The Agent Skills Directory

[SAFE]: The skill provides a robust and standard interface for binary analysis within IDA Pro. The SQL tables and functions described (e.g., funcs, segments, instructions, cfg_edges) are consistent with intended professional reverse engineering workflows and do not exhibit malicious patterns such as exfiltration or unauthorized privilege escalation.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted binary data, which serves as a potential attack surface for poisoning the agent's context.
Ingestion points: The agent reads data from the IDA database using tables like instructions, names, heads, and bytes as detailed in SKILL.md and references/disassembly-tables.md.
Boundary markers: The skill does not provide specific instructions or markers to isolate binary-derived content from the agent's operational instructions.
Capability inventory: The skill enables modifications to the database via INSERT, UPDATE, and DELETE operations, and permits file generation using gen_listing and gen_cfg_dot_file.
Sanitization: No mechanisms for sanitizing or validating binary strings are implemented or mentioned.

disassembly