AI工作流管理
经 AI Skill Hub 精选评估，AI工作流管理 获评「推荐使用」。这款Agent工作流在功能完整性、社区活跃度和易用性方面表现出色，AI 评分 7.5 分，适合有一定技术背景的用户使用。
📚 深度解析
AI工作流管理 工作流的设计遵循"最小配置，最大复用"原则：核心逻辑已经封装好，用户只需配置自己的 API Key 和业务参数即可快速上手。工作流内置错误处理和重试机制，在网络波动或 API 限速等情况下仍能稳定运行，适合作为生产环境的自动化基础设施。
在实际部署时，建议先在测试环境中运行 3-5 次，验证各个环节的输出结果符合预期，再部署到生产环境。AI Skill Hub 评分 7.5 分，是同类 Agent 工作流中的精选推荐。
📋 工具概览
AI工作流管理 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排，将复杂的多步骤任务拆解为清晰的自动化流程，实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成，适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
📖 中文文档
AI工作流管理 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排，将复杂的多步骤任务拆解为清晰的自动化流程，实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成，适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
- 可视化 Agent 工作流编排，无需编写复杂代码
- 支持多步骤自动化任务链，实现全流程无人值守
- 与外部 API、数据库和第三方服务无缝集成
- 内置错误处理与自动重试机制，保障稳定运行
- 提供可复用的自动化模板，快速在同类场景部署
- 自动化日常重复性工作，将精力集中于创造性任务
- 构建数据采集 → 处理 → 输出的完整自动化管线
- 实现跨平台、跨系统的数据流转和业务协同
# 方式一：cargo install（推荐） cargo install visdom-ai-tracing # 方式二：从源码编译 git clone https://github.com/VirtusLab/visdom-ai-tracing cd visdom-ai-tracing cargo build --release # 二进制在 ./target/release/visdom-ai-tracing
- 访问 GitHub 仓库获取工作流文件
- 在对应平台（Dify / Flowise / Make 等）中找到「导入工作流」功能
- 上传工作流文件
- 按照提示配置必要的环境变量和 API Key
- 运行测试确认流程正常后投入使用
# 查看帮助 visdom-ai-tracing --help # 基本运行 visdom-ai-tracing [options] <input> # 详细使用说明请查阅文档 # https://github.com/VirtusLab/visdom-ai-tracing
# visdom-ai-tracing 配置说明 # 查看配置选项 visdom-ai-tracing --config-example > config.yml # 常见配置项 # output_dir: ./output # log_level: info # workers: 4 # 环境变量（覆盖配置文件） export VISDOM_AI_TRACING_CONFIG="/path/to/config.yml"
Visdom Trace
Research Preview — This project is under heavy development. APIs, configuration, and features may change significantly between releases.
AI code governance platform for enterprises. Captures what AI coding agents do in your repos — which files they touch, how many tokens they burn, what tools they call, what percentage of code is AI-generated — then enforces policies and produces tamper-evident audit trails for regulatory compliance.
Built for financial institutions and regulated industries where AI-generated code needs the same audit rigor as human-written code.
Prerequisites
- Rust stable toolchain (install via rustup)
- PostgreSQL 14+ (or Docker) with the pgvector extension enabled (
CREATE EXTENSION vector;). pgvector is available out of the box on all major managed providers (AWS RDS, GCP Cloud SQL, Azure Database for PostgreSQL, Supabase, Neon). - Node.js 20+ and pnpm (for the web dashboard)
- Docker & Docker Compose (for containerized deployment)
Generate an encryption key (required — encrypts per-org signing keys in the DB)
export TRACEVAULT_ENCRYPTION_KEY=$(openssl rand -base64 32)
1. Build from source
cargo build --releaseBinaries land in target/release/: - tracevault (CLI) - tracevault-server
2. Start with Docker Compose (recommended)
```sh
Database credentials (defaults match docker-compose)
export DATABASE_URL=postgres://tracevault:tracevault@localhost:5432/tracevault
In Docker/CI/SSH-without-X11 the browser open is skipped automatically —
1. Install Claude Code
npm install -g @anthropic-ai/claude-code4. Install project-local MCP tools
This repo ships project-local MCP servers that Claude Code uses to satisfy pre-push policies.
