📄 工具详情 ⚙️ 安装教程 📚 使用教程
能力标签
🔌 MCP🔄 工作流💻 CLI🔗 REST API🎙 STT🧠 Claude
AI工作流助手
⚙️
Agent工作流

AI工作流助手

基于 TypeScript · 无代码搭建完整 AI 自动化流程
英文名：assist
⭐ 6 Stars 🍴 1 Forks 💻 TypeScript 📄 未公布协议 🏷 AI 7.5分
7.5AI 综合评分
LLM工作流自动化
⚙️ 配置说明 🔍 查看原项目
✦ AI Skill Hub 推荐

AI Skill Hub 推荐使用：AI工作流助手 是一款优质的Agent工作流。AI 综合评分 7.5 分，在同类工具中表现稳健。如果你正在寻找可靠的Agent工作流解决方案，这是一个值得深入了解的选择。

📚 深度解析

AI工作流助手 是一套完整的 AI Agent 自动化工作流方案。随着 AI 能力的不断提升，基于 Agent 的自动化工作流正在成为提升个人和团队效率的核心方式。区别于传统的 RPA 自动化（模拟鼠标键盘操作），AI Agent 工作流通过理解任务意图、动态规划执行路径，能够处理更复杂的非结构化任务。

AI工作流助手 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工作流助手 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。

GitHub Stars
⭐ 6
开发语言
TypeScript
支持平台
Windows / macOS / Linux
维护状态
轻量级项目，按需更新
开源协议
未公布
AI 综合评分
7.5 分
工具类型
Agent工作流
Forks
1

📖 中文文档

以下内容由 AI Skill Hub 根据项目信息自动整理，如需查看完整原始文档请访问底部「原始来源」。

AI工作流助手 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。

📌 核心特色
  • 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
🎯 主要使用场景
  • 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
以下安装命令基于项目开发语言和类型自动生成，实际以官方 README 为准。
安装命令
# 方式一：npm 全局安装
npm install -g assist

# 方式二：npx 直接运行（无需安装）
npx assist --help

# 方式三：项目依赖安装
npm install assist

# 方式四：从源码运行
git clone https://github.com/staff0rd/assist
cd assist
npm install
npm start
📋 安装步骤说明
  1. 访问 GitHub 仓库获取工作流文件
  2. 在对应平台（Dify / Flowise / Make 等）中找到「导入工作流」功能
  3. 上传工作流文件
  4. 按照提示配置必要的环境变量和 API Key
  5. 运行测试确认流程正常后投入使用
以下用法示例由 AI Skill Hub 整理，涵盖最常见的使用场景。
常用命令 / 代码示例
# 命令行使用
assist --help

# 基本用法
assist [options] <input>

# Node.js 代码中使用
const assist = require('assist');

const result = await assist.run(options);
console.log(result);
以下配置示例基于典型使用场景生成，具体参数请参照官方文档调整。
配置示例
# assist 配置说明
# 查看配置选项
assist --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 ASSIST_CONFIG="/path/to/config.yml"
📑 README 深度解析 真实文档 完整度 38/100 查看 GitHub 原文 →
以下内容由系统直接从 GitHub README 解析整理，保留代码块、表格与列表结构。

assist

A CLI tool for enforcing determinism in LLM development workflow automation.

See devlog for latest features.

Install dependencies

npm install

Installation

You can install assist globally using npm:

npm install -g @staff0rd/assist
assist sync

Build the project

npm run build

Install globally

npm install -g . ```

After installation, the assist command will be available globally. You can also use the shorter ast alias.

