Roadmap
Provael is CPU-first and model-agnostic. Shipped vs. planned, honestly marked.
Shipped
- Attacks: 4 families (instruction / visual / injection / action) + a
nonebaseline, mapped to the Embodied AI Security Top 10. - Policies:
stub(CPU);smolvla,pi0,pi05,pi0fast,groot(LeRobot);openvla(HF transformers). Bring-your-own via thePolicyAdapterABC. - Suites:
stub(scalar, CPU),reach(spatial, CPU),libero+metaworld(gated). - Evidence: SARIF, compliance crosswalk, pre-deployment scorecard, OSCAL export, AVID export.
- Reproductions: FreezeVLA, OpenVLA-patch, BadVLA, RoboPAIR.
- Integrations: promptfoo provider; garak/PyRIT reference plugins; multi-CI (GitHub/GitLab/ Azure) SARIF; pre-commit; MLflow/W&B logging; HF eval-results; Modal GPU-CI; Docker/devcontainer; supply-chain (model-signing + ML-BOM).
- Defense demo: action-stream firewall + ROS 2 guard node (sim/reference).
Planned (contributions welcome)
- Suites: RoboCasa, CALVIN, SimplerEnv, and the AI2 vla-evaluation-harness bridge (one adapter → ~18 benchmarks). See examples/suites.
- Public leaderboard with open submission; docs site versioning.
- Standards: MITRE ATLAS case study, OWASP Agentic embodied annex, OECD.AI listing (drafts in docs/standards).
- Stronger attacks: gradient/search-optimised variants beyond the templated screen.
Note
"Planned" means not yet shipped — we don't ship fabricated capability. Each lands behind tests and the same honesty discipline as the rest of the project.