Step 1 — Create your local .mcp.json:
cp .mcp.json.example .mcp.json.mcp.json is gitignored so it stays local. The example includes the three project servers needed for policy compliance. Add any private servers (deploy tools, DB access, etc.) to your local copy only.
Step 2 — Install dependencies for the required policy tools:
npm install --prefix tools/cargo-mcp
npm install --prefix tools/review-mcpWhat each server provides:
tools/cargo-mcp/ — required by pre-push policies: - mcp__cargo__cargo_fmt — runs cargo fmt. Must be called before committing. - mcp__cargo__cargo_check — runs cargo clippy then cargo test. Must be called and succeed before pushing. - mcp__cargo__cargo_audit — runs cargo audit. Required when Cargo.lock changes.
tools/review-mcp/ — required by the validation-scoped self-review policy: - mcp__review__agent_review — assembles the diff + full context of touched files and returns a review prompt. Call this inside a verification phase before pushing to Rust files.
Optional — session history queries:
npm install --prefix integrations/tracevault-mcpInstalls two tools:
ask_tracevault— lets agents query indexed session history in natural language ("Why was this refactored?", "What sessions touched the auth service?"). Not required by any policy. Needstracevault loginonce.agent_policies— returns the rendered policy instructions for the current repo (same output astracevault agent-policies). Call this at session start so the agent's behaviour matches the configured policies.
Claude Code picks up all servers from .mcp.json automatically on next session start. No further configuration needed.
4. Analyze — Usage Analytics
Understand how AI is used across your team:
- Token usage trends over time, per model, per author
- Model distribution — which models are used most and where
- Cost tracking — estimated cost breakdown by model and team member
- Cache savings — how much prompt caching saves
- AI attribution — what percentage of your codebase is AI-generated
- Author activity — commits, tokens, and cost per developer
All available through the web dashboard with filterable time ranges and drill-down views.
Quick Start
server_url = "https://your-server.example.com"
Policy types you can configure
In the Visdom Trace dashboard, under Repos → [your repo] → Policies, you can add:
| Condition | What it checks | Use case |
|---|---|---|
RequiredToolCall | A specific tool was called during the session | Mandate code formatters, linters, test runners |
ConditionalToolCall | A tool was called when specific files changed | Require security scan only when Cargo.lock changes |
AiPercentageThreshold | AI-authored lines exceed a threshold | Warn when AI writes > 80% of a module |
TokenBudget | Token or cost usage exceeds a limit | Cap AI spend per session |
Actions: Block push (exit non-zero, prevents git push) or Warn (logs but allows).
Scope: Session (evaluate all tool calls in the session) or Verification phase (evaluate only tools called after tracevault verify-start).
Configuration
CLI config (`.tracevault/config.toml`)
```toml agent = "claude-code"
Server environment variables
| Variable | Default | Description |
|---|---|---|
DATABASE_URL | postgres://tracevault:tracevault@localhost:5432/tracevault | PostgreSQL connection string |
HOST | 0.0.0.0 | Bind address |
PORT | 3000 | Bind port |
CORS_ORIGIN | _(permissive)_ | Allowed CORS origin for web dashboard |
TRACEVAULT_ENCRYPTION_KEY | — | **Required.** AES-256 encryption key (base64-encoded 32 bytes) for encrypting per-org signing keys, deploy keys, and API keys at rest. Generate with openssl rand -base64 32. |
