Security fixes are applied to the latest minor release line only. If a vulnerability is found in an older version, please upgrade to the latest 2.0.x release before reporting.
| Version | Supported |
|---|---|
| 2.0.x | ✅ |
| < 2.0 | ❌ |
Please do not report security vulnerabilities through public GitHub issues.
Report vulnerabilities privately through GitHub's Security Advisories flow. This keeps the details confidential until a fix is published.
A good report contains:
- A description of the vulnerability and its impact
- Affected versions (including the exact
tech-debt-mcpversion fromnpm listorpackage.json) - Steps to reproduce, ideally with a minimal example
- Any known mitigations or workarounds
Tech Debt MCP is a small open-source project maintained on a best-effort basis, but security reports are triaged with priority:
- Acknowledgement: within 7 days of the report
- Initial assessment: within 14 days (severity, affected versions, whether it reproduces)
- Fix timeline: depends on severity — critical issues are patched as quickly as possible, typically within 30 days
- Disclosure: coordinated with the reporter once a fix is released; credit is given in the release notes unless anonymity is requested
In scope:
- MCP tool handlers and resource handlers (
src/server/) - Input validation and path handling (
src/server/inputParser.ts,src/server/argValidation.ts) - User-supplied regex compilation in custom rules (
src/core/customRulesEngine.ts) - Dependency parsers that read manifest files (
src/analyzers/dependencies/) - The npm-published package itself (supply-chain issues, malicious publish, etc.)
Out of scope:
- Findings that require the attacker to already have full filesystem or process access
- Denial-of-service via extremely large input files beyond the documented caps (see
MAX_FILE_SIZE_BYTES,MAX_PATTERN_LENGTH,MAX_CODE_LENGTHinsrc/core/customRulesEngine.ts) - Issues in third-party dependencies — please report those upstream and let us know so we can bump the dependency
This repository runs the following automated scans:
- CodeQL — static analysis on every push to
develop/main, every pull request todevelop/main, and weekly on Mondays (see.github/workflows/codeql.yml) - Secret scanning — GitHub-managed, enabled for all pushes
- Dependabot alerts — GitHub-managed, monitors runtime and dev dependencies for known CVEs
Findings from these scans are reviewed by the maintainers. If you notice a scan result that looks like a genuine vulnerability, you can still file a private report via the Security Advisories flow — we'd rather have duplicates than miss something.