The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from an external social media platform (TikTok) via the Frevana API. \n
Ingestion points: The wait_for_result function in scripts/tiktok_task.py retrieves the final JSON array from the API result endpoint. \n
Boundary markers: The SKILL.md instructions mandate that result bytes be returned unchanged without wrapping or reshaping, which provides no delimitation between the data and the agent's context. \n
Capability inventory: The skill has the ability to perform network requests (urllib.request) and write files to the local system (shutil.copyfile). \n
Sanitization: There is no sanitization or filtering of the TikTok post content retrieved from the API before it is delivered to the agent.
[COMMAND_EXECUTION]: The skill employs a shell wrapper and subprocess calls for its core functionality and testing. \n
The file scripts/tiktok_task.sh executes the primary Python logic using exec python3. \n
The test suite tests/test_tiktok_task.py uses subprocess.run to call the script wrapper, which is standard for local tool execution but involves spawning new processes.

tiktok-posts-discover-by-keyword