惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

博客园 - 司徒正美
大猫的无限游戏
大猫的无限游戏
腾讯CDC
J
Java Code Geeks
博客园 - 【当耐特】
Microsoft Azure Blog
Microsoft Azure Blog
V
Visual Studio Blog
人人都是产品经理
人人都是产品经理
博客园 - Franky
博客园 - 聂微东
阮一峰的网络日志
阮一峰的网络日志
美团技术团队
云风的 BLOG
云风的 BLOG
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
U
Unit 42
雷峰网
雷峰网
B
Blog RSS Feed
博客园_首页
量子位
F
Fortinet All Blogs
罗磊的独立博客
H
Hackread – Cybersecurity News, Data Breaches, AI and More
酷 壳 – CoolShell
酷 壳 – CoolShell
C
Check Point Blog

Show HN

GitHub - astefanutti/shaderbang: Shebang for Shaders Show HN: Generate Claude Code Workflows using Spec Driven Development approach Show HN: AI agents for UK GDAD PCF roles and their skills The Two Pillars: Mixer Mode and Meta-Software in the Reorganization of Software Work After AI GitHub - JaiCode08/teleport-env What 1,000+ Harness Experiments Taught Me About Self-Improving Agents Show HN: Liiists, a Markdown-first, iOS and CLI list app SwiperTab – Get this Extension for 🦊 Firefox (en-US) GitHub - kouhxp/fftext: Summarize, explain, fact-check, or translate any text, URL, or file. No GPU. No cloud. One command GitHub - sweetpad-dev/sweetpad: Develop Swift/iOS projects using VSCode GitHub - dogmaticdev/IRON: IRON a.k.a. Intermediate Representation Object Notation is a Interpreter/Database that is used to create Programming Languages. GitHub - sjhalani7/vaen: Package your AI coding harness into a portable .agent file, and share it across repos, teams, & the community without ever having to copy-paste instructions, skills, MCP config, or secrets. Show HN: Gandalf the Grader Show HN: Citadeld – replay any CI failure locally from a single file GitHub - tdortman/cuSBF: High-Performance GPU Super Bloom Filter coral-ai/claude-code-token-xray at main · Coral-Bricks-AI/coral-ai GitHub - ulyssestenn/funes: Funes is a Git-based framework for LLM-managed knowledge work: an AI Librarian ingests raw sources, builds an interlinked Markdown knowledge base, and uses it to produce cited reports, analyses, and other outputs. GitHub - ThatXliner/gah: Git Add Hunk, built for agents to use GitHub - harmont-dev/harmont-cli: Command-line client for the Harmont CI platform GitHub - brooksmcmillin/mcp-authflow: OAuth 2.0 Authorization Server framework for MCP servers GitHub - javaid-codes/audit-supply-chain-agents GitHub - amorey/gochan: A small library of common channel architectures for Go, inspired by Rust GitHub - arifozgun/OpenGem: Free, Open-Source AI API Gateway with Gemini, OpenAI & Anthropic Compatibility in 1 file GitHub - Pranesh950/BioPetals: 🌸 Run BIOxAI models at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading GitHub - cnguyen14/bounty-doctor: Diagnose a GitHub bounty issue before you waste hours: detects honeypot scam repos, AI-bot attempt swarms, and stale contests. Show HN: CoreMCP – MCP Server for On-Prem DBs Show HN: KittyHTML – Render HTML/CSS as an inline image in your terminal GitHub - bingud/filemat: Web-based file manager Show HN: TruthLens – Free multi-signal deepfake image detector GitHub - apexlocal-jz/claude-usage-tray: Windows system-tray app showing your Claude Code rate-limit usage at a glance. Zero deps, ~300 lines of PowerShell. Cross-IDE (works regardless of VS Code, Cursor, plain terminal).
Autonomy — 自律型 AI 代理人系統 | Bill Liu
Bill Liu · 2026-06-20 · via Show HN

