v9.1.0 — Hemlock Score, intelligence loop, continuous security

Hemlock v9.1.0

First GitHub release after v4.5.0, bundling continuous-security work from v5.0 through v9.1 on master.

Highlights (v8.9–v9.1)

• Hemlock Score — pipeline-native 0–100 security metric with CI badge (hemlock score-pipeline, hemlock gate)
• Intelligence loop — auto replay capture, threat intel advisories, optional auto red-team after orchestrated scans
• Dashboard — Hemlock Score card, score trend chart, new attack techniques panel

Since v4.5.0 (summary)

• v5.x — SDK, red-team campaigns, genetic fuzzer, threat intel, OTel, marketplace
• v6.x — distributed scanner, fingerprinting, auto red-team, policy-as-code, benchmark registry, cloud prep
• v7.x — security baseline, finding lifecycle, executive reports, model inventory, attack replay, provider comparison, playbooks, scan orchestrator, risk scoring, framework adapters
• v8.x — operational CLI (orchestrate, risk-score, executive-report), operational dashboard, leaderboard, policy+risk gate, LLM judge, continuous security CI workflow, trend charts, org overview

Install

pip install "hemlock-rag @ git+https://github.com/M4NT/hemlock@v9.1.0"

Tests

1550+ pytest scenarios (mock mode, no API keys required for core suite).

Full per-version notes: CHANGELOG.md.