The Agent Skills Directory

[COMMAND_EXECUTION]: The skill contains multiple shell script templates that use tools like curl, jq, grep, and python3 to interact with external web servers and process local files (e.g., httpx.jsonl). These commands are used to probe for cache vulnerabilities by injecting headers and verifying reflections.
[EXTERNAL_DOWNLOADS]: The documentation suggests installing the web-cache-vulnerability-scanner package via pip. This is an external dependency that is not part of a standard trusted vendor list, representing a third-party code inclusion.
[PROMPT_INJECTION]: The skill is designed to ingest and analyze data from external web server responses, which is an untrusted source.
Ingestion points: HTTP response headers and bodies fetched via curl are processed in SKILL.md to identify reflections.
Boundary markers: The instructions do not define explicit boundaries or delimiters to isolate target output from the agent's internal reasoning context.
Capability inventory: The agent has the capability to execute shell commands, read local files, and perform network operations.
Sanitization: Content is passed through grep for string matching, but there is no explicit sanitization of the response content before it is presented back to the agent.

cache-poisoning