The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions guide users to host the Kanban board locally by running node server.js via the platform's preview tool. Additionally, the kb_export_for_sharing function in kanban_agent_skill/exports.py programmatically modifies a copy of the connector script and uses shutil to package it for distribution.
[EXTERNAL_DOWNLOADS]: The Kanban board's user interface (kanban.html) loads React and Babel libraries from unpkg.com, which is a well-known and trusted content delivery network.
[DATA_EXFILTRATION]: The Python connector in kanban_agent_skill/exports.py performs network operations using urllib.request to communicate with the Kanban board server. These requests target localhost or the vendor's community domain (community.iamstarchild.com) to manage tasks, which is the intended purpose of the skill.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by processing task data. 1. Ingestion points: Task content is ingested through kb_list_tasks, kb_get_task, and kb_summary functions in kanban_agent_skill/exports.py. 2. Boundary markers: Absent. There are no delimiters or specific instructions to the agent to disregard commands embedded within task data. 3. Capability inventory: The connector script in kanban_agent_skill/exports.py possesses network access (urllib.request) and file system access (shutil). 4. Sanitization: Absent. Data from the Kanban board is processed as-is without content filtering or validation.
[SAFE]: The skill is transparent about its operational requirements and security model. It explicitly documents that the API is unauthenticated and advises against the use of sensitive data. All external references and network communications are directed toward trusted or vendor-owned infrastructure.

kanban