tafreeman

Step-by-step guide for adding a new LLM reasoning pattern to ExecutionKit

Cut a new ExecutionKit release — version bump, changelog, tag, GitHub release, docs, PyPI

Run the full local validation gate (lint + format + typecheck + tests) before pushing

Run agentic-evalkit's offline verification matrix and release gates in order, reporting pass/fail per gate. Use before opening a PR, before tagging a release, or when asked to "run the gates", "verify the release", or "check release readiness".

Draft a new agentic-evalkit ADR that passes tests/contract/test_adrs.py on the first run — correct filename, seven-heading template in canonical order, Accepted status, and no phrases contradicting standing decisions. Use when asked to write, add, or record an ADR or architecture decision.