emergence-render-image（出现渲染图像）

Emergence Render Image Skill

This skill provides a programmatic interface to the Emergence Science Render API. It allows humans and AI agents to transform structured code into professional-grade scientific and technical visualizations.

1. Persona & Objective

The primary user of this skill is the Autonomous AI Agent. As many LLMs lack the ability to directly render pixels, this skill acts as the agent's "visual cortex" and "drawing hand," enabling it supplemented textual reasoning with high-fidelity diagrams.

Existing Pain Points

• Human-Centric Tools: Most online TikZ/Mermaid tools are interactive editors designed for humans, making them difficult for agents to automate.
• Syntactic Hallucination: LLMs often struggle with valid TikZ syntax. Without the ability to perform repetitive editing and validation via a stable API, agents are subject to hallucinations.
• Heavy Dependencies: TikZ and LaTeX libraries are resource-heavy to install and maintain locally. A REST API is the most efficient solution for agents to generate serious academic-level images on demand.

2. Authentication & Credits

Registration

Humans must register on the Emergence Science Web UI using GitHub OAuth.

Token Management

1. Navigate to the Web UI after login to obtain your EMERGENCE_API_KEY.
2. Paste this token into your Agent's environment configuration.
3. Scoped Access: This API key is utilized exclusively by this skill to call the rendering endpoint.
4. Incentive: Every new verified user is granted 1,000,000 micro-credits to be used across the Emergence Science ecosystem, including rendering services.

3. Usage & Examples

The service supports multiple diagramming engines and output formats.

Endpoint

https://api.emergence.science/tools/render

Method: POST Headers:

• Authorization: Bearer <EMERGENCE_API_KEY>
• Content-Type: application/json

[!WARNING] Response Latency: The REST API response time can be as slow as 1 minute due to the heavy computational overhead of LaTeX/TikZ rendering. Agents and callers should implement appropriate socket timeouts and be patient during large image generation.

Supported Formats

• png (Default)
• svg

[Engine: TikZ]

Used for high-rigor mathematical and scientific plots.

Request Payload:

{
  "engine": "tikz",
  "code": "\\begin{tikzpicture}[x=1cm, y=1cm]\
\\draw[blue, thick] (0,0) circle (1.5);\
\\
ode at (0,0) {Quantum Core};\
\\end{tikzpicture}",
  "format": "png"
}

[Engine: Mermaid]

Best for workflows, causal graphs, and sequence diagrams.

Request Payload:

{
  "engine": "mermaid",
  "code": "graph TD\
  Agent[AI Agent] -->|Auth| Hub[Emergence Hub]\
  Hub -->|Credits| Render[Render API]\
  Render -->|Image| Agent",
  "format": "svg"
}

[Engine: Graphviz]

Ideal for visualizing complex network topologies and tree structures.

Request Payload:

{
  "engine": "graphviz",
  "code": "digraph G {\
  rankdir=LR;\
  Input -> Processor -> Output;\
  Processor -> DB [label=\"cache\"];\
}",
  "format": "png"
}

[Engine: D2]

Modern, fast, and highly readable diagramming language.

Request Payload:

{
  "engine": "d2",
  "code": "User -> API: Request\
API -> Database: Query\
Database -> API: Results\
API -> User: Response",
  "format": "png"
}

[Response Schema]

The API returns a JSON object containing the status, the rendered image in Base64 format, and billing details.

Sample Response:

{
    "status": "success",
    "data":
    {
        "image_base64": "PD94bWwgdmVyc2lvbj0iMS4wIiBlbmNvZ...dmc+Cg==",
        "format": "svg"
    },
    "billing":
    {
        "cost": 0.001,
        "remaining_credit": 0.564
    }
}

Post-Processing: Agents are encouraged to decode the data.image_base64 string directly using the base64 command (e.g., echo "..." | base64 -d > output.png).

[Discovery & OpenAPI]

The full up-to-date REST API schema is available at: https://emergence.science/openapi.json

[!TIP] The openapi.json file is extensive. It is recommended to use the jq command for targeted inspection and filtering of endpoints.

4. Policy & Constraints

Rate Limiting

Users and agents must respect the 1-minute rate limit per account. Excessive requests may trigger temporary IP-based or Account-based blocks.

Governance & Security

[!CAUTION] No Malicious Code Injection: Use of the API to attempt sandbox escapes, network penetration, or injection of malicious LaTeX/Mermaid macros is strictly prohibited. All requests are logged and periodically audited. Violations will result in immediate forfeiture of credits and account suspension.

[!NOTE] Future Roadmap: Support for PlantUML and C4 architectural diagrams is scheduled for release in May 2026.