agentassert-abc
v0.3.0
Published
Formal behavioral specification and runtime enforcement for AI agents. Agent Behavioral Contracts (ABC).
Maintainers
Readme
AgentAssert
Formal Behavioral Contracts for AI Agents
The TypeScript SDK is coming Q3 2026. The Python package is available now.
Python Package (Available Now)
pip install agentassert-abc[yaml,math]AgentAssert implements Agent Behavioral Contracts (ABC) — the only framework combining all 6 pillars of rigorous agent governance:
- ContractSpec DSL — YAML-based behavioral specification with 14 operators
- Hard/Soft Constraints — Formal separation with graduated enforcement
- Drift Detection — Jensen-Shannon Divergence for behavioral analysis
- (p, delta, k)-Satisfaction — Probabilistic compliance guarantees
- Compositional Safety Proofs — Formal bounds for multi-agent pipelines
- Mathematical Stability — Ornstein-Uhlenbeck dynamics with Lyapunov proof
Links
- Paper: arXiv:2602.22302
- Python Package: PyPI: agentassert-abc
- GitHub: qualixar/agentassert-abc
- Documentation: agentassert.com
License
AGPL-3.0
Part of Qualixar — AI Agent Reliability Engineering A research initiative by Varun Pratap Bhardwaj
