The Agent Skills Directory

[SAFE]: Authentication is handled securely through environment variables (RECOUP_API_KEY or RECOUP_ACCESS_TOKEN). The instructions explicitly direct users to set these in the environment rather than hardcoding them into the skill scripts.
[SAFE]: Network operations are restricted to the vendor's official domain (api.recoupable.dev) for legitimate research, artist resolution, and content analysis purposes.
[SAFE]: The skill implements a quality control hook in SKILL.md that requires visual assets to pass an analysis check (POST /api/content/analyze) before the agent can claim the task is complete. This serves as a functional guardrail to ensure output quality and is not a malicious concealment attempt.
[INDIRECT_PROMPT_INJECTION]: The skill processes external research data and career feeds which represent a potential attack surface for indirect prompt injection.
Ingestion points: Career feeds, milestones, and web trend data retrieved from api.recoupable.dev/api/research/* (referenced in references/research-context.md).
Boundary markers: No explicit prompt boundary markers are used when interpolating these values into generation prompts.
Capability inventory: The skill uses curl for API requests, reads/writes within the artists/ workspace, and can route tasks to other content-specific skills.
Sanitization: Data is parsed using jq to extract specific structured fields, providing a level of protection by ignoring arbitrary fields or malicious injections outside the expected schema.

recoup-content-reactive-post