The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a 'Companion check' instruction that directs the agent to run an ls command on multiple hidden directories (such as ~/.claude/skills/ and ~/.agent/skills/). This constitutes environment discovery of other installed tools outside the skill's own data directory (~/.memory/).
[EXTERNAL_DOWNLOADS]: Installation instructions in the README and SKILL.md recommend using npx skills add AbsolutelySkilled/AbsolutelySkilled to install the skill and its recommended companions. This pattern involves fetching and executing code from an external repository that is not identified as a well-known or trusted service.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its automated memory loading mechanism. Ingestion points: It reads content from all markdown files in ~/.memory/ and the index.yaml registry file. Boundary markers: The loading logic does not specify the use of delimiters or 'ignore' instructions to prevent the agent from obeying commands embedded within the stored memories. Capability inventory: The skill has the ability to create directories (mkdir), list files (ls), and perform persistent writes/deletes within the user's home directory. Sanitization: There is no evidence of sanitization or verification of the content retrieved from the local memory files before it is injected into the agent's context.

second-brain