LazyBoy2/docs/ACCEPTANCE.md

GrokBoy acceptance

P0 — streaming chat

• Cargo workspace grokboy-core + grokboy
• Env: GROKBOY_API_KEY / XAI_API_KEY / OPENAI_API_KEY
• Default base https://api.x.ai/v1
• grokboy chat streams assistant tokens
• cargo test passes without API key

P1 — tools + ReAct

• shell, list_dir, read_file, write_file
• Multi-step tool loop
• Sessions under ~/.grokboy/sessions/
• grokboy run / grokboy smoke
• Default model grok-4.6

P2 — completion contract

• report_done / report_blocked stop the loop with a clear verdict
• Loop guard (identical tool rounds ×3 → blocked)
• Context truncation (GROKBOY_CONTEXT_CHARS, default ~100k)
• Offline smoke/tests cover P2 without API key

P3 — browser

• Playwright DOM path (no screenshot-first)
• Tools: browser_navigate, browser_snapshot, browser_click, browser_type, browser_eval
• Thin Node helper under tools/playwright/ (JSONL / one-shot JSON)
• Fail closed with install hint when Node/Playwright/Chromium missing
• cargo test / grokboy smoke pass without Playwright browsers installed

P4 — human browser handoff

• Spec: docs/PRODUCT.md (north star, P0–P3, P4, non-goals)
• Tool browser_handoff (reason required, optional timeout_secs)
• Headed Chromium for handoff (handoff_prepare; relaunch if was headless)
• Bilingual (繁中 + EN) terminal instructions; Enter continue / abort / timeout → fail-closed
• After resume: DOM snapshot returned as tool result
• Env docs: GROKBOY_BROWSER_HEADED, GROKBOY_HANDOFF_AUTO
• Wired into tool defs + agent system prompt; fail-closed if no browser
• Offline tests / smoke without API key or interactive stdin (GROKBOY_HANDOFF_AUTO)
• README status table updated

P5 — interactive multi-turn agent

• grokboy agent REPL: read line → ReAct with tools → print verdict → save session
• /exit /quit leave; /session show id; empty line ignored
• --session <id> resume; auto-create + print session id when omitted
• Keep run one-shot; keep chat streaming no-tools
• Same tool-capable system prompt (AGENT_SYSTEM); Traditional Chinese welcome
• Docs: ACCEPTANCE P5, PRODUCT.md note, README commands/status
• Offline smoke/tests: agent parses / help lists it; multi-turn session plumbing without API
• cargo test green without API key

P6 — scenario playbooks + confirm-before-post

• Reusable pattern docs: docs/scenarios/README.md, docs/SCENARIO-TEMPLATE.md
• Template prompts: prompts/templates/phase-a.txt, phase-b.txt (placeholders)
• Example (not sole path): Shopee→Threads under docs/scenarios/examples/ + prompts/examples/
• Tool request_user_confirm (reason, optional prompt, optional timeout_secs)
• Bilingual banner; yes/y/Enter approve; no/abort deny; timeout = deny (fail-closed)
• Env: GROKBOY_CONFIRM_AUTO (fallback GROKBOY_HANDOFF_AUTO)
• AGENT_SYSTEM: never irreversible public social publish without explicit approval this turn or confirm approved; prefer draft → confirm → act; handoff for auth only
• Wired in tools.rs / confirm.rs; offline unit + smoke
• PRODUCT.md / README pointer to scenario playbooks
• cargo test / grokboy smoke green without API key

P7 slice — agent UX polish

• AGENT_SYSTEM: natural text for greetings/small talk/no-tools; report_done only for finished tool workflows; report_blocked when stuck
• Max-rounds: progress summary via final no-tools complete (fallback to bare message); offline injectable completer covered in unit tests
• Env GROKBOY_MAX_ROUNDS (default 12) used by CLI run/agent; keep DEFAULT_MAX_ROUNDS = 12
• CLI Blocked recovery hint (Traditional Chinese) for run and agent; agent REPL stays open on blocked
• Docs note in PRODUCT.md / ACCEPTANCE; cargo test / grokboy smoke green without API key