This release includes 2 security fixes for security teams reviewing exposed deployments.
Topics
+5 more
Affected surfaces
Summary
AI summaryPublish gate honesty enforced, tokens.json taint closed, reproducible receipts added, output contracts tightened, and dogfood counting excludes test records.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Bugfix | Medium |
Enforces structural output contracts in the second‑agent benchmark (sentence limits, required shapes, fence‑only). Enforces structural output contracts in the second‑agent benchmark (sentence limits, required shapes, fence‑only). Source: llm_adapter@2026-07-18 Confidence: high |
— |
| Bugfix | Medium |
Excludes test records from `record_kind` counting; real count reported as 4. Excludes test records from `record_kind` counting; real count reported as 4. Source: llm_adapter@2026-07-18 Confidence: high |
— |
| Bugfix | Medium |
Fixes publish gate honesty by setting `publish_ready: false` when color governance is unresolved. Fixes publish gate honesty by setting `publish_ready: false` when color governance is unresolved. Source: llm_adapter@2026-07-18 Confidence: low |
— |
| Bugfix | Medium |
Sanitizes `tokens.json` keys and `$description` to prevent taint and collisions. Sanitizes `tokens.json` keys and `$description` to prevent taint and collisions. Source: llm_adapter@2026-07-18 Confidence: low |
— |
| Bugfix | Medium |
Records tree state, timestamp, and commit filenames for reproducible receipts. Records tree state, timestamp, and commit filenames for reproducible receipts. Source: llm_adapter@2026-07-18 Confidence: low |
— |
| Bugfix | Low |
Prevents strongest publication approval from blessing an unconfirmed primary when color governance is unresolved. Prevents strongest publication approval from blessing an unconfirmed primary when color governance is unresolved. Source: granite4.1:30b@2026-07-18-audit Confidence: low |
— |
Full changelog
Truth-and-trust patch from adversarial review of 0.13.4. All five findings fixed:
- Publish gate honesty:
pass_with_advisories+publish_ready: falsewhile color governance is unresolved — the strongest publication approval can no longer bless an unconfirmed primary - tokens.json taint closed + collision-safe keys: central
colorTokenKey()(sanitized slug + hex suffix) for runtime and DTCG unknown-role colors; sanitized$description - Reproducible receipts: tree state recorded, timestamp+commit filenames
- Structural output contracts in the second-agent benchmark (sentence limits, required shapes, fence-only)
- Honest dogfood counting (
record_kindexcludes test records; real count: 4)
Full changelog: https://github.com/Brandcode-Studio/brandsystem-mcp/blob/main/CHANGELOG.md
Security Fixes
- `pass_with_advisories` can no longer bless an unconfirmed primary when color governance is unresolved — fixes publish gate honesty issue.
- tokens.json taint closed; collision‑safe keys generated by centralized `colorTokenKey()` with sanitized slug and hex suffix; `$description` sanitization applied.
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]