complex-md is before 1.0. Security fixes go to the latest minor release only, as a patch release. npm view complex-md version prints the latest release.

Do not report a vulnerability in a public issue, discussion or pull request.

Report it in private through GitHub: https://github.com/KletoWorks/complex-md/security/advisories/new. On the repository page, this is Security, then "Report a vulnerability". Only the reporter and the people who administer the repository's security can read the report.

Put these facts in the report:

• the version, from npx complex-md --version
• the operating system and the agent harness, for example Claude Code or Cursor
• the steps that reproduce the problem, or a small repository that does
• what an attacker can do with it

You get a first answer within 7 days. When a fix ships, the advisory is published with the fixed version. The advisory credits you, unless you ask it not to.

• The hook commands that wire writes into .claude/settings.json and .cursor/hooks.json. The agent harness runs them on each edit, so a way to make them run other code is in scope.
• A file that wire or generate writes outside the repository root.
• A repository whose file names or history make the CLI write a COMPLEX.md or a brief that injects instructions into an agent.
• The model call. With a provider key, generate sends the signals and the hotspot excerpts to that provider. Before the call, it replaces each secret-shaped value in the request with [redacted:<rule>] (cli/src/redact.js, with the rules in cli/rules/exposure.json). A credential file that reaches the request is in scope, and so is a key in a known token format that reaches it.
• complex-md exposure. A secret value in its report, its SARIF file, its baseline or its terminal output is in scope. So is a rule file in a cloned repository (.complex-md/exposure.json) that makes the scan run code, stall, or fetch from a remote.
• The MCP server, complex-md mcp. wire registers it only with --mcp.
• The release workflow and the npm package provenance.

• A vulnerability in a model provider or an agent harness. Report it to that vendor.
• A known vulnerability in a dependency with no complex-md specific path to exploit it. Report it upstream.