The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches official documentation from docs.pluggy.ai via an MCP or WebFetch of an llms.txt index. These are well-known resources belonging to the vendor (Pluggy) and are used solely for the stated purpose of code review.
[COMMAND_EXECUTION]: The skill mentions command-line instructions (e.g., claude mcp add) but does not execute them automatically. They are provided as documentation to the user for setting up the environment.
[PROMPT_INJECTION]: The skill instructions contain strong formatting and guidance but do not attempt to bypass AI safety filters or override system-level instructions in a malicious way. The use of 'Golden rule' and 'Requirement' phrases are internal logic for the code review workflow.
[DATA_EXFILTRATION]: No evidence of sensitive data exfiltration. The skill reads local integration files for analysis but only sends queries to the official documentation endpoints specified by the vendor.

pluggy-doctor