AI小说生成 是 AI Skill Hub 本期精选Agent工作流之一。综合评分 8.0 分,整体质量较高。我们强烈推荐将其纳入你的 AI 工具库,帮助提升工作效率。
AI小说生成 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。
AI小说生成 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。
# 方式一:go install(推荐) go install github.com/spiritLHLS/novelbuilder@latest # 方式二:从源码编译 git clone https://github.com/spiritLHLS/novelbuilder cd novelbuilder go build -o novelbuilder . # 方式三:下载预编译二进制 # 访问 Releases 页面下载对应平台二进制文件 # https://github.com/spiritLHLS/novelbuilder/releases
# 查看帮助 novelbuilder --help # 基本运行 novelbuilder [options] <input> # 详细使用说明请查阅文档 # https://github.com/spiritLHLS/novelbuilder
# novelbuilder 配置说明 # 查看配置选项 novelbuilder --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 NOVELBUILDER_CONFIG="/path/to/config.yml"
NovelBuilder is an AI long-form fiction workbench with a Vue UI, Go API gateway, Python agent sidecar, optional graph/vector memory, and deployment profiles from SQLite-only local mode to full all-in-one Docker.
| Tag | Dockerfile | Shape | Suggested resources | Notes |
|---|---|---|---|---|
latest, full, YYYYMMDD | Dockerfile | Single container with PostgreSQL, Redis, Qdrant, Neo4j, Python, Go, Vue, Playwright | 4 CPU, 8 GB RAM, 20 GB disk | Complete local deployment |
standard, YYYYMMDD-standard | Dockerfile.standard | Single container with PostgreSQL, Redis, Python, Go, Vue | 2 CPU, 4 GB RAM, 10 GB disk | Installs base Python deps only; graph/vector/browser routes are disabled |
app, YYYYMMDD-app | Dockerfile.app | App, sidecar, and Vue only | 2 CPU, 2 GB RAM plus external services | Includes graph/vector/browser Python deps for external services and upload automation |
sqlite | Dockerfile.sqlite | Independent minimal image with SQLite and optional services disabled | 1 CPU, 2 GB RAM, 5 GB disk | Base Python deps only; intended for single-user local use |
no-neo4j | Dockerfile.no-neo4j | Independent single container with PostgreSQL, Redis, Qdrant, browser automation, Python, Go, Vue | 3 CPU, 6 GB RAM, 15 GB disk | Omits Neo4j and graph Python/runtime deps |
no-qdrant | Dockerfile.no-qdrant | Independent single container with PostgreSQL, Redis, Neo4j, browser automation, Python, Go, Vue | 3 CPU, 6 GB RAM, 15 GB disk | Omits Qdrant and vector Python/runtime deps |
no-graph-vector | Dockerfile.no-graph-vector | Independent single container with PostgreSQL, Redis, Python, Go, Vue | 2 CPU, 4 GB RAM, 10 GB disk | Omits graph/vector/browser runtime deps |
no-redis | Dockerfile.no-redis | Independent single container with PostgreSQL, Python, Go, Vue | 2 CPU, 3 GB RAM, 10 GB disk | Omits Redis service and optional graph/vector/browser runtime deps |
The release workflow builds full, standard, app, and each variant from its own Dockerfile. Variant Dockerfiles no longer inherit from a same-run base tag.
VERSION=dev UPX_ENABLED=auto ./scripts/build-binaries.sh
TARGETS="linux amd64,windows amd64" ./scripts/build-binaries.sh
Go binaries are built with -trimpath, stripped symbols, and an empty build id. If upx is installed, Linux and Windows binary packages are compressed automatically. Docker builds use Node 24 builder images, npm ci, no pip cache, no Python bytecode writes, and split sidecar requirements (base, graph, vector, browser) per profile. GitHub Actions also run JavaScript actions on Node 24-compatible action versions.
All-in-one full profile:
```bash cp .env.example .env
docker compose up -d open http://127.0.0.1:8080/setup
Standard profile without graph/vector services:
bash cp .env.example .env
docker compose -f docker-compose.standard.yml up -d
Minimal SQLite profile:
bash docker compose -f docker-compose.sqlite.yml up -d
Source or binary local mode:
Prerequisites for source builds: Go 1.22+, Python 3.11+, Node.js 20.19+.
bash ./scripts/install.sh ./scripts/run-local.sh
Windows:
powershell powershell -ExecutionPolicy Bypass -File .\scripts\install.ps1 powershell -ExecutionPolicy Bypass -File .\scripts\run-local.ps1 ```
Open /setup first. It checks runtime readiness and then the in-app guide walks through model configuration, project creation, references, blueprint generation, and chapter generation.
