The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses sensitive conversation logs located at ~/.pi/agent/sessions/. While the primary output is described as local, the "Analyze" feature spawns "subagents," which typically involves sending these private transcripts to external LLM providers for processing.
[COMMAND_EXECUTION]: The skill executes local Javascript files (analyze.js and model-performance.js) using the Node.js runtime. This gives the scripts full access to the user's filesystem and environment context.
[EXTERNAL_DOWNLOADS]: The setup instructions require running npm install within a directory containing code from an unverified external source (github.com/ferologics/pi-skills). This may download numerous third-party dependencies that have not been vetted for security.
[PROMPT_INJECTION]: The skill processes session transcripts, which are considered untrusted data. This introduces a surface for Indirect Prompt Injection (Category 8):
Ingestion points: Reads transcripts from ~/.pi/agent/sessions/ via analyze.js.
Boundary markers: None identified in the provided documentation to prevent subagents from obeying instructions found within the transcripts.
Capability inventory: The skill can read local files and execute commands via Node.js scripts.
Sanitization: No sanitization or filtering of the transcript content is mentioned before it is processed by the AI subagents.

session-analyzer