做什么
与 Neo4j 命令行工具工作时使用 – New4j- cli( 现代统一)
技能库 智客分类:Agent 工作流 neo4j-cli-tools-skill
与 Neo4j 命令行工具工作时使用 – New4j- cli( 现代统一)
官方网址:skills.sh
先看中文介绍;官方 description 原文单独保留,不改写 SKILL.md。
与 Neo4j 命令行工具工作时使用 – New4j- cli( 现代统一)
- 密码查询、计划检查、Aura、Docker、证书、代理技能-`neo4j-cli'(**首选**)
按 Agent Skills 渐进披露:启动时只加载 name 与 description(约 100 token);任务匹配后才读入整份 SKILL.md 正文;scripts/、references/、assets/ 仅在需要时再读。 本文件正文结构:Neo4j CLI Tools skill、When to Use、When NOT to Use、Available CLI Tools、1. neo4j-cli、Schema-first workflow: always inspect schema before writing Cypher。 其中含规范建议的小节:分步指令。
文件分析:除 SKILL.md 外,正文引用了 references/neo4j-admin-reference.md、references/cypher-shell-reference.md、references/aura-cli-reference.md、references/neo4j-mcp-reference.md,属于带资源的技能包,这些文件按需再读。
Use when working with Neo4j command-line tools — neo4j-cli (modern unified
Neo4j CLI Tools skillWhen to UseWhen NOT to UseAvailable CLI Tools1. neo4j-cliSchema-first workflow: always inspect schema before writing CypherRun Cypher (bolt URIs auto-rewritten; HTTP URIs rewritten to Bolt)Inline embedding parameter (text → vector; $q bound to []float32)Store a connection profile, then query without flagsLocal Docker: persistent or ephemeralInstall skills from catalog (neo4j-contrib/neo4j-skills)2. neo4j-admin
· allowed-tools:WebFetch, Bash
来源分类:skills.sh agent-skill
nameneo4j-cli-tools-skilldescriptionallowed-toolsreferences/neo4j-admin-reference.mdreferences/cypher-shell-reference.mdreferences/aura-cli-reference.mdreferences/neo4j-mcp-reference.md以下路径提取自原文;文件是否齐全请以来源仓库中的完整目录为准。
具体调用语法与可用工具以目标 Agent 客户端为准。 查看调用机制说明 ↗
先选择目标 Agent 和安装范围,保留技能包的附属文件,安装后检查客户端能否发现该技能。
该技能引用了附属文件,请从来源获取完整目录;仅复制 SKILL.md 可能缺少依赖。
复制安装指令给支持 Agent Skills 的代理,确认其中的目标目录与客户端匹配。
把 Agent Skill「neo4j-cli-tools-skill」安装到我的项目:SKILL.md 原文与官方 description 见 https://zicq.com/zh/skills/skl-b60de8701f16a0da-Neo4j-Cli-%E5%B7%A5%E5%85%B7%E6%8A%80%E8%83%BD.html 请存为 .cursor/skills/neo4j-cli-tools-skill/SKILL.md 或 .claude/skills/neo4j-cli-tools-skill/SKILL.md,frontmatter 的 name 与 description 保持原样,不要改写。 该技能还带 scripts/、references/、assets/ 等文件,请从 https://github.com/neo4j-contrib/neo4j-skills 取完整目录,不要只建一个 SKILL.md。
需要 Node.js 与 npx。先查看仓库技能列表,确认实际名称。
npx skills add 'https://github.com/neo4j-contrib/neo4j-skills' --list
npx skills add 'https://github.com/neo4j-contrib/neo4j-skills' --skill 'neo4j-cli-tools-skill'
CLI 会交互选择目标 Agent,默认安装到项目;用户级安装使用 -g。先通过查看命令核对仓库内容,再用 npx skills list 检查已安装技能。
neo4j-cli (preferred)neo4j-admincypher-shellneo4j-mcpneo4j-cypher-skillneo4j-migration-skillneo4j-getting-started-skillModern unified CLI. Bolt-native Cypher, schema inspection, Aura management, local Docker, credential storage, agent skill catalog install. Single binary — no Java required.