CLI Commands

- assist activity [--since <date>] - Chart GitHub commit activity per day (defaults to last 30 days) - assist init - Initialize project with VS Code and verify configurations - assist new vite - Initialize a new Vite React TypeScript project - assist new cli - Initialize a new tsup CLI project - assist sync - Copy command files to ~/.claude/commands - assist commit status - Show git status and diff - assist commit <message> - Commit staged changes with validation - assist commit <message> [files...] - Stage files and create a git commit with validation - assist prs - List pull requests for the current repository - assist prs create --title <title> --body <body> - Create a pull request via gh pr create - assist prs list-comments - List all comments on the current branch's pull request - assist prs fixed <comment-id> <sha> - Reply with commit link and resolve thread - assist prs wontfix <comment-id> <reason> - Reply with reason and resolve thread - assist prs comment <path> <line> <body> - Add a line comment to the pending review - assist review [number] [options] - Run Claude and Codex in parallel to review the open PR for the current branch. The diff is fetched from GitHub (base SHA → head SHA via gh pr diff), so stale local base branches don't pollute the review; fails fast if no PR is open. By default, prompts before posting line-bound comments and then prompts again to submit the pending review (defaulting to no). Findings whose lines fall outside the diff are skipped with a warning rather than being silently dropped by GitHub. Cached claude.md / codex.md / synthesis.md are reused when present; if any reviewer is re-run, the synthesis is invalidated. - [number] - Run gh pr checkout <number> first, then review that PR's branch. If the checkout fails (dirty working tree, unknown PR number), the review aborts - --no-prompt - Skip all confirmations - --submit - Default the submit prompt to yes (or auto-submit when combined with --no-prompt) - --force - Clear all cached files and re-run every phase - --refine - Skip posting; launch an interactive Claude session that walks through synthesis.md and edits it in place. A subsequent assist review reuses the refined file and posts only the surviving findings - --apply - Skip posting; launch an interactive Claude session that walks through each finding asking apply/skip. Applied findings are fixed in the working tree (unstaged) and removed from synthesis.md; skipped findings stay so a subsequent assist review posts them. Cannot be combined with --refine - --backlog - Skip posting; launch an interactive Claude session running /bug that files all findings (including already-raised) as a single bug backlog item with one phase per finding. synthesis.md is left untouched; --submit is ignored. Cannot be combined with --refine or --apply - --verbose - Disable the stacked-spinner UI and fall back to per-line log output. Non-TTY environments (CI) automatically use this mode - assist github commits <org> - Report commit activity across a GitHub organisation over the last 30 days: repos ranked by commits to their default branch, top committers, and a per-repo author breakdown (empty repos are skipped; commits with no linked GitHub account fall back to the raw author name) - --since <date> - Start of the window as YYYY-MM-DD instead of the default 30 days ago - --top <n> - Only report the top n repos by commit count; committers and the author breakdown then cover those repos only (also caps the per-repo author queries, which speeds up large orgs) - --json - Output all three views as structured JSON instead of tables - assist news - Start the news web UI showing latest RSS feed items (same as news web) - assist news add [url] - Add an RSS feed URL to the config - assist news web [-p, --port <number>] - Start a web view of the news feeds (default port 3001) Backlog data is stored in a global Postgres database (shared across all repos, scoped per repository by git origin), so a connection string is required. Set it via the ASSIST_BACKLOG_DATABASE_URL environment variable or the backlog.databaseUrl key in assist.yml; the environment variable takes precedence. Without one, every assist backlog command exits with a setup message. (There is no SQLite/JSONL fallback.) Commands default to the current repository's items; pass --all-repos to span every repository.

The first backlog command in a repository that still has a local .assist/backlog.jsonl automatically migrates it into Postgres — but only as a one-time bootstrap into an empty origin. If Postgres has no items for the repo's origin yet, it runs git pull (best-effort) to fetch the latest committed copy, imports every item under the origin with fresh global IDs (rewriting links to other items), and verifies the result. If Postgres already has items for that origin (a prior run, another clone, or a pre-seeded database), the import is skipped to avoid creating duplicates. Either way the local .assist/backlog.jsonl and .assist/backlog.db are renamed to *.bak, so the migration never re-runs and a local copy is retained.

