openclaw-lark-integration
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The documentation provides instructions to install the
openclawcore and the@larksuite/openclaw-larkplugin from the public NPM registry.\n- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of external data from Lark workspaces (messages, docs, etc.), which creates a potential surface for indirect prompt injection.\n - Ingestion points: Untrusted data enters the agent context via messaging and document retrieval functions as shown in the
SKILL.mdcode examples.\n - Boundary markers: The documentation supports a
customSystemPromptfor agent instructions, though specific data delimiters are not shown in the snippets.\n - Capability inventory: The skill grants the agent capabilities to send messages, manage documents, and edit database records through the Lark API.\n
- Sanitization: The skill documents robust architectural mitigations, including disabling group chat by default, supporting group whitelists, and requiring confirmation for destructive operations.\n- [COMMAND_EXECUTION]: The skill includes standard CLI commands for installation and service management, which are appropriate for a developer-focused integration.
Audit Metadata