Skip to content

Latest commit

 

History

History
89 lines (71 loc) · 3.99 KB

File metadata and controls

89 lines (71 loc) · 3.99 KB

Prahari OpenShell — Execution Checklist

What's ready (delivered in this archive)

Code (13 tests passing, 0.012ms evaluation latency)

  • GRDL schema — universal, jurisdiction-agnostic governance rule definitions
  • GRDL loader — YAML parser with validation
  • GRDL compiler — three-pass: static → OpenShell YAML, dynamic → network policies, runtime → CFAIS sidecar
  • CFAIS engine — deterministic rule evaluator with graceful degradation
  • FastAPI sidecar — HTTP server with /evaluate, /healthz, /metrics, /rules
  • CLI — compile, validate, evaluate, serve commands
  • OpenShell sandbox — Dockerfile, policy.yaml, README.md
  • Test suite — 13 tests across 3 governance templates

Governance templates

  • Enterprise Agent Governance — budget caps, privilege escalation, cascading depth, PII, human-in-the-loop
  • DAO / Cooperative Governance — treasury limits, quorum, voting weight caps
  • AI Safety — tool allowlists, output validation, rate limiting

Licensing

  • LICENSE-APACHE — core engine
  • LICENSE-BSL — enterprise modules
  • docs/pricing.md — FRAND terms

Outreach materials

  • docs/nvidia-inception-application.md — ready to paste into Inception portal
  • docs/openshell-pr-description.md — ready to paste into GitHub PR
  • docs/openshell-discussion-post.md — ready to paste into GitHub Discussions
  • LinkedIn message to Kari Briski (2 variants drafted)
  • Email to Naska Yankova (drafted)

Execution sequence

Week 1: Foundation

# Task Time Status
1 Extract archive to VPS, run python tests/test_pipeline.py 30 min
2 Create GitHub repo prahari-ai/prahari-openshell 15 min
3 Push code to GitHub 15 min
4 Apply for NVIDIA Inception at nvidia.com/en-us/startups/ 45 min
5 Send email to inceptionprogram@nvidia.com referencing application 15 min

Week 2: Integration testing

# Task Time Status
6 Install OpenShell CLI on VPS: curl -fsSL nvidia.com/openshell.sh | bash 30 min
7 Build Docker image: docker build -t prahari-cfais -f core/sandbox/Dockerfile . 15 min
8 Create OpenShell sandbox manually and test sidecar integration 2 hrs
9 Record 2-minute demo video showing: compile → deploy → deny → allow 1 hr

Week 3: Publish

# Task Time Status
10 Fork NVIDIA/OpenShell, add sandbox to sandboxes/prahari-constitutional/ 30 min
11 Submit PR using docs/openshell-pr-description.md 30 min
12 Post Discussion using docs/openshell-discussion-post.md 15 min
13 Update prahari.ai website with OpenShell integration page 2 hrs

Week 4: Outreach

# Task Time Status
14 LinkedIn message to Kari Briski (AFTER PR is submitted) 15 min
15 Email Naska Yankova with Patio/NVIDIA angle 15 min
16 Weave NemoClaw integration into DAWO26 paper (due April 15) 4 hrs
17 Post on Moltbook agentic-governance submolt 30 min

Key files reference

File Purpose
cli.py Main entry point — compile, validate, evaluate, serve
core/grdl_compiler/schema.py Type definitions (GovernanceLaw, Rule, Ruleset)
core/grdl_compiler/compiler.py GRDL → OpenShell YAML + CFAIS config
core/cfais_engine/engine.py Runtime evaluator (12μs per eval)
core/sidecar.py FastAPI HTTP server for OpenShell integration
core/sandbox/ Dockerfile + policy.yaml for OpenShell community catalog
examples/templates/ Three governance templates ready to customize
tests/test_pipeline.py 13-test suite covering all three templates
docs/nvidia-inception-application.md Inception portal application text
docs/openshell-pr-description.md GitHub PR description
docs/openshell-discussion-post.md GitHub Discussion post