Integrated agent workspace guardrails and verifiable software-delivery pipeline for autonomous coding workflows.
README
Safe Agent Engineering Suite (Bundle A)
Complete in-process filesystem guardrails, command execution policy, deterministic packaging, and cryptographic review receipts for autonomous AI agents.
What Is This Bundle?
The Safe Agent Engineering Suite unites two critical infrastructure products extracted from MIMARIO into a unified agent safety workflow:
1. Agent Workspace Guard SDK (P1): In-process runtime guardrails restricting AI agents to authorized workspace boundaries with 120 adversarial red-team tests (blocking path traversal, multi-dot evasion, symlink breakouts, dangerous commands, and shell chaining).
2. Verifiable Delivery Packager (P5): Deterministic packaging with canonical JSON manifests, executable bit auditing (mode & 0o111 === 0), and cryptographically signed delivery receipts.
Pricing
- Standalone P1: $149
- Standalone P5: $79
- Bundle Founding Price: $199 (Instant $29 saving)
Quickstart
# 1. Run integrated test
node --test tests/integrated.test.mjs
# 2. Run end-to-end demo
node examples/safe-agent-pipeline.mjs
16 files
file contents open where the seller allows a preview; the rest is what you are buying
examples/1
p1-workspace-guard/4
p5-delivery-packager/3
tests/1
- .gitignore30 B
- LICENSE.md2 KB
- PREVIEW.md1 KB
- README.md1 KB
- commercial-manifest.json385 B
- package.json316 B
- release-manifest.json2 KB