Install
$ agentstack add mcp-appergb-aegis-mesh ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Aegis Mesh
[](https://github.com/appergb/aegis-mesh/actions/workflows/ci.yml) [](LICENSE) [](.python-version)
Aegis Mesh is a framework-neutral, defensive multi-agent security package for detection, protection, incident response, recovery, observability, evidence governance, and risk analysis. It is designed to be embedded into different model runtimes and operational environments without granting a model implicit production authority.
The repository includes twelve role-bound agents, three shared defensive skills, closed JSON Schemas, strict configuration validation, synthetic scenarios, evidence-chain contracts, and disabled-by-default MCP integration specifications. It does not connect to real services, scan assets, send notifications, store production evidence, or ship credentials.
Why Aegis Mesh
- Portable: provider adapters are separate from the security contract.
- Governed: permissions, action classes, approval roles, and tool inventories come from validated configuration.
- Efficient: policy-authorized read-only work does not gain unnecessary human approval; irreversible external actions keep the approvals they actually require.
- Fail closed: unknown tools, assets, schemas, scopes, identities, or configuration are blocked.
- Verifiable: 23 scenario contracts, package validation, forward-result validation, and a reproducible source-bundle build are included.
Quick start
Prerequisites: Python 3.11.15 and Git.
git clone https://github.com/appergb/aegis-mesh.git
cd aegis-mesh
./setup.sh
make validate
make build
make build creates versioned source archives and SHA-256 checksums in dist/. It does not create or deploy a live agent runtime.
Repository map
- Root governance files: identity, safety, routing, and incident response.
skills/: role agents and reusable defensive workflows.schemas/: closed structural request and response contracts.configs/: governance facts and a disabled runtime example.integrations/: pinned, disabled MCP policies and upstream evidence.forensics/: custody schemas, policy, and synthetic templates.adapters/: model-provider and MCP compatibility guidance.scenarios/: synthetic evaluation contracts.scripts/andtests/: validators, release build, and regression tests.
Load order
SYSTEM_PROMPT.md
→ SECURITY_POLICY.md
→ AGENT_ROUTER.md
→ INCIDENT_RESPONSE.md when an incident is active
→ validated deployment configuration
→ selected agent system prompt
→ selected agent/shared skill
→ least-privilege tool policy
→ incident-scoped untrusted context
Later layers may narrow, but cannot expand, earlier boundaries. A deployment adapter may translate provider mechanics but cannot add permissions, targets, approvals, disclosure, or tools.
Customize and deploy
Read [Architecture](docs/architecture.md), [Customization](docs/customization.md), and [Deployment](docs/deployment.md). Copy configs/runtime.example.yaml into a deployment-owned location, keep secrets in a secret manager, and replace synthetic thresholds only with reviewed measurements from your environment.
Production remains blocked until provider identity, secrets, endpoints, owned assets, SLO/RTO/RPO, legal and vendor review, evidence storage, gateway/MCP conformance, and independent forward results are supplied. Static tests do not prove live protection or recovery performance.
Security boundary
Aegis Mesh is defensive only. It prohibits hack-back, credential theft, persistence, destructive payloads, unauthorized scans, hidden disclosure, wildcard targets, and direct upstream MCP access. See [SECURITY.md](SECURITY.md) for vulnerability reporting and [SECURITYPOLICY.md](SECURITYPOLICY.md) for the runtime policy.
Contributing
Contributions are welcome under the [contribution guide](CONTRIBUTING.md) and [code of conduct](CODEOFCONDUCT.md). By contributing, you agree that your contribution is licensed under Apache-2.0.
Using with Codex
Open the cloned repository in Codex and ask it to follow [AGENTS.md](AGENTS.md). That guide contains the exact setup, focused-test, validation, build, architecture, and configuration entry points. Codex operates on this source package only; it does not gain production credentials, network targets, or deployment authority from the repository.
License
Copyright 2026 Aegis Mesh Contributors. Licensed under the [Apache License 2.0](LICENSE).
Source & license
This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: appergb
- Source: appergb/aegis-mesh
- License: Apache-2.0
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.