The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, including binary files, firmware samples, and network captures for the purpose of reverse engineering. Malicious files could contain embedded strings or payloads designed to manipulate the agent's logic or influence its analysis conclusions.
Ingestion points: Binary files, firmware images, memory dumps, and code snippets provided by the user for analysis as defined in SKILL.md.
Boundary markers: The skill implements a structured state machine and an "Analysis Contract" (references/analysis-contract.md) to maintain state, but it does not specify explicit instruction-guarding delimiters when presenting data extracted from untrusted binaries to the agent's context.
Capability inventory: The skill provides access to powerful analysis capabilities including debugging, decompilation, memory dumping, dynamic instrumentation (Frida), and binary patching (references/capabilities.md).
Sanitization: There is no documented process for sanitizing, escaping, or filtering strings and metadata extracted from target binaries before they are processed by the agent.

re-analyze