Infrastructure settings are environment variables. Application settings, LLM profiles, prompt presets, and runtime snapshots are stored in the database.
| Variable | Default | Notes |
|---|---|---|
APP_PROFILE | full in Docker, binary in local scripts | Displayed in setup diagnostics |
SERVER_HOST, SERVER_PORT, SERVER_MODE | 0.0.0.0, 8080, release | Go gateway listener |
ALLOWED_ORIGINS | localhost dev and :8080 origins | Comma-separated CORS allowlist; use your HTTPS origin for public deployments |
TRUSTED_PROXIES | empty | Comma-separated proxy CIDRs; set only when running behind a trusted reverse proxy |
ADMIN_USERNAME, ADMIN_PASSWORD | admin, generated at runtime when unset | Set a strong ADMIN_PASSWORD; if omitted, read the temporary password from startup logs |
SESSION_TTL_HOURS | 24 | Sliding session lifetime |
LOGIN_MAX_ATTEMPTS | 5 | Failed login attempts before lockout |
LOGIN_WINDOW_SECONDS | 300 | Counting window for failed logins |
LOGIN_LOCKOUT_SECONDS | 900 | Lockout duration after too many failures |
DB_DRIVER | postgres in containers, sqlite in local scripts | sqlite/sqlite3 or postgres |
SQLITE_PATH | /data/novelbuilder.db or ./data/novelbuilder.db | Used when DB_DRIVER=sqlite |
DB_HOST, DB_PORT, DB_USER, DB_PASSWORD, DB_NAME, DB_SSLMODE | host/user/name defaults; password required in PostgreSQL Docker profiles | Used when DB_DRIVER=postgres |
DB_MAX_OPEN_CONNS, DB_MAX_IDLE_CONNS, DB_CONN_MAX_LIFETIME_MINUTES | 25, 5, 60 | Go database pool sizing; open/idle values are clamped to at least 20/5, lifetime to at most 60 minutes |
REDIS_ENABLED, REDIS_ADDR, REDIS_URL, REDIS_PASSWORD, REDIS_DB | profile-specific | Go uses REDIS_ADDR; Python uses REDIS_URL |
SIDECAR_URL, SIDECAR_TIMEOUT | http://127.0.0.1:8081, 600 | Go to Python sidecar |
SIDECAR_DB_MIN_CONNS, SIDECAR_DB_MAX_CONNS | 5, 20 | Python sidecar PostgreSQL pool sizing for legacy DB-backed analysis routes |
NEO4J_URI, NEO4J_USER, NEO4J_PASSWORD | profile-specific; password required when Neo4j is enabled | Empty NEO4J_URI disables graph services |
QDRANT_URL | profile-specific | Empty disables vector services |
TASK_WORKERS, TASK_MAX_RETRIES | 4, 3 | Background task queue |
NB_ACCELERATOR | auto | auto, cpu, cuda, rocm, mps, or npu |
VECTOR_EMBED_CONCURRENCY | 4 | Python-sidecar embedding concurrency during vector rebuilds |
Reference uploads accept .txt, .md, .markdown, .pdf, and .epub files up to 50 MiB. Files are stored under /data/uploads, and the sidecar only reads paths inside that directory.
Authenticated API documentation is available at /api/docs and /api/docs/openapi.json. The docs routes use the same admin session middleware as the main API; CLI callers should pass Authorization: Bearer <token>, while direct browser access can use /api/docs?token=<token>.
创新性的AI小说生成平台
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。
⚠️ GPL 3.0 — 强 Copyleft,衍生作品须开源,含专利保护条款,不可闭源使用。
经综合评估,AI小说生成 在Agent工作流赛道中表现稳健,质量优秀。如果你已有明确的使用需求,可以直接上手体验;如果还在评估阶段,建议对比同类工具后再做决策。
| 原始名称 | novelbuilder |
| Topics | AI小说生成工作流 |
| GitHub | https://github.com/spiritLHLS/novelbuilder |
| License | GPL-3.0 |
| 语言 | Go |
收录时间:2026-07-05 · 更新时间:2026-07-05 · License:GPL-3.0 · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端