The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The SKILL.md file instructs users to install dependencies by piping a remote script directly into the shell using curl -LsSf https://astral.sh/uv/install.sh | sh. This is a high-risk pattern that executes unverified remote code.
[EXTERNAL_DOWNLOADS]: The skill uses uvx to download and execute the minimax-coding-plan-mcp package. This package is an unverified dependency from an external registry, and its source code is not audited by the skill.
[COMMAND_EXECUTION]: The Python script scripts/web_search.py uses subprocess.Popen to launch the uvx command and interact with the minimax-coding-plan-mcp server. It passes user-provided search queries into this process.
[DATA_EXFILTRATION]: The skill reads API keys from sensitive local paths, including ~/.openclaw/config/minimax.json and ~/.openclaw/agents/main/agent/auth-profiles.json. While these keys are used for the search functionality, accessing centralized credential stores increases the risk of accidental exposure or theft.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8).
Ingestion points: The output of the web_search.py script is returned to the agent context.
Boundary markers: There are no delimiters or instructions to prevent the agent from obeying commands embedded in the search results.
Capability inventory: The skill has the ability to execute subprocesses and shell commands.
Sanitization: There is no sanitization or filtering of the content retrieved from the web search.

minimax-web-search