Brotli Checker finds copies of the Brotli decoder that may need an update for CVE-2026-111497. Install it as a Claude Code or Codex skill, then ask it to check your project.
The skill scans source, dependencies, app packages, and unpacked container filesystems. It checks source fingerprints and supported ELF binary code, investigates which copies the app uses, and explains what to update or review. Each copy keeps its own result.
It also recognizes decoder tables in stripped or LTO binaries. Table recognition finds copies to review; it does not establish their patch status.
For source review, Trailmark maps calls to the decoder and fetch APIs, helping the assistant find callers to inspect. The scanner finds suspect parts; Trailmark helps show how the code connects. The assistant checks whether those connections apply to the running app.
Ask Claude Code or Codex:
Install the brotli-checker skill from https://github.com/trailofbits/brotli-checker/tree/main/skills/brotli-checker
For manual installation, copy the entire skills/brotli-checker folder into
~/.claude/skills/ (Claude Code) or
~/.codex/skills/ (Codex). Start a new session, open the project to check, and ask:
Use brotli-checker to check this project and tell me whether I need to update anything.
You can also name a folder or app package. With only a website address, the skill explains what is missing and prepares a request for your hosting provider. HTTP headers cannot establish the decoder's patch status.
Requires Python 3.12–3.15 and uv for Trailmark source discovery. The standalone scanner uses only the standard library.
To collect local evidence without an agent:
git clone https://github.com/trailofbits/brotli-checker.git
cd brotli-checker
python3 skills/brotli-checker/scripts/brotli_inventory.py /path/to/applicationThe default output explains findings and next steps. Add --json for per-copy
evidence, coverage gaps, and a plain-language user_summary. See the
scanner guide for flags, exit codes, Trailmark source discovery,
binary inventory, and disassembly review.
Affected code calls for an update or vendor-backport review. Version and dependency findings need confirmation against the app's actual build. A fixed copy does not clear other bundled copies, and an empty scan does not establish that an app is safe.
The scanner reads files without executing application code or sending network probes. Packaged compiled references cover x86, x86-64, ARM/Thumb, AArch64, RISC-V64, PowerPC64 little-endian, s390x, MIPS32/64 in both byte orders, and LoongArch64 ELF builds. Coverage depends on compiler and build flags; see the build matrix. Unsupported formats, skipped files, and resource limits produce coverage gaps. Finding affected code does not establish that a running app uses it or that an attacker can reach it.
Rules use an advisory snapshot dated 2026-10-23. Check current vendor guidance for updates and backports.
- Scanner guide: commands, output, limits, and binary review.
- Ecosystem guide: native bindings, runtime decoders, independent ports, and WASM integrations.
- Advisory: fixed versions, fingerprints, and provenance.
- Reachability: investigating Node, Electron, Java, fonts, TLS, and compiled code.
- Contributing: development setup and checks.
- Changelog: release history.
- Security policy: private vulnerability reporting.
MIT. Compiled Brotli references retain their upstream notices.