larksync-feishu-local-cache_skill

Overview

This skill synchronizes Feishu (Lark) documents into a local cache so OpenClaw can read files locally first, greatly reducing Feishu API calls and token costs. It converts high-frequency query-time API access into low-frequency scheduled syncs, preserving existing user workflows while improving stability and cost efficiency. The default mode is download-only with daily sync, and advanced modes (bidirectional or upload-only) are available only with explicit consent.

How this skill works

The skill verifies the local LarkSync service and Feishu authorization, then creates and manages scheduled sync tasks that mirror cloud folders into a local directory. On setup it can run an immediate sync to build a baseline cache; thereafter OpenClaw queries use the local Markdown/attachment store before falling back to cloud APIs. The tool returns JSON outputs from helper scripts to simplify automation and auditing.

When to use it

• You want to reduce Feishu API frequency and token consumption for OpenClaw document Q&A.
• You need a reliable local copy of team documents for offline or low-latency retrieval.
• You prefer scheduled, low-frequency synchronization rather than per-query cloud access.
• You want to set up a safe download-only mirror before enabling any cloud-to-local writes.
• You need to automate audit-friendly sync tasks with JSON output for orchestration.

Best practices

• Start with download_only and a daily low-frequency schedule to establish a safe baseline.
• Run the check command first to verify authorization and local service availability before creating tasks.
• Keep the sync target in a dedicated test folder when evaluating bidirectional mode.
• Preserve helper script JSON outputs for tracing, auditing, and automated workflows.
• Inform and obtain explicit user consent before enabling bidirectional or upload behaviors.

Example use cases

• Mirror a production Feishu folder to D:\Knowledge\FeishuMirror every day at 01:00 and use that local cache for document Q&A.
• Bootstrap a new OpenClaw deployment by running an immediate sync to create the initial local dataset.
• Check LarkSync authorization and task health during onboarding or troubleshooting.
• Switch a directory to bidirectional sync after risk assessment and test in a sandbox folder first.
• Auto-reuse an existing conflicting task (same path + cloud folder) instead of creating duplicates when a 409 occurs.