The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a feature for power users to integrate custom publishing logic. The lib/backend_selector.py script reads a command from the THREADS_SKILLS_CUSTOM_POSTER environment variable and executes it using subprocess.run. This is a documented developer feature and is implemented without using a shell, mitigating standard injection risks.
[EXTERNAL_DOWNLOADS]: The skill communicates with the Publora API (api.publora.com) for post publishing, the Apify API (api.apify.com) for niche research, and the Pixfaro API (api.pixfaro.com) for AI image generation. These external connections are core to the skill's functionality and target well-known service providers.
[INDIRECT_PROMPT_INJECTION]: The skill bundle contains various tools that ingest and process untrusted text (e.g., social media posts from LinkedIn, X, or Threads) to facilitate content repurposing and 'humanizing'.
Ingestion points: Skills like threads-repurposer and threads-reply-drafter ingest user-pasted text strings.
Boundary markers: The skill instructions include explicit directives (found in sub-skills/story-bank-interview.md) for the agent to treat all external content as data and never as instructions, reducing the risk of obeying embedded commands.
Capability inventory: The skill has network access and local script execution capabilities (via configuration).
Sanitization: Platforms artifacts are removed during the transformation process, though inputs are primarily managed through LLM reasoning instructions.

threads-marketing