The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on a custom CLI tool named skill to perform all operations, including interacting with external APIs (Front, Inngest, Linear).
[DATA_EXFILTRATION]: Several commands allow exporting or pulling data to local files (e.g., skill front pull-conversations --output data.json, skill responses export -o bad-responses.json). This behavior is consistent with the skill's purpose for data analysis and evaluation but involves sensitive customer information.
[CREDENTIALS_UNSAFE]: The skill manages and configures sensitive API tokens (FRONT_API_TOKEN, INNGEST_SIGNING_KEY, LINEAR_API_KEY). It includes a configuration command (skill config set) to store these secrets locally.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from external sources that may contain malicious instructions for the agent.
Ingestion points: Data enters the agent's context through commands like skill front conversation (which includes message history), skill inngest events, and skill linear issues.
Boundary markers: The instructions do not define clear delimiters or provide the agent with warnings to ignore instructions embedded within the retrieved data.
Capability inventory: The agent has access to powerful write and administrative tools, including skill front reply, skill front archive, skill linear create, and a raw API 'escape hatch' (skill front api).
Sanitization: There is no mention of sanitization or filtering of the content retrieved from external platforms before it is processed by the agent.

skill-cli