Install:
curl -sSfL https://neo4j.sh/install.sh | bash # recommended
brew install neo4j-labs/tap/neo4j-cli # Homebrew
pip install neo4j-cli / pipx install neo4j-cli # PyPI
npm i -g @neo4j-labs/cli # npm
Self-update: neo4j-cli update (also refreshes installed skills automatically)
Key subcommands:
| Subcommand | Purpose |
|---|---|
| neo4j-cli query [cypher] | Run Cypher via Bolt — auto-rewrites HTTP URIs |
| neo4j-cli query :schema | Inspect labels, rel types, indexes, constraints |
| neo4j-cli query :embed [text] | Compute embedding vector standalone |
| neo4j-cli aura instance ... | Provision and manage Aura instances |
| neo4j-cli aura agent ... | Manage Aura Agents (list/get/create/invoke) |
| neo4j-cli docker create/start/stop/delete | Manage local Neo4j Docker containers |
| neo4j-cli credential aura-client ... | Store Aura API credentials |
| neo4j-cli credential dbms ... | Store Neo4j connection profiles (URI + auth) |
| neo4j-cli credential embed ... | Store embedding provider credentials |
| neo4j-cli skill install [skill-name] | Install self-skill or any catalog skill |
| neo4j-cli skill list / check / remove | Manage installed agent skills |
| neo4j-cli update | Self-update binary + refresh installed skills |
Key patterns:
# Schema-first workflow: always inspect schema before writing Cypher
neo4j-cli query :schema --format toon
# Run Cypher (bolt URIs auto-rewritten; HTTP URIs rewritten to Bolt)
neo4j-cli query "MATCH (n:Person) RETURN n.name LIMIT 5" \
--uri neo4j+s://xxx.databases.neo4j.io --username neo4j --password $PASS
# Inline embedding parameter (text → vector; $q bound to []float32)
neo4j-cli query "MATCH (c) SEARCH c IN (VECTOR INDEX chunk_embedding FOR \$q LIMIT 5) SCORE AS score RETURN c.text, score" \
--param q:embed="graph database performance"
# Store a connection profile, then query without flags
neo4j-cli credential dbms add --name prod \
--uri neo4j+s://xxx.databases.neo4j.io --username neo4j --password $PASS
neo4j-cli query --credential prod "MATCH (n) RETURN count(n)"
# Local Docker: persistent or ephemeral
neo4j-cli docker create --name dev --wait --rw
neo4j-cli query --credential dev 'RETURN 1 AS n'
neo4j-cli docker create --name tmp --ephemeral --env-out-file /tmp/n.env --wait --rw
neo4j-cli query --env /tmp/n.env 'RETURN 1 AS n'
# Install skills from catalog (neo4j-contrib/neo4j-skills)
neo4j-cli skill install # self-skill into all detected agents
neo4j-cli skill install neo4j-cypher-skill # one catalog skill, all agents
neo4j-cli skill install --all # self-skill + every catalog skill
neo4j-cli skill install --agent claude-code # scope to one agent
neo4j-cli skill check # detect drift after upgrades
Agent output: Always use --format toon — ~40% fewer tokens than JSON. Set as default: neo4j-cli config set format toon.
Write gate: Write operations require --rw under agent harnesses. Do NOT add it preemptively — if a command fails with "this command writes; pass --rw to allow it", surface the error and ask the user once.
Async operations: instance create/resize/destroy and docker create accept --wait to block until terminal state.
Credentials precedence: flag > OS env (NEO4J_URI, NEO4J_USERNAME, NEO4J_PASSWORD, NEO4J_DATABASE) > .env walk-up > stored credential.
Supported agents: Claude Code, Cursor, Windsurf, Copilot, Gemini CLI, Cline, Codex, OpenCode, Junie, and more.
Full reference: neo4j-cli skill install keeps an always-in-sync skill in your agent.
Included with Neo4j. Backup, restore, dump/load, CSV import, memory sizing, password reset. Run as Neo4j system user.
Reference: neo4j-admin-reference.md
Interactive Cypher REPL. Requires Java 21. Included with Neo4j. Use for ad-hoc queries, file-based batch scripting, CI/CD pipelines.
Reference: cypher-shell-reference.md
Legacy Aura cloud CLI. Prefer neo4j-cli aura for new work — covers all aura-cli operations plus credential storage, Docker, and agent skills.
Reference: aura-cli-reference.md
MCP server for AI agent integration. For full install + editor config → use neo4j-mcp-skill (covers Claude Code, Desktop, Cursor, Windsurf, VS Code, Kiro, stdio vs HTTP transport, troubleshooting).
pip install neo4j-mcp-server && neo4j-mcp --version
Reference: neo4j-mcp-reference.md
Online backup requires Enterprise Edition. Community Edition users must use dump/load only (database must be offline).
