Back to browse
GitHub Repository

DevOps MCP server, Flight recorder for AI infrastructure agents. The prescribe/report protocol captures intent before execution and outcome after — in a signed, tamper-evident evidence chain. Detects behavioral patterns. Computes reliability scorecards.

13 starsGo

Evidra – a fail-closed MCP guardrail for AI infrastructure ops

by vitass·Mar 3, 2026·1 point·1 comment

AI Analysis

●●●BangerSolve My ProblemWizardry

Fail-closed guardrail for AI agent infrastructure access—kills unsafe ops before they ship.

Strengths
  • Fail-closed default (deny unless explicitly safe) is the correct stance for production access
  • SHA-256 hash-chained evidence log makes tampering detectable and auditable
  • OPA policy engine sidesteps natural language brittleness of pure LLM safety measures
Weaknesses
  • Early project (6 stars, 0 issues public)—real-world deployment feedback lacking
  • Requires structured tool invocation; won't protect against shell injection or unstructured agent output
Target Audience

DevOps engineers, SRE teams, organizations giving AI agents production cluster access

Similar To

OPA (Open Policy Agent) · Kyverno · Kubewarden

Similar Projects

Security●●●Banger

VellaVeto — blocks unsafe MCP tool calls by default

Fail-closed MCP gateway with formal verification and MCPSEC benchmark suite.

Big BrainWizardryZero to One
paolovella
212mo ago
Infrastructure●●Solid

Make AI and automation pipelines fail-closed

Deterministic offline verification of AI pipeline outputs with Merkle hashing—novel framing, early stage.

Big BrainZero to One
oneinx
112mo ago