The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes HTTP response headers from external target APIs to determine rate-limiting configurations. This creates a surface for indirect prompt injection if the agent interprets responses from a malicious server.
Ingestion points: The probe_rate_limit function in SKILL.md and the detect_rate_limit_headers function in scripts/agent.py read headers like X-RateLimit-Limit and Retry-After directly from HTTP responses.
Boundary markers: No boundary markers or ignore instructions are used when processing these external inputs.
Capability inventory: The skill has the capability to perform network requests using the requests and aiohttp Python libraries.
Sanitization: The script does not sanitize or validate the content of the response headers before processing and displaying them.
[EXTERNAL_DOWNLOADS]: The skill references standard Python dependencies (requests, aiohttp, asyncio) and external cybersecurity resources.
The references/api-reference.md file links to a well-known cybersecurity educational site (HackTricks) for further research on rate-limiting bypass techniques.
The instructions specify the use of common security tools including Burp Suite with the Turbo Intruder extension.
[COMMAND_EXECUTION]: The documentation includes instructions for using command-line tools such as cURL to perform manual header manipulation and testing.

performing-api-rate-limiting-bypass