  • assist backlog [--dir <path>] - Open the backlog tab in the web dashboard (same as backlog web). --dir overrides the directory used to resolve the current repository's git origin
  • assist backlog init - Create an empty backlog
  • assist backlog list [--status <type>] [-a, --all] [--all-repos] [-v] - List backlog items with status icons (alias: ls). Defaults to the current repository's todo/in-progress items; --all includes done/wontdo, --all-repos lists items across all repositories. Also available as the top-level shortcut assist list / assist ls with the same flags
  • assist backlog add - Add a new backlog item interactively (prompts for type: story/bug)
  • assist backlog add --name <n> --type <t> --desc <d> --ac <criterion...> - Add a backlog item from CLI options (used by /draft)
  • assist backlog add-phase <id> <name> --task <t...> [--manual-check <c...>] [--position <pos>] - Add a phase (appends by default; --position inserts at a 1-indexed position)
  • assist backlog update-field <id> [--name <n>] [--desc <d>] [--type <t>] [--ac <criterion...>] - Update fields on a backlog item
  • assist backlog update-field <id> [--add-ac <text>] [--edit-ac <n> <text>] [--remove-ac <n>] - Granular acceptance-criteria edits using 1-based indices matching backlog show: --add-ac appends (repeatable), --edit-ac replaces criterion n in place, --remove-ac deletes criterion n and renumbers the rest (cannot be combined with the whole-list --ac)
  • assist backlog update-phase <id> <phase> [--name <n>] [--task <t...>] [--manual-check <c...>] - Modify a plan phase (name, tasks, or manual checks)
  • assist backlog update-phase <id> <phase> [--add-task <text>] [--edit-task <n> <text>] [--remove-task <n>] [--add-check <text>] [--edit-check <n> <text>] [--remove-check <n>] - Granular task and manual-check edits using 1-based indices matching backlog show: --add-* appends (repeatable), --edit-* replaces entry n in place, --remove-* deletes entry n and renumbers the rest (task ops cannot be combined with --task; check ops cannot be combined with --manual-check)
  • assist backlog remove-phase <id> <phase> - Remove a plan phase from a backlog item
  • assist backlog next [id] [--once] - Pick and run the next backlog item, or open /draft if none remain; pass an id to run that item first, then continue chaining; --once exits after the first completed item run instead of prompting for another
  • assist backlog refine [id] [--once] - Alias for refine
  • assist backlog start <id> - Set a backlog item to in-progress
  • assist backlog stop - Revert all in-progress backlog items to todo and reset their phase to 1
  • assist backlog done <id> - Set a backlog item to done
  • assist backlog wontdo <id> [reason] - Set a backlog item to won't do
  • assist backlog delete <id> - Delete a backlog item
  • assist backlog show <id> - Display full detail for a backlog item (alias: view)
  • assist backlog plan <id> - Display the phased plan for a backlog item
  • assist backlog comment <id> <text> - Add a comment to a backlog item
  • assist backlog comments <id> - List comments and summaries for a backlog item
  • assist backlog delete-comment <id> <comment-id> - Delete a comment from a backlog item (summaries cannot be deleted)
  • assist backlog phase-done <id> <phase> <summary> - Signal that a plan phase is complete with a summary (used by orchestrator)
  • assist backlog rewind <id> <phase> --reason <reason> - Rewind a backlog item to an earlier phase, setting status to in-progress
  • assist backlog run <id> - Run a backlog item's plan phase-by-phase with Claude; --resume-session <id> resumes an interrupted Claude session for the current phase (used by the sessions daemon when it restarts a running item)
  • assist backlog export [file] - Export the entire backlog database (all tables, all repos) to a file, or stdout if omitted
  • assist backlog import [file] - Restore the entire backlog database from a dump (file or stdin), faithfully replacing all data; prompts for confirmation (use -y, --yes to skip; required when reading from stdin)
  • assist backlog move-repo <old-origin> [new-origin] - Retag all items from one origin to another after a repo rename; the new origin defaults to the current repo's remote, both accept URL or git@ forms, and a bare repo name works for the old origin when unambiguous. Prompts for confirmation (use -y, --yes to skip)
  • assist backlog web [-p, --port <number>] [--no-open] - Open the backlog tab in the web dashboard (default port 3100); --no-open skips opening a browser on startup
  • assist roam auth - Authenticate with Roam via OAuth (opens browser, saves tokens to ~/.assist.yml)
  • assist roam show-claude-code-icon - Forward Claude Code hook activity to Roam local API
  • assist run <name> [params...] - Run a configured command from assist.yml (positional params are matched to params config; supports pre array of commands to run first). If <name> is purely numeric and matches no configured command, it is treated as an alias for assist backlog run <name> and forwards --write/--no-write/-w.
  • assist run add - Add a new run configuration to assist.yml and create a Claude command file
  • assist run link <path> --prefix <prefix> - Link run configurations from another project's assist.yml
  • assist run remove <name> - Remove a run configuration from assist.yml and delete its Claude command file
  • assist config set <key> <value> - Set a config value (e.g. commit.push true)
  • assist config get <key> - Get a config value
  • assist config list - List all config values