# Full online backup — database stays online during backup
neo4j-admin database backup \
--to-path=/backups/ \
--database=neo4j \
--compress
# Differential backup — only changes since last full backup (faster)
neo4j-admin database backup \
--to-path=/backups/ \
--database=neo4j \
--type=DIFF \
--compress
# Backup to cloud storage (S3, GCS, or HTTPS)
neo4j-admin database backup \
--to-path=s3://my-bucket/neo4j-backups/ \
--database=neo4j
AGENT GATE — destructive operation: Before running restore, show the user the exact command and target database name, and wait for explicit confirmation. A restore overwrites the existing database.
# Restore from a full or differential backup
# Requires DB to be stopped, or use --force-offline for a running instance
neo4j-admin database restore \
--from-path=/backups/neo4j-2026-01-15T10-00-00/ \
--database=neo4j \
--overwrite-destination=true
# Restore to a new database name (non-destructive path)
neo4j-admin database restore \
--from-path=/backups/neo4j-2026-01-15T10-00-00/ \
--database=neo4j-restored
Use for migrations, dev/test data transfers, and Community Edition backups.
# Dump — stop the database first, or pass --force-offline
neo4j-admin database dump --to-path=/exports/ neo4j
# Load — overwrites if target DB exists
neo4j-admin database load \
--from-path=/exports/neo4j.dump \
--database=neo4j \
--overwrite-destination=true
AGENT GATE — destructive operation: Before running load with --overwrite-destination=true, confirm target database name and path with the user.
| Flag | Notes |
|---|---|
| --compress | Zstd compression on backup archives |
| --type=DIFF | Differential: only changes since last full backup |
| --to-path | Local path or s3://, gs://, https:// |
| --overwrite-destination=true | Required if target database already exists |
| --force-offline | Allow backup/restore of a running database in some scenarios |
/backups/neo4j-2026-01-19T02-00-00/# Example: restore Sunday full + Monday + Tuesday differentials
neo4j-admin database restore \
--from-path=/backups/neo4j-2026-01-19T02-00-00/ \
--database=neo4j --overwrite-destination=true
neo4j-admin database restore \
--from-path=/backups/neo4j-2026-01-20T02-00-00/ \
--database=neo4j --overwrite-destination=true
neo4j-admin database restore \
--from-path=/backups/neo4j-2026-01-21T02-00-00/ \
--database=neo4j --overwrite-destination=true
NEO4J_URI / NEO4J_ADDRESS — connection URINEO4J_USERNAME / NEO4J_PASSWORD / NEO4J_DATABASENEO4J_CONF — path to neo4j.conf directoryNEO4J_HOME — Neo4j installation directoryPrecedence: flags > env vars > config files. neo4j-cli uses its own credential store — see neo4j-cli credential dbms.
neo4j-cli skill install run after install or upgrade (keeps skill in sync)credential dbms; not hardcoded--format toon for agent-facing neo4j-cli outputSHOW INDEXES + countAgent 工作流
创造有效技能指南。 当用户想创造出新的技能(或更新现有的技能),以专业知识,工作流程,或工具集成来扩展克洛德的能力时,应该使用这种技能.
Agent 工作流
使用ClawdHub CLI搜索,安装,更新并发布从taladhub.com的代理技能. 需要获取苍蝇上的新技能时使用,将安装的技能同步到最新版本或特定版本,或者发布 npm-instainddhub CLI 的新/更新的技能文件夹.
Agent 工作流
管弦乐团多代理团队,任务设定周期,交接协议,审查工作流程. 使用时间: (1)建立2+特派员队伍,具有不同专业,(2)确定任务路线和生命周期(收录框_ spec_建设_审查_完成),(3)在特派员之间制定交接协议,(4)建立审查和质量关口,(5)管理特派员之间的交流和文物共享.
Agent 工作流
Spec-first,TDD,子代理驱动的软件开发工作流程. 当:(1)构建任何新功能或应用——触发脑暴_计划_子代理执行回路,(2)调试出一个bug或测试失败——触发系统性的根起过程,(3)用户说"让我们构建","帮助我计划","我想添加X",或"这个被打破",(4)完成一个功…