perplexity-ai
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFECREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill identifies a local sensitive path at
~/.clawlink/credentials.jsonfor storing authentication keys. While this is the intended configuration for the integration tool, referencing specific credential file paths increases the potential for targeted local data exposure. - [EXTERNAL_DOWNLOADS]: The skill utilizes
npx @useclawlink/clito fetch and execute the ClawLink CLI from the NPM registry. This involves acquiring and running code from an external well-known service. - [COMMAND_EXECUTION]: The skill performs various operations (login, connect, run actions) by executing shell commands via the
@useclawlink/cliinterface. - [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through external data processing.
- Ingestion points: Perplexity AI search results and chat completions retrieved via the API actions defined in
SKILL.md. - Boundary markers: Absent. The skill provides no instructions or delimiters to isolate untrusted external content from agent instructions.
- Capability inventory: Shell command execution via the
@useclawlink/cliinSKILL.md. - Sanitization: Absent. There is no evidence of validation or sanitization for the content returned by Perplexity AI.
Audit Metadata