The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied "raw ideas" and external "brand notes" to generate social media drafts, presenting an attack surface where instructions embedded in that data could influence agent behavior.
Ingestion points: User-provided text for post ideas and brand voice notes retrieved via the manage_brand and inspect_voice tools (SKILL.md, references/lane.md).
Boundary markers: The instructions do not specify the use of delimiters or protective instructions to isolate the ingested content from the agent's internal logic.
Capability inventory: The skill utilizes manage_post to create and publish content and manage_media to upload data to an external API (SKILL.md).
Sanitization: No explicit sanitization or filtering of the ingested content is described before it is interpolated into drafts.
[DATA_EXFILTRATION]: Transmits post content and media assets to the vendor's official API at app.andypartner.com as part of its core functionality (SKILL.md, references/lane.md).
[COMMAND_EXECUTION]: Includes example initialization steps using shell echo commands to display service configuration and available tools (SKILL.md).
[CREDENTIALS_UNSAFE]: References the use of workspace API keys (andy_sk_) and provides safety instructions to pipe keys instead of echoing them to prevent log exposure (references/lane.md).

andy-post