The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from past sessions and codebases, which constitutes a surface for indirect prompt injection.
Ingestion points: Data is retrieved via search, timeline, and get_observations tools as defined in SKILL.md.
Boundary markers: Absent; there are no instructions for the agent to use delimiters or specific ignore-tags when processing retrieved content.
Capability inventory: The skill utilizes read-only MCP tools; it does not have the capability to write files, perform network operations, or execute arbitrary shell commands.
Sanitization: Not present; however, the skill explicitly instructs the agent to treat retrieved memory as a "starting point for investigation" rather than ground truth.
[COMMAND_EXECUTION]: The skill instructs the agent on how to execute specific MCP tools (search, timeline, get_observations, smart_search, smart_outline, smart_unfold) to interact with the local environment and session history.

memory-recall