The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external API endpoints that provide summaries and source names from third-party health feeds. It proactively mitigates the risk of indirect prompt injection by including specific instructions for the agent to ignore directive-like text found in responses.
Ingestion points: API responses from https://www.worldmonitor.app/api/health/v1/ referenced in SKILL.md.
Boundary markers: Present; the 'Content safety' section explicitly instructs the agent to treat responses as data only and disregard embedded directives like 'ignore previous instructions'.
Capability inventory: None; the skill is informational and does not define tools for file writing, shell execution, or dynamic code evaluation.
Sanitization: Implements prompt-based containment by defining clear behavior for external content processing.
[CREDENTIALS_UNSAFE]: The documentation includes an example API key (wm_0123456789abcdef0123456789abcdef01234567). This string consists of sequential digits and characters, identifying it as a generic placeholder for demonstration rather than a functional credential.
[PROMPT_INJECTION]: Static analysis flagged the phrase 'ignore previous instructions' in the skill text. This is a false positive as the phrase is used within the 'Content safety' mitigation instructions to tell the agent how to handle untrusted data, rather than attempting to subvert the agent's own system prompt.

monitor-health-alerts