  • assist verify - Run all verify:* commands in parallel (from run configs in assist.yml and scripts in package.json)
  • assist verify all - Run all checks, ignoring diff-based filters
  • assist verify init - Add verify scripts to a project (writes to assist.yml by default; pass --package-json to write to package.json scripts instead)
  • assist verify hardcoded-colors - Check for hardcoded hex colors in src/ (supports hardcodedColors.ignore globs in config)
  • assist verify comment-policy - Flag comments added on changed lines (staged + unstaged) unless they carry a justification marker; supports commentPolicy.markers and commentPolicy.ignore globs in config
  • assist lint [-f, --fix] - Run lint checks for conventions not enforced by biomejs (use -f to auto-fix)
  • assist lint init - Initialize Biome with standard linter config
  • assist refactor check [pattern] - Check for files that exceed the maximum line count
  • assist refactor ignore <file> - Add a file to the refactor ignore list
  • assist refactor rename file <source> <destination> - Rename/move a TypeScript file and update all imports (dry-run by default, use --apply to execute)
  • assist refactor rename symbol <file> <oldName> <newName> - Rename a variable, function, class, or type across the project (dry-run by default, use --apply to execute)
  • assist refactor extract <file> <functionName> <destination> - Extract a function and its private dependencies to a new file (dry-run by default, use --apply to execute)
  • assist refactor restructure [pattern] - Analyze import graph and restructure tightly-coupled files into nested directories
  • assist devlog list - Group git commits by date
  • assist devlog next - Show commits for the day after the last versioned entry
  • assist devlog repos - Show which github.com/staff0rd repos are missing devlog entries
  • assist devlog skip <date> - Add a date to the skip list
  • assist devlog version - Show current repo name and version info
  • assist cli-hook - PreToolUse hook for auto-approving CLI commands (reads from allowed.cli-reads and allowed.cli-writes, also auto-approves read-only gh api calls). Supports compound commands (|, &&, ||, ;) by checking each sub-command independently.
  • assist cli-hook add <cli> - Discover a CLI's commands and auto-permit read-only ones
  • assist cli-hook check <command> [--tool <tool>] - Check whether a command would be auto-approved by cli-hook (tool defaults to Bash)
  • assist cli-hook deny - List all deny rules
  • assist cli-hook deny add <pattern> <message> - Add a deny rule for a command pattern
  • assist cli-hook deny remove <pattern> - Remove a deny rule by pattern
  • assist update - Update assist to the latest version and sync commands
  • assist vscode init - Add VS Code configuration files
  • assist deploy init - Initialize Netlify project and configure deployment
  • assist deploy redirect - Add trailing slash redirect script to index.html
  • assist notify - Show desktop notification from JSON stdin (supports macOS, Windows, WSL)
  • assist status-line - Format Claude Code status line from JSON stdin
  • assist dotnet inspect [sln] - Run JetBrains inspections on changed .cs files to find dead code
  • assist dotnet inspect [sln] --scope all - Inspect the full solution
  • assist dotnet inspect [sln] --scope base:<ref> - Inspect all .cs files changed since diverging from a base ref (e.g. --scope base:main for a full PR)
  • assist dotnet inspect [sln] --scope commit:<ref> - Inspect .cs files changed in a specific commit
  • assist dotnet inspect [sln] --only <ids...> - Show only the specified issue type IDs (e.g. --only CommentedCode)
  • assist dotnet inspect [sln] --suppress <ids...> - Suppress specific issue type IDs on the command line
  • assist dotnet inspect [sln] --roslyn - Use Roslyn analyzers via msbuild instead of JetBrains
  • assist dotnet inspect [sln] --swea - Enable solution-wide error analysis (slower but more thorough)
  • assist dotnet check-locks - Check if build output files are locked by a debugger
  • assist dotnet deps <csproj> - Show .csproj project dependency tree and solution membership
  • assist dotnet in-sln <csproj> - Check whether a .csproj is referenced by any .sln file
  • assist jira auth - Authenticate with Jira via API token (saves site/email to ~/.assist/jira.json)
  • assist jira ac <issue-key> - Print acceptance criteria for a Jira issue
  • assist jira view <issue-key> - Print the title and description of a Jira issue
  • Note: Claude is wired to the MCP Atlassian server (mcp__claude_ai_Atlassian__getJiraIssue) for fetching Jira context, so the /jira slash command and Jira-key mentions go through MCP. These assist jira CLI commands remain for direct human use.
  • assist ravendb auth add - Add a new RavenDB connection (prompts for name, URL, database, op:// secret reference)
  • assist ravendb auth list - List configured RavenDB connections
  • assist ravendb auth remove <name> - Remove a configured connection
  • assist ravendb set-connection <name> - Set the default connection for query/collections commands
  • assist ravendb query [connection] [collection] - Query a RavenDB collection (outputs JSON to stdout)
  • assist ravendb query [connection] [collection] --page-size <n> --sort <field> --query <lucene> --limit <n> - Query with options
  • assist ravendb collections [connection] - List collections and document counts in a database
  • assist seq auth add - Add a new Seq connection (prompts for name, URL, API token)
  • assist seq auth list - List configured Seq connections
