The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of API error responses which could theoretically be manipulated by a malicious server. 1. Ingestion points: The Quick Diagnostic Script saves API responses from https://api.exa.ai/search to /tmp/exa-test.json. 2. Boundary markers: None are used when displaying the raw JSON output from the temporary file. 3. Capability inventory: The skill utilizes Bash, Grep, and Read tools to execute diagnostics and inspect files. 4. Sanitization: Instructions advise users to redact evidence, though the diagnostic script outputs raw response data upon failure.
[COMMAND_EXECUTION]: The skill uses Bash to perform network requests and process JSON data as part of its diagnostic functions. 1. Uses curl to interact with the official Exa API endpoint for connectivity testing. 2. Employs python3 -m json.tool to format diagnostic output for readability. 3. These operations are standard for debugging and align with the skill's stated purpose.

exa-common-errors