codesift-mcp@0.21.0
MCP server for code intelligence — 146 tools for symbol search, call graph, semantic search, route tracing, community detection, LSP bridge, secret detection, conversation search, and Hono framework intelligence
AI Security Review
scanned 9d ago · by lpm-firewall-aiThe automatic postinstall performs first-party agent integration setup and modifies agent configuration, which warrants a warning under the agent extension lifecycle policy. The reviewed runtime account-action candidates are indexing and cache operations, not automatic external account changes.
Decision evidence
public snapshot- package.json runs postinstall.mjs, which imports dist/postinstall.js when present; that script launches the CLI with setup all.
- The setup command defaults hooks and rules on, and setup all processes the supported platforms. This installs the package’s own agent integrations automatically during npm installation.
- Codex setup writes its MCP configuration under the user’s .codex directory. The inspected agent setup code configures the package’s own MCP server.
- The runtime locations flagged for account actions resolve to file watching, indexing, index freshness, and cached search operations; they do not show automatic external account changes.
Source & flagged code
22 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgPackage source references dynamic require/import behavior.
dist/tools/pg-introspection.jsView on unpkg · L13Findings
Affected versions and remediation
This report applies to codesift-mcp@0.21.0.
See all versions of codesift-mcp for other recorded verdicts.
- Review the evidence and your use of codesift-mcp@0.21.0 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.