Asosiy tarkibga o'tish

​
9. Agent Runtime 🤖

Workspace (majburiy)

​
SOUL.md — persona, chegaralar, ohang

  1. OpenClaw agent uchun yagona ishchi katalog (cwd) sifatida asboblar va kontekst uchun bitta agent ish maydoni katalogidan (agents.defaults.workspace) foydalanadi.
  2. Tavsiya etiladi: agar mavjud bo‘lmasa, ~/.openclaw/openclaw.json ni yaratish va ish maydoni fayllarini boshlang‘ich holatga keltirish uchun openclaw setup dan foydalaning.
  3. Ish maydonining to‘liq tuzilishi va zaxira nusxa qo‘llanmasi: Agent workspace
  4. Agar agents.defaults.sandbox yoqilgan bo‘lsa, asosiy bo‘lmagan sessiyalar buni agents.defaults.sandbox.workspaceRoot ostidagi sessiya-bo‘yicha ish maydonlari bilan almashtirishi mumkin (qarang: Gateway configuration).

​
16. Bootstrap fayllari (kiritiladi)

  1. agents.defaults.workspace ichida OpenClaw quyidagi foydalanuvchi tahrirlashi mumkin bo‘lgan fayllarni kutadi:
    1. AGENTS.md — ishlash bo‘yicha ko‘rsatmalar + “xotira”
  • Queue rejimi steer bo‘lganda, kiruvchi xabarlar joriy run ichiga kiritiladi.
    1. TOOLS.md — foydalanuvchi tomonidan yuritiladigan asboblar bo‘yicha eslatmalar (masalan, imsg, sag, konventsiyalar)
    1. BOOTSTRAP.md — bir martalik birinchi ishga tushirish marosimi (yakunlangach o‘chiriladi)
    1. IDENTITY.md — agent nomi / vibe / emoji
    1. USER.md — foydalanuvchi profili + afzal murojaat shakli
  1. Yangi sessiyaning birinchi navbatida OpenClaw ushbu fayllarning mazmunini to‘g‘ridan-to‘g‘ri agent kontekstiga kiritadi.
  2. Bo‘sh fayllar o‘tkazib yuboriladi. 26. Katta fayllar qisqartiriladi va belgi bilan kesib tashlanadi, shunda promptlar ixcham bo‘lib qoladi (to‘liq mazmun uchun faylni o‘qing).
  3. Agar fayl mavjud bo‘lmasa, OpenClaw bitta “fayl yo‘q” belgisi qatorini kiritadi (va openclaw setup xavfsiz standart shablonni yaratadi).
  4. BOOTSTRAP.md faqat mutlaqo yangi ish maydoni uchun yaratiladi (boshqa bootstrap fayllari mavjud bo‘lmaganida). 29. Marosimni tugatgach uni o‘chirib tashlasangiz, keyingi qayta ishga tushirishlarda u qayta yaratilmasligi kerak.
  5. Bootstrap fayllarini yaratishni butunlay o‘chirish uchun (oldindan to‘ldirilgan ish maydonlari uchun), quyidagini sozlang: 
31. { agent: { skipBootstrap: true } }

​
32. Ichki asboblar

  1. Asosiy asboblar (o‘qish/ijro/tahrirlash/yozish va tegishli tizim asboblari) asboblar siyosatiga bo‘ysungan holda har doim mavjud. 34. apply_patch ixtiyoriy va tools.exec.applyPatch orqali cheklanadi. 35. TOOLS.md qaysi asboblar mavjudligini boshqarmaydi; u ularni siz qanday ishlatilishini xohlayotganingiz bo‘yicha ko‘rsatma beradi.

​
36. Ko‘nikmalar

  1. OpenClaw ko‘nikmalarni uchta joydan yuklaydi (nomlar to‘qnashganda ish maydoni ustun):
    1. Paketlangan (o‘rnatish bilan birga yetkaziladi)
    1. Boshqariladigan/mahalliy: ~/.openclaw/skills
    1. Ish maydoni: <workspace>/skills
  1. Ko‘nikmalar konfiguratsiya/muhit orqali cheklanishi mumkin (qarang: Gateway configuration dagi skills).

​
42. pi-mono integratsiyasi

Ulanish hayotiy sikli (bitta mijoz) The queue is checked after each tool call; if a queued message is present, remaining tool calls from the current assistant message are skipped (error tool results with “Skipped due to queued user message.”), then the queued user message is injected before the next assistant response.
    1. pi-coding agent runtime yo‘q.
    1. ~/.pi/agent yoki <workspace>/.pi sozlamalari hisobga olinmaydi.

​
46. Sessiyalar

  1. Sessiya transkriptlari JSONL formatida quyida saqlanadi:
  • ~/.openclaw/agents/<agentId>/sessions/<SessionId>.jsonl
  1. Sessiya ID barqaror bo‘lib, OpenClaw tomonidan tanlanadi.
  2. Meros bo‘lib qolgan Pi/Tau sessiya papkalari o‘qilmaydi.

​
Steering while streaming

Ulanish hayotiy sikli (bitta mijoz) The queue is checked after each tool call; if a queued message is present, remaining tool calls from the current assistant message are skipped (error tool results with “Skipped due to queued user message.”), then the queued user message is injected before the next assistant response. When queue mode is followup or collect, inbound messages are held until the current turn ends, then a new agent turn starts with the queued payloads. See Queue for mode + debounce/cap behavior. Block streaming sends completed assistant blocks as soon as they finish; it is off by default (agents.defaults.blockStreamingDefault: "off"). Tune the boundary via agents.defaults.blockStreamingBreak (text_end vs message_end; defaults to text_end). Control soft block chunking with agents.defaults.blockStreamingChunk (defaults to 800–1200 chars; prefers paragraph breaks, then newlines; sentences last). Coalesce streamed chunks with agents.defaults.blockStreamingCoalesce to reduce single-line spam (idle-based merging before send). Non-Telegram channels require explicit *.blockStreaming: true to enable block replies. Verbose tool summaries are emitted at tool start (no debounce); Control UI streams tool output via agent events when available. More details: Streaming + chunking.

​
Model refs

Model refs in config (for example agents.defaults.model and agents.defaults.models) are parsed by splitting on the first /.
  • Use provider/model when configuring models.
  • If the model ID itself contains / (OpenRouter-style), include the provider prefix (example: openrouter/moonshotai/kimi-k2).
  • If you omit the provider, OpenClaw treats the input as an alias or a model for the default provider (only works when there is no / in the model ID).

​
Configuration (minimal)

At minimum, set:
  • agents.defaults.workspace
  • channels.whatsapp.allowFrom (strongly recommended)
Next: Group Chats 🦞