README.md

metadata

title: README
emoji: 👁
colorFrom: red
colorTo: green
sdk: static
pinned: false
short_description: Clean feedback and promotion gates for AI agents.

Verifiable Labs

Verifiable Labs builds clean feedback and promotion gates for increasingly general AI agents.

We develop infrastructure for evaluating agent improvements, checking whether those improvements transfer to unseen/OOD/adversarial situations, and producing synthetic/redacted evidence artifacts for promotion decisions.

Public resources

• GitHub: https://github.com/verifiablelabs
• PyPI: https://pypi.org/project/vlabs-sdk/0.0.2/
• Install: pip install "vlabs-sdk==0.0.2"
• Dataset: https://huggingface.co/datasets/verifiablelabs/vlabs-clean-gate-evidence
• W&B: https://wandb.ai/verifiable-labs/clean-generalization-gate

Evidence policy

Public evidence is synthetic/redacted and is not a training dataset.

It does not include customer data, hidden evals, gold answers, raw traces, private traps, private engine internals, secrets, or provider keys.

Formal scope

Selected mathematical properties behind the contamination-resistant promotion gate are machine-verified in Lean 4. The implementation is property-tested against the formal specification.