The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies and extracts patterns from session history, which includes contents of files read or modified that may originate from untrusted sources, creating an attack surface for indirect prompt injection.
Ingestion points: The "Analysis Process" section in SKILL.md instructs the agent to examine all files read and modified during the session to identify patterns and insights.
Boundary markers: The instructions lack specific delimiters or negative constraints to prevent the agent from following instructions hidden within the session data being analyzed.
Capability inventory: The skill leverages the Edit tool to write extracted knowledge to CLAUDE.md and CLAUDE.local.md in the project root.
Sanitization: The skill implements a robust mitigation by requiring mandatory user confirmation via the AskUserQuestion tool before any data is persisted, allowing the user to filter out unwanted or suspicious extracted items.

learn