The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes 'memory-hub' CLI commands via Bash to initialize the project structure, write knowledge base files, and update the project catalog. These operations are intended and necessary for the skill's functionality.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it reads and processes untrusted project files to generate documentation. 1. Ingestion points: Reads package manifests, entry files, and configuration examples from the project root (SKILL.md). 2. Boundary markers: Uses shell heredocs (EOF) to pass data to the tool. 3. Capability inventory: Uses Bash, Read, and Glob tools. 4. Sanitization: File contents are processed by the LLM without specific sanitization steps.
[SAFE]: No evidence of data exfiltration, credential theft, or malicious remote code execution was found. The skill operates locally on the project files as described.

memory-init