lsp-index-engineer
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to install language servers globally (
npm install -g) and verify their connectivity via stdin. - [EXTERNAL_DOWNLOADS]: The agent is directed to download and install several language servers from the NPM registry. While most are well-known tools (e.g.,
typescript-language-server,pyright), others likegoplsandrust-analyzerare typically distributed through other package managers (Go/Rust specific), which may lead to the installation of non-standard or unexpected packages if fetched from the NPM registry. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process arbitrary project source code to build a semantic index and provide hover documentation.
- Ingestion points: The agent reads source files (
.ts,.php,.py, etc.) and processes LSP responses containing symbol names and documentation strings. - Boundary markers: No specific delimiters or instructions are provided to the agent to distinguish between its own logic and potentially malicious instructions embedded within the source code data it processes.
- Capability inventory: The agent has file system access (reading code, writing cache files) and can interact with external language server processes.
- Sanitization: There are no explicit instructions for the agent to sanitize or validate the data extracted from source files before adding it to the index or presenting it during navigation tasks.
Audit Metadata