Runtime verification framework for software systems
Turn real software repos into agent-ready engineering environments with entrypoints, validation, evidence, and delivery records.
AgentHook: the open runtime evidence standard for AI agents. Apache 2.0. Stewarded by Agentic Thinking Limited (UK).
Runtime evidence layer for agentic AI — signed audit chain, deterministic gate verdicts with record-bound audit replay. pip install phionyx-core.
Primitives for recording Letta memory mutations as hash-chained runtime-evidence envelopes; optional signing; automatic instrumentation not yet included.
Portable Agent Skills for using Allure Report as the evidence layer for test work
Local-first AI video workflow workbench for planning, evaluating, iterating, and exporting generation-ready video production packages.
Runtime-governance conformance scenarios: reviewer-runnable, first-party measurements with published method
Bounded-authority household AI for families. AI proposes; the responsible adult executes. Browser demo + policy gates.
Building Phionyx: runtime governance and evidence infrastructure for agentic AI — deterministic control paths, measurement discipline, decision evidence, auditable execution boundaries. PyPI · Zenodo · ORCID 0009-0002-3718-4010.
Inspect AI bridge — Phionyx runtime evidence exported into Inspect eval logs. Offline-inspectable evaluation records.
LangChain + LangGraph adapters for Phionyx runtime evidence — chain / tool / LLM events and supervisor handoffs recorded as hash-chained envelopes, optionally Ed25519-signed.
Phionyx evaluation tooling — LLM-as-judge primitive + Langfuse/LangSmith cross-runtime importers. Caller supplies the LLM client. AGPL-3.0.
OpenAI Agents SDK tracing bridge for Phionyx runtime evidence — Traces and Spans recorded as hash-chained envelopes, optionally Ed25519-signed.
MCP trust boundary — descriptor hash pinning and hash-chained evidence envelopes (optional signing) over third-party MCP tool calls.
Open, vendor-neutral runtime evidence format for AI decisions, control delivery, execution, and observed effects. Signed, hash-linked, offline-checkable. Experimental.
Evidence-grade compliance report drafts from Phionyx runtime evidence — maps audit chain envelopes against EU AI Act Article 13, NIST AI RMF, ISO/IEC 42001, and OWASP Agentic AI Threats
Operational AI risk and evidence lab for RAG, agentic systems, observability, workflow integrity, and controllability.
To associate your repository with the runtime-evidence topic, visit your repo's landing page and select "manage topics."