TRACEVAULT_REPOS_DIR | ./data/repos | Directory for cloned git repos (used by code browser) |
TRACEVAULT_LLM_PROVIDER | — | LLM provider for story generation (anthropic or openai) |
TRACEVAULT_LLM_API_KEY | — | API key for the LLM provider |
TRACEVAULT_LLM_MODEL | — | LLM model name (defaults: Claude Sonnet 4 for Anthropic, GPT-4o for OpenAI) |
TRACEVAULT_LLM_BASE_URL | — | Custom LLM endpoint URL |
RUST_LOG | — | Log level (e.g. info, debug) |
Web dashboard environment variables
| Variable | Default | Description |
|---|---|---|
PUBLIC_API_URL | http://localhost:3000 | Backend server URL the SvelteKit proxy forwards API calls to. In Docker Compose this is set to http://server:3000 automatically. |
CLI Commands
| Command | Description |
|---|---|
tracevault init [--server-url URL] [--claude-settings shared\|local] | Initialize Visdom Trace in current repo, install pre-push hook and Claude Code hooks. --claude-settings chooses between .claude/settings.json (default) and .claude/settings.local.json; prompts interactively if omitted on a TTY |
tracevault login --server-url URL [--no-browser] | Authenticate via device auth flow. Prints the URL and opens a browser when possible; --no-browser (or a headless env) skips the auto-open. |
tracevault logout | Clear local credentials |
tracevault stream --event <type> | Handle a Claude Code hook event (reads JSON from stdin) and stream it to the server |
tracevault sync | Sync repo metadata with the server |
tracevault check | Evaluate policies against server rules, exit non-zero if blocked |
tracevault verify-start [--session-id ID] | Open a verification phase. Call this when work is complete and you are ready to run pre-push validation tools. Verification-phase-scoped policies only evaluate tools called after this point. Calling it again invalidates the previous phase. |
tracevault stats | Show local session statistics |
tracevault verify | Verify commits are registered and sealed on the server (--commits or --range) |
tracevault status | Show current session status (not yet implemented) |
api_key = "your-api-key"
```
Run integration tests
source .env.test && cargo test -p tracevault-server --test '*'
高质量的AI工作流管理平台
- 需要 visdom-ai-tracing 解决具体问题的开发者与运营人员
- 先在测试环境跑通最小用例，再接入生产数据
- API key 直接提交到 git 仓库（请用 .env 并加入 .gitignore）
- 云端托管：可放在 Vercel / Railway / Fly.io 等 PaaS 平台
⚡ 核心功能
- 可视化 Agent 工作流编排，无需编写复杂代码
- 支持多步骤自动化任务链，实现全流程无人值守
- 与外部 API、数据库和第三方服务无缝集成
- 内置错误处理与自动重试机制，保障稳定运行
- 提供可复用的自动化模板，快速在同类场景部署
- 需要 visdom-ai-tracing 解决具体问题的开发者与运营人员
- 先在测试环境跑通最小用例，再接入生产数据
- API key 直接提交到 git 仓库（请用 .env 并加入 .gitignore）
👥 适合人群
🎯 使用场景
- 自动化日常重复性工作，将精力集中于创造性任务
- 构建数据采集 → 处理 → 输出的完整自动化管线
- 实现跨平台、跨系统的数据流转和业务协同
⚖️ 优点与不足
- +Apache-2.0 协议，可免费商用
- +大幅减少重复性人工操作
- +可视化流程，清晰直观
- +可扩展性强，支持复杂场景
- −初始配置和调试需投入一定时间
- −强依赖外部服务的稳定性
- −复杂场景需具备一定技术基础
AI Skill Hub 为第三方内容聚合平台，本页面信息基于公开数据整理，不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后，再部署至生产环境，并做好必要的安全评估。
✅ Apache 2.0 — 宽松开源协议，可商用，需保留版权声明和 NOTICE 文件，含专利授权条款。
🔗 相关工具推荐
❓ 常见问题 FAQ
AI Skill Hub 点评：AI工作流管理 的核心功能完整，质量良好。对于自动化工程师和运维人员来说，这是一个值得纳入个人工具库的选择。建议先在非生产环境试用，再逐步推广。
| 原始名称 | visdom-ai-tracing |
| 原始描述 | 开源AI工作流：AI code governance platform for enterprises. Captures, enforces policies, audits。⭐6 · Rust |
| Topics | AI工作流Rust |
| GitHub | https://github.com/VirtusLab/visdom-ai-tracing |
| License | Apache-2.0 |
| 语言 | Rust |
收录时间：2026-06-11 · 更新时间：2026-06-11 · License：Apache-2.0 · AI Skill Hub 不对第三方内容的准确性作法律背书。