awesome-openclaw-agents
Warn
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches agent templates and bot code from a third-party GitHub repository (
github.com/mergisi/awesome-openclaw-agents.git). - [REMOTE_CODE_EXECUTION]: Instructions direct the user to clone an external repository and run its code using
npm installfollowed bynode bot.js. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of agents designed to ingest untrusted data from multiple external sources.
- Ingestion points: GitHub webhooks (
SKILL.md), Slack channel messages (SKILL.md), and Gmail inbox content (SKILL.md). - Boundary markers: Relies on
SOUL.mdguardrails sections; however, the provided examples lack explicit structural delimiters or "ignore" instructions for processed data. - Capability inventory: The agents are granted write-access capabilities including posting GitHub comments, sending Slack messages, and modifying Gmail labels.
- Sanitization: Implementation snippets do not demonstrate sanitization, filtering, or validation of data retrieved from external APIs before processing by the agent.
- [COMMAND_EXECUTION]: Provides instructions for running local CLI tools (
openclaw,crewclaw) and environment configuration that involves handling API keys for multiple services.
Audit Metadata