AgentLoop · ConversationLoop Python 3.13 · 自律執行中

Beam Search 候選排序受管行動閘道程序技能學習為核心,
打造真正能自主感知、規劃、執行並持續進化的 AI 代理人系統。

探索核心模組 查看原始碼

0 內建 Procedure Skills

0 + 內建工具

0 支援 AI Provider

0 Beam Width

AgentLoop 是 Autonomy 的核心驅動器——一個「自導向任務迴圈」,不需要人工介入即可從目標出發,自主選擇工具、執行行動並從結果中學習。

每個 Turn 的執行流程:①從技能庫選取相關 ProcedureSkill,②由 LLM 提案 + RecipeEngine 產出候選動作路徑,③以 5 維評分排序(Beam Width=3),④透過 ActionGateway 受管執行,⑤評估結果並觸發學習。

終止條件明確:ACHIEVED(目標完成)、BLOCKED(無法繼續)、NO_CANDIDATESAPPROVAL_DENIEDMAX_STEPS_REACHED、或 FAILED——系統永遠不會陷入無限循環。

AgentLoop ActionGateway CandidateSelector LearningLoop CuratorDaemon

📚

disclose_procedure_skills() 選取相關技能

💡

generate_candidates() LLM 提案 + Recipe Engine

📊

rank_candidates() 5 維評分 · Beam Width=3

ActionGateway.execute_next() ApprovalPolicy 授權 · 受管執行

🧠

evaluate_outcome() + learn() 結果評估 · Recipe 學習

自導向代理人迴圈

AgentLoop.run(goal, max_steps=12) 啟動後完全自主執行,每步驟透過 run_turn() 推進。支援互動模式(interactive=True)與批次模式,所有事件均以 run_id 記錄於 AutonomyStore

受管行動閘道

所有工具呼叫必須通過 ActionGatewayApprovalPolicy 授權。每個動作帶有 RiskLevel(LOW / MEDIUM / HIGH)、expected_effectverification_plan,確保執行可控、可審計。

5 維 Beam Search 評分

CandidateSelector 以 evidence_strength(+0.30)、purpose(+0.10)、risk(−0.35)、side_effects(−0.20)、penalty(−1.0)五個維度計算總分,選出前 beam_width=3 名候選路徑送往執行。

混合結果評估

DeterministicOutcomeEvaluator 優先判斷執行失敗(直接 BLOCKED),成功但結果不確定時再交由 ModelAssistedOutcomeEvaluator 呼叫 LLM 語義判斷——最小化昂貴的模型呼叫次數。

程序技能庫(13 內建)

ProcedureSkillLibrary 內建 13 個 bundled skills,涵蓋 api-debugging、browser-navigation、code-editing、codebase-documentation、plan、procedure-skill-authoring、process-management、requesting-code-review、systematic-debugging、technical-spike、TDD、website-inspection、writing-plans,每步動態選取最相關技能載入。

後執行學習迴路

每次 Run 結束後,LearningLoop.review_run() 自動觸發:若結果為 ACHIEVED 且有 ≥2 個成功步驟,自動起草新 ProcedureSkillDraft(信心度 0.85),生成 LearningProposal 等待審核。

Action Recipe 引擎

RecipeEngine 監控每個成功動作的 SHA-256 指紋——當同一動作成功次數 ≥ candidate_threshold=2,自動升格為候選 Recipe,下次直接作為行動選項(source=action_skill)提出,無需 LLM 重新推理。

技能策展守護程序

每次 Run 結束後,CuratorDaemon 在後台執行緒(daemon=True)非同步啟動 SkillCurator.apply_auto_merges(),自動偵測並合併重複(信心 0.95)或子集(信心 0.90)技能,防止技能庫膨脹。

可擴展工具集目錄

TOOLSET_CATALOG 涵蓋 4 個預設啟用工具集:file、terminal、search、skills,加上 opt-in 的 project(git / JSON / YAML / 測試探索)與 browser(Playwright headless Chromium,11 工具)。規劃中:memory、delegate、cronjob、computer_use。

