The Agent Skills Directory

[SAFE]: The skill is a legitimate tool for documentation management and project memory maintenance. It follows security best practices by explicitly instructing the agent to never store secrets, API keys, or sensitive infrastructure details in the managed files.
[INDIRECT_PROMPT_INJECTION]: The skill identifies a potential attack surface by ingesting data from the claude-mem MCP tool (via search, timeline, and get_observations tools). This risk is effectively mitigated by clear instructions that forbid verbatim copying of raw observations, require the agent to curate and compress information into high-signal guidance, and mandate human review/acceptance of all generated patches before they are committed to the repository.
[REMOTE_CODE_EXECUTION]: There are no patterns of remote code execution or unauthorized package installations. The skill relies on MCP tools that must be pre-installed by the user in a controlled environment.

claude-mem-coded-assistant