ready audit, all in one process that runs in your own environment.
- [x] **Fail-closed by design**: offline guardrails never delegate to third parties, no traffic while scanning is unhealthy
- [x] **Agents are identities**: A2A protocol, agent registry, mTLS, Kerberos, OIDC SSO, SCIM provisioning
- [x] **Cost savings without safety holes**: semantic-gated prompt compression and inspection-gated caching
- [x] **Audit you can hand to a regulator**: metadata-only transaction log with encrypted export to your SIEM
- [x] **Surfaces others cannot reach**: Claude inference hooks, Cursor Agent Hooks, MCP tool governance, M365 Copilot admin with agent sync
- [x] **Opt-in endpoint gateway**: images, audio, files, fine-tuning, and batches under the same inspection pipeline
## πŸ›‘οΈ Features
Every feature is documented on aifw.io/docs and in this repository's `docs/` folder. Product updates and practical guidance are published on AI-FW Insights, with an RSS feed.
Legend: βœ… native Β· 🧩 via partner or plugin Β· ❌ not supported
| Category | Feature | Doc | Native |
|---|---|---|---|
| Security & guardrails | Prompt & response guardrails | Read | βœ… |
| Security & guardrails | Semantic intent analysis | Read | βœ… |
| Security & guardrails | Fail-open options (opt-in) | Read | βœ… |
| Security & guardrails | Claude inference hooks | Read | βœ… |
| Identity & agents | Identity & access (IDAM) | Read | βœ… |
| Identity & agents | Admin-issued API keys | Read | βœ… |
| Identity & agents | A2A agent protocol | Read | βœ… |
| Identity & agents | Chain of Command | Read | βœ… |
| Identity & agents | Agent self-enrollment | Read | βœ… |
| Cost | Completion cache | Read | βœ… |
| Cost | Prompt compression | Read | βœ… |
| Observability & audit | Observability & dashboard | Read | βœ… |
| Observability & audit | Risk profiles & auto-block | Read | βœ… |
| Observability & audit | Audit logs & export | Read | βœ… |
| Governance & compliance | Compliance module administration | Read | βœ… |
| Governance & compliance | Skills & Tools governance | Read | βœ… |
| Governance & compliance | IDE Artifacts | Read | βœ… |
| Governance & compliance | Configuration backup and restore | Read | βœ… |
| Integrations | M365 Copilot admin | Read | βœ… |
| Integrations | Claude Code, Cursor & MCP | Read | βœ… |
| Integrations | Cursor Agent Hooks and shims | Read | βœ… |
| Integrations | securetron-aifw hook package | Use the package | βœ… |
| Integrations | Endpoint gateway | Read | βœ… |
| Core & reference | Model routing & registry | Read | βœ… |
| Core & reference | Reliability | Read | βœ… |
| Core & reference | Advanced parameters | Read | βœ… |
| Core & reference | Prompt