The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the ruby interpreter to run local vendor scripts located at lib/buttercut/report.rb and lib/buttercut/library.rb. These tools are used to check the software edition and transmit feature requests.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it ingests user-supplied text for the title and narrative fields of the report. Evidence: Title and narrative arguments in Step 3 of SKILL.md. Ingestion points: User words drafted in Step 2. Boundary markers: The skill instructs the agent to read the draft back for user approval before execution. Capability inventory: Accesses internal Ruby scripts with shell arguments. Sanitization: Explicit instructions are provided to exclude sensitive identifiers such as library, clip, client, or subject names from the narrative.
[DATA_EXFILTRATION]: The skill facilitates the transmission of user-approved feature ideas to the development team. It accesses local configuration files like libraries/settings.yaml to check for user consent and associated contact information. This behavior is the stated purpose of the skill and is performed with user confirmation.

request-feature