@papi-ai/server@0.7.113
PAPI MCP server — AI-powered sprint planning, build execution, and strategy review for software projects
AI Security Review
scanned 4d ago · by lpm-firewall-aiThe reviewed credential flow sends a PAPI-specific key only to an endpoint explicitly configured for prompt retrieval, and otherwise uses bundled prompts. The flagged settings changes are part of explicit setup or status-line installation; no malicious install-time or automatic credential redirection was found.
Decision evidence
public snapshot- The manifest exposes dist/index.js as the server entry and declares no install lifecycle hook.
- Prompt fetching falls back to bundled prompts unless PAPI_PROMPT_ENDPOINT is configured; any PAPI_PROMPT_API_KEY is sent to that configured endpoint.
- The CLI handles setup as an explicit subcommand; project scaffold and Claude settings changes occur as part of setup.
- The separate status-line installer writes Claude settings only when run with --install; otherwise it reads PAPI-specific configuration and fetches from its configured endpoint.
- The server can run git and GitHub CLI commands to detect the user's email for project ownership.
- Setup and status-line installation can write AI-client settings and hooks, though these actions are tied to their respective user-invoked setup/install flows.
Source & flagged code
7 flaggedPackage source references child process execution.
dist/backfill-cycle-metrics.jsView on unpkg · L271This package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
dist/index.jsView on unpkgFindings
Affected versions and remediation
This report applies to @papi-ai/server@0.7.113.
See all versions of @papi-ai/server for other recorded verdicts.
- Review the evidence and your use of @papi-ai/server@0.7.113 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.