Named after Liezi’s artificer · Explicit, inspectable, replayable structure

Turn raw sources into finished deliverables on one Rust runtime

Feed an idea, URL, GitHub repository, academic PDF, or recording into deterministic DAG pipelines and agent-native loops. Emit native editable OOXML .pptx, 16:9 .pdf, narrated 1080p .mp4, normalized .mp3, .xlsx workbooks, and verified vector diagrams.

$yanshi workflow run examples/generators/illustrated_slides.yml --input subject "Rust Zero-Copy I/O" --output-dir out/slides
23 + 1 Rust Crates + Studio UI
15 Built-in Generators
6 LLM Multi-Provider Native Tools
< 1ms Scheduler & Guard Overhead
Content Studio · 5-Stage Pipeline Select a stage to inspect

01 · Sandboxed Multimodal Source Ingestion

SSRF-Pinned + Typed Chunks

Ingest web pages with DNS-pinned SSRF protection, snapshot GitHub repos without running untrusted code, parse PDFs/arXiv tarballs, or extract captions + ASR transcripts.

- id: ingest_repo
  type: repo_snapshot
  parameters:
    max_files: 180
    include_imports: true
- id: ingest_paper
  type: pdf_extract
  parameters:
    extract_tables: true
web_search repo_snapshot pdf_extract arxiv video_ingest + asr

Three-Way Slide Visual Routing: Beyond Flat AI Image Decks

All three slides below were produced in a single Yanshi run. Each slide is routed by intent to the right visual engine and compiled together into native editable OOXML .pptx, 16:9 .pdf, and interactive .html.

Visual Routing Spec
Yanshi slide routed to IllustrationNode showing isometric pipeline bottleneck breaking into smooth zero-copy channel
Route 01 · illustration IllustrationNode

Art-Directed Conceptual Illustration

Cover, analogy, and takeaway slides run through LLM art-director prompt refinement with canvas palette harmonization and strict anti-garbled-text guardrails.

editorial_tech isometric_3d flat_vector blueprint_schematic scientific_diagram cinematic_keynote
Yanshi slide routed to FigureNode compiling D2 kernel SQ and CQ ring buffer architecture diagram
Route 02 · arch_diagram D2 · Typst · TikZ

Verified Vector Topology & Schematics

System topologies, state machines, and dataflows compile cleanly via D2, Typst, TikZ, or isolated Python — automatically repairing syntax or layout errors before export.

language: d2 typst tikz / circuitikz container python
Yanshi slide routed to structured_panel showing syntax-highlighted Rust io_uring code and editable callout cards
Route 03 · structured_panel Pure-Rust OOXML

Native Editable Shapes, Tables & Code

Code walkthroughs, benchmarks, and checklists emit real PowerPoint shapes (<p:sp>), <a:tbl> tables, and connectors — 100% editable in PowerPoint or Keynote.

editable .pptx 16:9 .pdf 1080p narrated .mp4

Four Execution Paradigms Sharing One Narrow-Waist Kernel

Span the full binding-time spectrum — from fixed production DAGs to autonomous agent loops and runtime-compiled flows — sharing one ToolRegistry, checkpoint format, and security harness.

Deterministic DAG: Explicit I/O, Cost Caps & Checkpoints

Built for repeatable Studio pipelines, RAG indexing, and batch automation. Validate expressions and node schemas statically before spending a token.

  • Dependency-ready concurrent scheduler with map, while, and run_if
  • Built-in 44-model pricing ledger with hard cost_limit_usd enforcement
  • Embed an autonomous agent inside any DAG step via AgentNode
Read Workflow Schema Guide
YAML examples/generators/paper_reader.yml
# Validate statically, cap spend, and export numeric Excel sheets
name: paper_reader
cost_limit_usd: 1.50
nodes:
  - id: extract
    type: pdf_extract
    parameters:
      source: "{{ source }}"
      extract_tables: true

  - id: bilingual_digest
    type: llm
    dependencies: ["extract"]
    parameters:
      model: "gpt-4o"
      response_format: json_schema
      prompt: "Synthesize key contributions and benchmarks..."

  - id: export_excel
    type: render_table
    dependencies: ["bilingual_digest"]
    parameters:
      format: "xlsx"

15 Built-in Content Studio Generator Workflows

Select any template card below to inspect its full syntax-highlighted YAML source, DAG node topology, and CLI command right on this page.

Web-grounded keynote deck routing slides across 6 art-directed illustration styles, D2 diagrams, and native OOXML shapes.

.pptx 16:9 .pdf .html

Step-by-step progressive slide builds synced millisecond-accurately to TTS narration, spotlight rings, and optional PiP presenter.

1080p .mp4 .srt / .vtt .pptx .pdf

slides.yml

View YAML

Source-grounded presentation with self-repairing vector diagrams; works with zero image API keys out of the box.

.pptx 16:9 .pdf .html

video_overview.yml

View YAML

Parallel multi-scene visual generation (stills or optional motion clips) muxed with per-scene voiceover and subtitles.

.mp4 .srt

audio_overview.yml

View YAML

Two-speaker conversational deep-dive voiced in parallel and loudness-normalized to EBU R128 via render_audio.

.mp3

Scans repository manifests and internal imports safely without executing code, producing verified D2 architecture diagrams and PDF dossier.

.svg / .png .pdf .html

Interactive HITL clarification followed by system architecture, UML class/sequence, ER diagrams, PostgreSQL .sql DDL, and spec PDF.

