The Agent Skills Directory

[DATA_EXPOSURE]: The skill implements secure credential management by instructing users to export the OPENROUTER_API_KEY as an environment variable and accessing it via os.environ in Python and process.env in TypeScript. No hardcoded secrets are present.
[COMMAND_EXECUTION]: The skill uses Bash(curl:*) and Bash(jq:*) within its instructions to verify API functionality and test fallback logic. These operations are limited to the intended utility of the skill and target a well-known service (OpenRouter).
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface as it processes untrusted data.
Ingestion points: Untrusted data enters the agent context through the messages list parameter in the client.chat.completions.create calls located in SKILL.md and references/examples.md.
Boundary markers: Explicit delimiters or warnings to ignore embedded instructions are absent in the provided code snippets.
Capability inventory: The skill has network access capabilities via the OpenAI SDK and curl commands.
Sanitization: No sanitization or validation of the content within the messages array is demonstrated in the examples.
[DYNAMIC_EXECUTION]: The skill correctly uses yaml.safe_load() in references/configuration-examples.md to prevent unsafe deserialization vulnerabilities when loading fallback configurations.

openrouter-fallback-config