  • assist seq auth remove <name> - Remove a configured connection
  • assist seq set-connection <name> - Set the default Seq connection
  • assist seq query <filter> - Query Seq events with a filter expression
  • assist seq query <filter> -c <connection> - Query using a specific connection
  • assist seq query <filter> --json - Output raw JSON
  • assist seq query <filter> -n <count> - Fetch a specific number of events (default 50)
  • assist seq query <filter> --from <date> - Start of query window (UTC date or relative e.g. 5m, 1h, 2d)
  • assist seq query <filter> --to <date> - End of query window (UTC date or relative e.g. 5m, 1h, 2d)
  • assist sql auth add - Add a new MSSQL connection (prompts for name, server, port, user, password, database)
  • assist sql auth list - List configured SQL connections
  • assist sql auth remove <name> - Remove a configured connection
  • assist sql set-connection <name> - Set the default SQL connection
  • assist sql query "<sql>" [connection] - Execute a read-only SQL statement and print results in table format (rejects INSERT/UPDATE/DELETE/DROP/CREATE/ALTER/TRUNCATE/MERGE/GRANT/REVOKE/EXEC)
  • assist sql mutate "<sql>" [connection] - Execute a mutating SQL statement and print rows affected (rejects non-mutating statements like pure SELECTs)
  • assist sql tables [connection] - List tables in the connected database (via INFORMATION_SCHEMA.TABLES)
  • assist sql columns <table> [connection] - List columns for a table (use schema.table for non-default schema; via INFORMATION_SCHEMA.COLUMNS)
  • assist screenshot <process> - Capture a screenshot of a running application window (e.g. assist screenshot notepad). Output directory is configurable via screenshot.outputDir (default ./screenshots)
  • assist handover archive [--suffix <s>] - Archive the current .assist/HANDOVER.md to .assist/handovers/archive/<ISO-ts>[-<suffix>].md. Prints the archive path; no-op when no handover exists
  • assist handover summarise <jsonl> - Print a one-line summary of a session JSONL via claude -p --model haiku. Filters sdk-cli-only transcripts and sets a recursion-guard env var so the inner SessionStart hook short-circuits
  • assist handover load - SessionStart hook entry point: reads {cwd, session_id} from stdin, and only when .assist/HANDOVER.md exists archives it and emits { hookSpecificOutput: { hookEventName: "SessionStart", additionalContext }, systemMessage }. Emits nothing when no handover exists (use /recall to pull prior-session context on demand). Honors _CLAUDE_HOOK_SUMMARISE_RUNNING so nested invocations short-circuit silently
  • assist mermaid export [file.md] - Render each fenced mermaid block to <stem>-<index>.svg via Kroki. With no file, scans *.md in the current directory (non-recursive). Use --out <dir> to override the output directory. Use --index <n> to render only the nth mermaid block (1-based; requires a file argument). Endpoint is configurable via mermaid.krokiUrl (default https://kroki.io).
  • assist prompts - Show top 10 denied tool calls by frequency with count and repo breakdown
  • assist coverage - Print global statement coverage percentage
  • assist complexity <pattern> - Analyze a file (all metrics if single match, maintainability if multiple)
  • assist complexity cyclomatic [pattern] - Calculate cyclomatic complexity per function
  • assist complexity halstead [pattern] - Calculate Halstead metrics per function
  • assist complexity maintainability [pattern] - Calculate maintainability index per file (--ignore <glob>, repeatable, excludes extra files on top of complexity.ignore)
  • assist complexity sloc [pattern] - Count source lines of code per file
  • assist transcript configure - Configure transcript directories
  • assist transcript format - Convert VTT files to formatted markdown transcripts
  • assist transcript summarise - List transcripts that do not have summaries
  • assist voice setup - Download required voice models (VAD, STT)
  • assist voice start - Start the voice daemon (always-on, listens for wake word)
  • assist voice start --foreground - Start in foreground for debugging
  • assist voice stop - Stop the voice daemon
  • assist voice status - Check voice daemon status and recent events
  • assist voice devices - List available audio input devices
  • assist voice logs [-n <count>] - Show recent voice daemon log entries
  • assist sessions - Start the web dashboard (same as sessions web)
  • assist sessions web [-p, --port <number>] [--no-open] - Start the web dashboard with Sessions and Backlog tabs, xterm.js terminals with clickable http(s) links (default port 3100); --no-open skips opening a browser on startup; press Ctrl+R in the foreground terminal for an in-terminal restart menu (Restart daemon, Restart webserver, Restart both); Restart webserver re-execs the foreground process (passing --no-open so no browser pops on restart) so the connected browser auto-reconnects
  • assist sessions summarise [-f, --force] [-n, --limit <count>] - Generate one-line summaries for unsummarised Claude sessions (force re-generates all; limit caps how many to process)
  • assist daemon run - Run the sessions daemon in the foreground (normally auto-spawned detached by assist sessions)
  • assist daemon status - Show sessions daemon status, live sessions, and any stray daemon processes or stolen socket
  • assist daemon stop - Stop the sessions daemon; running claude sessions resume on next start
  • assist daemon restart - Restart the sessions daemon, resuming previously running claude sessions

