The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute the ax (axctl) command-line tool using parameters derived from user input (SHAs, dates, topics). There is a risk of command injection if the agent implementation does not correctly sanitize these inputs before shell execution. Examples include ax sessions near <sha>, ax sessions around <date>, and ax recall "<topic>". It also instructs the agent to recommend the execution of a local script scripts/db-start.sh if database connection fails.
[PROMPT_INJECTION]: The skill processes historical session data and tool outputs, which may contain untrusted content from past interactions, creating a surface for indirect prompt injection. Ingestion points: Data is ingested through ax sessions show --json and ax recall commands referenced in SKILL.md. Boundary markers: Absent. The skill does not provide instructions to treat historical data as untrusted or provide delimiters to separate retrieved data from agent instructions. Capability inventory: The skill uses shell command execution via the ax CLI to query and show session details. Sanitization: Absent. The skill does not specify any filtering, escaping, or validation of the retrieved session content before it is narrated to the user.

ax-extract-workflow