This release adds 3 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Topics
+5 more
Summary
AI summaryExtraction fixes improve color handling and corpus precision, with benchmark validation.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Dependency | Low |
Upgraded vitest to version 4 Upgraded vitest to version 4 Source: llm_adapter@2026-07-18 Confidence: high |
— |
| Dependency | Low |
Upgraded actions/checkout and setup-node to version v7 with SHA re-pins Upgraded actions/checkout and setup-node to version v7 with SHA re-pins Source: llm_adapter@2026-07-18 Confidence: high |
— |
| Dependency | Low |
Upgraded puppeteer-core to version 25 Upgraded puppeteer-core to version 25 Source: llm_adapter@2026-07-18 Confidence: high |
— |
| Dependency | Low |
Upgraded pdfjs-dist to version 6 Upgraded pdfjs-dist to version 6 Source: llm_adapter@2026-07-18 Confidence: high |
— |
| Dependency | Low |
Upgraded TypeScript to version 7 with clean typechecks; migration parked for later PR Upgraded TypeScript to version 7 with clean typechecks; migration parked for later PR Source: llm_adapter@2026-07-18 Confidence: high |
— |
| Bugfix | Medium |
Dark chromatic brand colors can now become primary candidates (HSL-based) Dark chromatic brand colors can now become primary candidates (HSL-based) Source: llm_adapter@2026-07-18 Confidence: high |
— |
| Bugfix | Medium |
Corpus precision threshold for generic web-safe fallbacks restored to 0.95 Corpus precision threshold for generic web-safe fallbacks restored to 0.95 Source: llm_adapter@2026-07-18 Confidence: high |
— |
| Bugfix | Medium |
Instruction-shaped extracted color names replaced with generated names Instruction-shaped extracted color names replaced with generated names Source: llm_adapter@2026-07-18 Confidence: high |
— |
Full changelog
Task-completion evidence release.
First live benchmark numbers
- Second-agent job completion: 5/5 — a fresh model given only brand_context output produced content the real brand_check + compliance gate accepts (0 flags, ~144 output / ~325 context tokens per artifact)
- Negative routing: 0% false-positive invocation (0/7) on vocabulary-baited prompts
- Published with caveats in eval/RESULTS.md
Extraction fixes (#35, corpus expectations flipped, no thresholds lowered)
- Dark chromatic brand colors (navy/forest/burgundy) can now become primary candidates (HSL-based chromaticity)
- Generic web-safe fallbacks (Arial et al.) no longer leak from font stacks; corpus precision threshold restored to 0.95
- Instruction-shaped extracted color names replaced with generated names
Dependencies — individually triaged
- vitest 4, actions/checkout + setup-node v7 (SHA re-pins), puppeteer-core 25, pdfjs-dist 6
- TypeScript 7 typechecks the project clean; parked for a standalone migration PR (#18)
Full changelog: https://github.com/Brandcode-Studio/brandsystem-mcp/blob/main/CHANGELOG.md
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About Brand-System/brandsystem-mcp
Make your brand machine-readable. Extract brand identity (colors, fonts, logo, voice, visual rules) from any website via static CSS + rendered-page extraction, compile into DTCG tokens, brand runtime contracts, and interaction policies. 34 tools across 4 progressive sessions. Subscribable `brand://runtime` and `brand://policy` MCP resources. Content compliance scoring (0-100), pass/fail gate, and HTML/CSS preflight. Brandcode Studio connector for hosted brand sync.
Related context
Beta — feedback welcome: [email protected]