6x .svg .sql .pdf

paper_reader.yml

View YAML

Extracts experimental benchmark tables into typed Excel workbooks alongside a bilingual digest PDF, method diagram, and mindmap.

.pdf .xlsx / .csv .svg

Prefers native captions with automatic fallback to FFmpeg audio extraction + ASR, emitting timestamped notes, .srt, and mindmap.

.pdf / .html .srt .svg

report.yml

View YAML

Parallel section drafting with verbatim source quotes and chunk-level footnote citations, typeset cleanly into HTML and PDF.

.pdf .html .md

infographic.yml

View YAML

Extracts schema-checked headline metrics and renders crisp deterministic SVG and high-DPI PNG posters.

.svg .png

mindmap.yml

View YAML

Hierarchical knowledge tree rendered simultaneously as vector SVG, PNG, and interactive collapsible HTML.

.svg .png .html

Extracts a typed parts list first, validates every component ID, and compiles publication-grade circuitikz schematics.

.svg .png

flashcards.yml

View YAML

Turns source material into a self-contained offline flip-card study deck and structured JSON dataset.

.html .json

quiz.yml

View YAML

Self-grading multiple-choice assessment page with instant answer rationales and score tracking.

.html .json

Five-Layer Architecture Enforced by Eight Dependency Laws

Runtimes never depend on each other — only on the zero-logic L0 contract kernel, guarded continuously in CI by cargo xtask check-arch.

Architecture Docs
L4 · Operations Gateway, Tracing, Distributed Worker & Web/Desktop UI
yanshi-server yanshi-worker (gRPC) yanshi-db (Postgres) yanshi-tracing (OTLP) yanshi-ui (React + Tauri)
L3 · Orchestration Agent Runtimes, Skills, Governance Harness & Config CLI
yanshi-agents yanshi-skills yanshi-harness yanshi-config yanshi-cli
L2 · Capabilities Deterministic Renderers, AI Nodes, LLM, OS Sandbox, MCP & RAG
yanshi-nodes yanshi-nodes-ai yanshi-llm yanshi-tools yanshi-mcp yanshi-rag yanshi-memory
L1 · Execution Core Concurrent Scheduler, Checkpointing, Retry & Resource Limits
yanshi-core (FlowExt::run)
L0 · Contract Kernel Narrow-Waist Contract Kernel (Shared Types Only, Zero Logic)
yanshi-value yanshi-graph yanshi-tool yanshi-agent-spi yanshi-store-spi yanshi-async-util

CLI & Checkpointed Automation

Validate, execute, checkpoint-resume, and inspect runs interactively in the terminal TUI with yanshi trace replay.

cargo install --path yanshi-cli

Web Studio & Axum Gateway

Launch yanshi-server for /ui/app template forms and /ui/studio DAG debugging with per-node breakpoints and inline media preview.

cargo run --package yanshi-server

Yanshi Studio Desktop App

Standalone Tauri v2 app embedding the local Rust engine — zero Postgres or external server process required.

./scripts/install_desktop.sh --open

Technical FAQ & Architecture Evaluation

Answers on native editable deliverables, model keys, OS-level sandboxing, and how Yanshi compares to Python agent stacks.

Are generated .pptx decks flat AI images or native editable PowerPoint shapes?

They are real, native OOXML presentations. Yanshi’s pure-Rust OOXML writer emits standard PowerPoint shapes (<p:sp>), native tables (<a:tbl>), code callouts, and connectors that open directly for text and layout editing in PowerPoint or Keynote — combined per-slide with verified D2/Typst/TikZ diagrams and 6 art-directed illustration styles, plus matching 16:9 .pdf and .html exports.

Can I run workflows without image or TTS API keys, and which LLM providers are supported?

Yes. A single text LLM key runs most Studio generators out of the box — including slides.yml, report.yml, repo_architecture.yml, backend_project_planner.yml, paper_reader.yml, and mindmap.yml. Yanshi natively supports OpenAI, Anthropic Claude, Google Gemini, Moonshot, StepFun, and any OpenAI-compatible endpoint (DeepSeek, Qwen, or local Ollama).

How does Yanshi differ from LangGraph, Dify, or single-purpose AI slide tools?

Yanshi unifies deterministic concurrent DAGs, native autonomous agent loops (ReAct, Plan-Execute, Handoff/Blackboard/Debate), and runtime-compiled dynamic workflows on one narrow-waist Rust kernel. You get a single fast binary (< 1ms scheduling overhead, built-in 44-model cost_limit_usd enforcement) and first-class FlowValue::File multimodal renderers (.pptx, 1080p .mp4, .pdf, .xlsx, .mp3) with zero Python dependency drift or Base64 memory bloat.

How does Yanshi isolate untrusted repos, web fetches, and LLM-generated code?

Under SecurityProfile::Production, security is fail-closed by default: shell and renderer subprocesses run inside OS sandboxes (macOS sandbox-exec or Linux Landlock + seccomp + cgroup v2), code_exec requires a zero-network ContainerBackend, repo_snapshot inspects manifests statically without executing code, HttpTool pins DNS per request to block TOCTOU SSRF, and marketplace skills/plugins verify Ed25519 signatures.

If a long multi-stage video or report run is interrupted, do resumed runs re-bill completed steps?

No. With --checkpoint (and automatically on the gateway/Studio), Yanshi persists node outputs, SHA-256 file manifests, and accrued spend after every node. Running yanshi workflow resume <run_id> replays completed upstream outputs from disk and executes only unfinished downstream nodes under the remaining cost budget.

examples/generators/illustrated_slides.yml

Download GitHub
DAG Nodes