The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The integrity verification step in the procedure reads content from the project's memory index file and processes it using a shell loop. If this file contains specially crafted markdown links, it could lead to the execution of arbitrary commands. * Ingestion points: The skill reads MEMORY.md and associated topic files from the local filesystem. * Boundary markers: The instructions do not define clear boundaries or escape sequences for data parsed from these files. * Capability inventory: The skill utilizes Bash, Grep, Read, and Write tools to perform its tasks. * Sanitization: There is no validation or sanitization of the strings extracted from the markdown links before they are executed in the shell loop.
[COMMAND_EXECUTION]: The skill executes shell commands via Bash to perform line counts and file listings. Specifically, the loop in Step 6 (for f in $(grep ...); do ls ...; done) presents a risk because it dynamically constructs commands using data retrieved from a user-editable markdown file.

manage-memory