The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute grep and sed. These commands are used to calculate file paths and search through the project's memory directory based on the user's current working directory.
[DATA_EXFILTRATION]: The skill accesses sensitive file paths within the $HOME/.claude/projects/ directory. This location contains project-specific conversation histories and memory files that may include proprietary information or sensitive code snippets.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection when processing historical project data.
Ingestion points: Data is ingested from the project's memory directory via grep in Step 1 of SKILL.md.
Boundary markers: No explicit delimiters or instructions are provided to distinguish between the skill's core instructions and the potentially untrusted content retrieved from memory.
Capability inventory: The skill has access to the filesystem through file-writing tools and can execute shell commands via Bash.
Sanitization: There is no evidence of sanitization, filtering, or validation of the content extracted from the project memory files before it is processed by the AI agent.

extract