lsp-understand
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is strictly limited to read-only code analysis using LSP tools. All functions are designed to retrieve information about symbol definitions, implementations, and call hierarchies.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of source code and hover documentation. While this creates a potential surface for indirect prompt injection, the risk is mitigated by the read-only nature of the toolset.
- Ingestion points: Source code and hover signatures are ingested through
mcp__lsp__get_symbol_sourceandmcp__lsp__inspect_symbolinSKILL.md. - Boundary markers: The synthesized Code Map output does not explicitly use delimiters to separate ingested code from instructions.
- Capability inventory: The skill only utilizes read-only LSP inspection tools; it cannot write to files, access the network, or execute system commands.
- Sanitization: No sanitization is applied to the retrieved source code or documentation before it is presented to the agent.
Audit Metadata