The Agent Skills Directory

[SAFE]: The skill defines a benign workflow for forensic network mapping. It uses standard static analysis techniques, such as cross-referencing (xrefs) and string extraction, to identify remote endpoints and data transmission logic without executing the target program.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface as it is designed to ingest and process data from potentially untrusted programs.
Ingestion points: Data from analyzed programs enters the agent context via list_procedures, list_strings, and xrefs as described in SKILL.md.
Boundary markers: No specific delimiters are mentioned in the instructions to distinguish external program data from the agent's internal instructions.
Capability inventory: The skill's capabilities are limited to modifying the analysis session through rename_function, add_note, and open_function.
Sanitization: The instructions do not specify any sanitization or validation of the strings extracted from the target program.
Note: This surface is inherent to the functionality of a code analyzer and does not present a specific threat in the context of the defined workflow.

network-map