The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/burp-search.sh passes all command-line arguments directly to a Java subprocess using the "$@" variable. This pattern lacks programmatic sanitization, relying entirely on the agent following the documented instructions to avoid passing dangerous flags or malformed input to the shell.
[PROMPT_INJECTION]: The skill processes untrusted external data from Burp Suite projects, which could contain malicious instructions embedded in HTTP headers or bodies.
Ingestion points: Data is read through the burp-search.sh wrapper script.
Boundary markers: The SKILL.md file provides instructions for the agent to use head -c and jq for output control, but these boundaries are not programmatically enforced by the tool logic.
Capability inventory: The skill provides access to the Bash execution environment and Read tool.
Sanitization: There is no automated sanitization or escaping of extracted content before it is returned to the agent's context.
[EXTERNAL_DOWNLOADS]: The documentation references and requires an external third-party extension from a GitHub repository (BuffaloWill/burpsuite-project-file-parser) which must be manually installed by the user.

burpsuite-project-parser