editor-context

Pass

Audited by Gen Agent Trust Hub on Apr 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the one-ide-cli command-line interface to interact with the user's IDE (VS Code, Cursor, JetBrains, etc.) and retrieve project metadata.
  • [EXTERNAL_DOWNLOADS]: The skill references the one-ide-cli package and instructs installation via npm (npm install -g one-ide-cli). It also provides a link to the vendor's repository at github.com/tangcent/one-IDE. These are vendor-owned resources used for the skill's primary functionality.
  • [PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection as it ingests untrusted data from the user's local workspace into the AI's context.
  • Ingestion points: Full file content (activeFileText) and editor selections (activeSelectionText) are retrieved via the one-ide-cli context command.
  • Boundary markers: No specific delimiters or instructions are used to wrap or isolate the retrieved file content from the agent's instructions.
  • Capability inventory: The skill's primary capability is executing the one-ide-cli tool to read workspace information.
  • Sanitization: No sanitization or filtering is applied to the content of the active files before they are provided to the model.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 2, 2026, 10:53 PM
Security Audit — agent-trust-hub — editor-context