完整事件溯源稽核

AutonomyStore 以 Event Sourcing 模式記錄整個執行生命週期:run_started → skills_selected → candidates_ranked → action_selected → approval_decision → observation → outcome_evaluated → recipe_learned → learning_review → run_finished,完整可重播。

TUI 終端互動介面

autonomy tui 啟動豐富終端介面,提供 session 總覽面板、回合記錄、Action trail、compact/full 切換模式與 / 指令面板。自然語言直接流入 AgentLoop,UI 永遠不直接執行工具。

多 Provider 支援

支援 ollama(本地)與 8 個 OpenAI-compatible providers:openai-api、nvidia、openrouter、deepseek、xai、zai、kimi-coding、alibaba。提供 API 金鑰管理與 provider 設定,autonomy doctor 可檢查端點走通性。

score =

evidence_strength × +0.30 purpose × +0.10 risk × −0.35 side_effects × −0.20 penalty × −1.0

Entry · Interactive TUI / CLI run

SessionShell → ConversationLoop autonomy tui · 互動式對話

AgentLoop.run(goal) autonomy run · max_steps · UUID run_id

Turn Loop · run_turn()

ProcedureSkillLibrary index() · load_selected()

RecipeEngine candidates_for() · learn()

CandidateSelector beam_width=3 · 5-dim score

Execution · ActionGateway

ApprovalPolicy authorize(action, interactive)

ActionGateway execute_next() · risk assessment

ToolRegistry execute(action) → Observation

Post-Run · finish_run()

OutcomeEvaluator Deterministic + Model-assisted

LearningLoop review_run() · draft skill

CuratorDaemon async thread · auto-merge

Persistence · AutonomyStore

📁 runs

📡 events

⚗️ recipes

📚 skills

💡 proposals

🗂️ curator_log

TOOLSET_CATALOG — 工具集 & AI Providers

fileread / write / patch / diff / outline / symbol_search / syntax_check…

terminalshell.execute / process.start / poll / log / wait / stop

projectgit.status/diff/log · json.parse · yaml.parse · python.test_discover (opt-in)

browserPlaywright Chromium · navigate / snapshot / click / screenshot… (opt-in)

memoryPersistent Workspace Memory & Context Injection

delegate / cronjob / computer_useplanned

支援 AI Provider

Ollama本地部署 · http://127.0.0.1:11434/v1

OpenAI APIopenai-compatible · OPENAI_API_KEY

NVIDIAintegrate.api.nvidia.com · kimi-k2.6

OpenRouterOPENROUTER_API_KEY

xAI · Kimi · Alibaba · ZAIXAI_API_KEY · KIMI_API_KEY · DASHSCOPE_API_KEY…

api-debugging

browser-navigation

code-editing

codebase-documentation

plan

process-management

requesting-code-review

systematic-debugging

technical-spike

test-driven-development

website-inspection

writing-plans

每次執行 Turn 時,model.select_procedure_skills() 依可用工具名稱動態篩選並載入最相關技能;執行成功且達到門檻後,LearningLoop 將自動起草新技能並生成 LearningProposal(CANDIDATE 狀態)。

Bill Liu

Bill Liu

Autonomy 框架設計者 & AI Agent 系統工程師

設計並實作了 Autonomy 自律型 AI 代理人框架,核心貢獻包括:Beam Search 候選評分系統、受管行動閘道(ActionGateway)、後執行學習迴路(LearningLoop)、以及後台技能策展守護程序(CuratorDaemon)。

致力於讓 AI 代理人在真實工程環境中可靠運行——從程式碼編輯、瀏覽器自動化到系統除錯,透過可組合的技能庫與事件溯源架構實現完整可審計性。

LinkedIn GitHub

準備好打造你的自律型 AI 代理人了嗎?

Fork Autonomy,添加你的工具集、訓練專屬技能庫,讓代理人在你的領域自主作業。

前往 GitHub 聯絡作者