windsurf-common-errors
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for the agent to ingest and analyze local application configuration files, creating a surface for potential indirect prompt injection.
- Ingestion points:
~/.codeium/windsurf/mcp_config.json(as described in Error 7 instructions) - Boundary markers: None present in the instructions.
- Capability inventory: The skill utilizes
Read,Grep, andBash(curl:*)tools. - Sanitization: No specific sanitization or filtering logic is provided for the ingested content.
- [COMMAND_EXECUTION]: The skill uses standard shell commands to verify the local environment and application status.
- Evidence:
windsurf --versionandls ~/.codeium/are used as diagnostic examples to check installation and authentication state.
Audit Metadata