The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes raw material from a preceding 'Madman' phase to generate mindmaps. It lacks instructions to use boundary markers or to ignore instructions embedded within this external content, which could allow malicious data to influence agent behavior.
Ingestion points: The skill ingests raw material dumps from the 'Madman' phase, as described in Step 1 of the Core Workflow in SKILL.md.
Boundary markers: The instructions do not define delimiters or explicit warnings for the agent to disregard potential instructions within the ingested material.
Capability inventory: The skill directs the agent to generate and write Markdown files for each whirlybird option (SKILL.md).
Sanitization: No specific filtering or validation steps for the input data are described.
[EXTERNAL_DOWNLOADS]: The skill includes references to external documentation and author profiles hosted on GitHub and vendor domains.
Automated security scanners flagged the documentation URL https://jeffallan.github.io/writing-with-agents/skills/structure/whirlybird/ as malicious.
This URL is a vendor resource hosted on a well-known service (github.io) and is documented neutrally as a link to the skill's official documentation and author information.

whirlybird