Web sessions are owned by a long-lived daemon process, not the web server: the server is a thin client that relays WebSocket traffic to the daemon over a local IPC socket (unix domain socket at ~/.assist/daemon/daemon.sock; named pipe \\.\pipe\assist-sessions-daemon on Windows). Restarting the web server leaves sessions running with scrollback intact. The daemon logs to ~/.assist/daemon/daemon.log (timestamped lines tagged with the daemon's pid, including why it spawned and which sessions it restored) and auto-exits once no sessions remain and no client has been connected for 60 seconds (it is respawned on demand by the web server). Daemon spawning is arbitrated by an O_EXCL lockfile so racing clients start at most one daemon; sessions are only restored after the daemon owns the IPC socket, and a daemon that loses ownership of daemon.pid shuts down its sessions and exits rather than running orphaned.

When iterating on assist itself: web server changes only need the assist sessions process restarted — sessions survive. Daemon/session-core changes need assist daemon restart to load the new code; this kills the PTYs, then claude sessions — including assist sessions that wrap claude, like assist draft — are auto-respawned via claude --resume with scrollback starting fresh, while run sessions (and assist sessions whose claude sessionId was never discovered) reappear as not-restored tiles that can be retried. - assist next [id] [--once] - Alias for backlog next [id]; --once exits after the first completed item run instead of prompting for another - assist draft [description] [--once] (alias: feat) - Launch Claude in /draft mode, chain into next on /next signal; an optional description is forwarded as /draft <description>; --once exits when the done signal arrives after the initial draft completes - assist bug [description] [--once] - Launch Claude in /bug mode, chain into next on /next signal; an optional description is forwarded as /bug <description>; --once exits when the done signal arrives after the initial bug report completes - assist refine [id] [--once] - Launch Claude in /refine mode to refine a backlog item; prompts for selection when no id given; --once exits when the done signal arrives after refinement completes - assist review-comments [number] - Launch Claude in /review-comments mode to process PR review comments (single session, no chaining); when a PR number is supplied, checks out that PR via gh pr checkout first - assist signal next [id] - Write a next signal to chain into assist next; when id is supplied, the parent launcher runs that backlog item directly - assist signal done [id] - Write a done signal marking the session's initial task complete; an optional id surfaces the backlog item the session created onto its session card; --once launch sessions exit when it arrives, plain sessions ignore it

