Smart Poller

Periodically polls a Feishu task board to auto-detect and execute tasks assigned to the current AI agent with optional silent mode and feedback.

This skill appears to do what it says, but review and run it cautiously: - Provide a Feishu App (app_id/app_secret) with the minimal scopes required for reading/writing the target document; avoid using org-wide/high-privilege credentials. - Be aware credentials are stored in a local config.json file; keep that file protected (filesystem permissions) or adapt the code to use your platform's secret store if preferred. - Ensure your runtime (python3 or node) is available and run the Python single-run mode first (python3 poller.py config.json --once) or test with test.py to verify parsing behavior. - Inspect the code yourself (it's included) and, if deploying long-term, run the poller in an isolated environment/network and restrict outbound access to open.feishu.cn only. - If you need the skill declared to the platform as requiring specific runtimes or secret-handling, request the author update registry metadata to list Python/Node requirements and to document how secrets should be supplied.

Type: OpenClaw Skill Name: smart-poller Version: 1.0.0 The Smart Poller skill is a legitimate automation tool designed to synchronize tasks between a Feishu (Lark) task board and an AI agent. The code in poller.py and poller.js implements standard OAuth2-style token management and document parsing using official Feishu APIs (open.feishu.cn). There is no evidence of data exfiltration, shell command execution, or malicious prompt injection; the 'execution' logic is currently limited to string-based status updates. The skill uses only standard libraries and follows best practices by requiring credentials to be supplied via a local config.json file.