@cachly-dev/mcp-server@0.10.201
Solve it once. From then on, your AI knows. Experience for coding agents, EU-hosted: an MCP server for Claude Code, Cursor, Copilot & Windsurf. Learns from every bug fix, commit and CI run.
AI Security Review
scanned 8d ago · by lpm-firewall-aiThe source shows a guarded, first-party configuration change after device authorization, not an unconsented install-hook mutation or unrelated credential transfer. Classify as suspicious under the agent extension lifecycle risk policy.
Decision evidence
public snapshot- The postinstall hook only displays setup instructions; the manifest points it to a script, and that script recommends running the setup command.
- An unauthenticated MCP tool call enters the sign-in flow, which presents a browser link and code for user authorization.
- After receiving an access token, the runtime stores it in the global Claude Code MCP configuration for Cachly.
- The package also updates Cachly configuration in existing editor entries; this is first-party integration setup, but it mutates editor configuration after sign-in.
- The configuration write follows device authentication, and the install hook itself only prints instructions.
- The credential is used for the Cachly service, and the inspected source does not establish an unrelated recipient or a broader agent-control hijack.
Source & flagged code
18 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 unpkgFindings
Affected versions and remediation
This report applies to @cachly-dev/mcp-server@0.10.201.
See all versions of @cachly-dev/mcp-server for other recorded verdicts.
- Review the evidence and your use of @cachly-dev/mcp-server@0.10.201 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.