The package is not shown to be malicious, but its explicit agent-configuration capability merits a behavior-specific warning under the agent integration policy. There is no evidence of automatic lifecycle execution, credential theft, or remote payload delivery.
Static reason
One or more suspicious static signals were detected.
Trigger
A user runs the diffprism setup command, optionally with the global flag.
Impact
This expands an agent's permitted DiffPrism tool capability after explicit user invocation; no confirmed covert exfiltration or install-time mutation was found.
Mechanism
The command updates Claude settings with the package's MCP tool permissions and writes package-owned local server state.
Evidence
.claude/settings.json.claude/skills/review/SKILL.md.mcp.json.diffprism/server.json