When commit.pull is enabled in config, assist draft, assist bug, assist refine, assist next, and assist backlog run run git pull --ff-only before doing anything else; if the pull fails the command aborts. assist next pulls once per invocation, not per item in its loop.

🎯 aiskill88 AI 点评 A 级 2026-06-09

assist是一个有用的LLM开发工作流自动化工具

📚 实用指南（长尾问题）
适合谁
  • 需要 assist 解决具体问题的开发者与运营人员
最佳实践
  • 先在测试环境跑通最小用例，再接入生产数据
常见错误
  • API key 直接提交到 git 仓库（请用 .env 并加入 .gitignore）
部署方案
  • 云端托管：可放在 Vercel / Railway / Fly.io 等 PaaS 平台
相关搜索
assist 中文教程assist 安装报错怎么办assist 与同类工具对比assist 最佳实践assist 适合谁用

⚡ 核心功能

👥 适合谁
  • 需要 assist 解决具体问题的开发者与运营人员
⭐ 最佳实践
  • 先在测试环境跑通最小用例，再接入生产数据
⚠️ 常见错误
  • API key 直接提交到 git 仓库（请用 .env 并加入 .gitignore）

👥 适合人群

自动化工程师和运维人员项目经理和业务分析师希望减少重复性工作的专业人士数字化转型团队

🎯 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

✅ 优点
  • +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
⚠️ 不足
  • −未明确开源协议，商用场景需谨慎评估
  • −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
⚠️ 使用须知

该工具未明确声明开源协议，商业使用前请联系原作者确认授权范围，避免侵权风险。

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。

🔗 相关工具推荐

Graphiti知识图谱引擎
Agent工作流
browser-use Agent工作流
Agent工作流
LLM资源合集（精选）
精选100+可直接运行的AI Agent和RAG应用集合。包含完整工作流示例、智能代理框架和检索增强生成系统。适合AI开
Flowise 可视化AI工作流
Agent工作流
📰 相关 AI 新闻
AI 前沿资讯：A2A, how it looks in an enterp…
AI 资讯 · 知识关联
AI 开源生态最新进展
AI 资讯 · 知识关联
MCP 生态最新动态：工具与配置进展
AI 资讯 · 知识关联
Anthropic 研究与产品动态
AI 资讯 · 知识关联
🍿 AI 圈相关吃瓜
配了5个 MCP 工具，Claude 一个都没用
AI 圈观察
Filesystem MCP 帮 Claude 找文件，找了整个 node_modules
AI 圈观察
我用了 Karpathy 的提示词，Claude 终于不瞎编了
AI 圈观察
🗺️ 相关解决方案
mcp
mcp-server-setup
workflow
ai-workflow-templates
cli
cli-productivity
stt
speech-recognition
🧩 你可能还需要
基于当前 Skill 的能力图谱，自动补全的工具组合
技能寻求者
MCP · Agent · 工作流
DeepCode Agent工作流
MCP · Agent · 工作流
natively-cluely-ai-assistant — Claude Skill 中文使用文档
免费开源的AI面试助手，实时转录，隐蔽模式，局部RAG，BYOK。无订阅，防止数据泄露。
gptme Agent工作流
MCP · Agent · 工作流
HexStrike AI MCP智能体
MCP · Agent · 工作流
WordPress MCP管理工具
基于MCP协议的WordPress内容管理服务器，为Claude AI提供WordPress CMS系统的完整管理能力。支持文章、页面、分类

❓ 常见问题 FAQ

使用assist CLI工具来强制确定性LLM开发工作流自动化
💡 AI Skill Hub 点评

总体来看，AI工作流助手 是一款质量良好的Agent工作流，在同类工具中具备一定竞争力。AI Skill Hub 将持续追踪其更新动态，建议收藏备用，结合自身场景选择合适时机引入使用。

⬇️ 获取与下载
⚠️ 该工具未声明开源协议，不提供直接下载。请访问原项目了解使用条款。
📚 深入学习 AI工作流助手
查看分步骤安装教程和完整使用指南，快速上手这款工具
⚙️ 安装教程 📚 使用教程
🌐 原始信息
原始名称 assist
原始描述 开源AI工作流：A CLI tool for enforcing determinism in LLM development workflow automation.。⭐6 · TypeScript
Topics LLM工作流自动化
GitHub https://github.com/staff0rd/assist
语言 TypeScript
🔗 原始来源
🐙 GitHub 仓库  https://github.com/staff0rd/assist 🌐 官方网站  https://staffordwilliams.com/devlog/assist/

收录时间：2026-06-09 · 更新时间：2026-06-09 · License：未公布 · AI Skill Hub 不对第三方内容的准确性作法律背书。