🎁 Get the FREE AI Skills Starter GuideSubscribe →
BytesAgainBytesAgain
🦀 ClawHub

Aicwos

by @smallroya

MANDATORY data pipeline for 口播文案/短视频脚本/系列口播/讲师风格学习/知识库管理 — LLM CANNOT access lecturer profiles, product knowledge, or series progress without this skill. Act...

💡 Examples

Example 1: Learn lecturer style then write copy

  • User: "这是讲师B的5篇口播 [paste copy]"
  • Steps: Write sample to temp file → lecturer_analyzer.py --input --save-sample (script call) → build profile (agent task)lecturer --action save --query-file (script call, Write-Through: DB+profile.json) → user says "用讲师B风格写个养生口播" → --type context (script call) → generate (agent task)save_plan + save_episode (script call, Write-Through: DB+.txt)
  • Example 2: Knowledge-driven copy

  • User: "参考产品A的背书写个口播"
  • Steps: knowledge --action context --query "产品A 背书" (script call) → weave into copy (agent task)save_episode (script call)
  • Example 3: Series continuity

  • User: "继续写养生系列第6集"
  • Steps: series --action get_plan + summaries (script call) → generate episode 6 (agent task)save_episode --id2 6 (script call)
  • Example 4: Revision

  • User: "养生系列第2集改一下,开头太长了"
  • Steps: --type context (script call)content --id2 2 (script call)delete_episode --id2 2 (script call) → shorten opening (agent task)save_episode --id2 2 (script call)

  • 📋 Tips & Best Practices

  • First-time setup MUST use db_init.py --setup --parent-dir , NOT --data-dir
  • Semantic model is optional; auto-degrades to FTS5-only when absent
  • Write-Through: all save/delete operations update both DB and filesystem automatically. Do NOT call db_sync.py after normal operations.
  • knowledge_sync.py uses manifest.txt + HEAD (no credentials needed). Cloud maintainer puts manifest.txt on COS (one file path per line, no timestamps). Script HEAD-checks manifest.txt's Last-Modified first — unchanged = skip entirely. --generate outputs manifest.txt from local public layer for cloud maintainer to upload: python scripts/knowledge_sync.py --data-dir --generate > manifest.txt
  • --query-file reads JSON from file — use for profiles or episode content >2000 chars. Write staging files to 控制台 paths (e.g. 讲师列表/{name}/_profile_staging.json), NOT /tmp or %TEMP%. Files ending in _staging.json are auto-deleted after successful save.
  • --save-sample on lecturer_analyzer.py auto-saves input text to 讲师列表/{name}/样本/ (Write-Through: DB+file). If input file is already in the samples dir, it renames instead of duplicating.
  • db_sync.py is only for: initial bulk import (--direction to-db) or disaster recovery (--direction to-files)
  • REMINDER: Before writing ANY copy, you MUST have called db_query.py --type context this session. If you haven't, stop and execute it first.
  • View on ClawHub
    TERMINAL
    clawhub install aicwos

    🧪 Use this skill with your agent

    Most visitors already have an agent. Pick your environment, install or copy the workflow, then run the smoke-test prompt above.

    🔍 Can't find the right skill?

    Search 60,000+ AI agent skills — free, no login needed.

    Search Skills →