Model behaviour
Figma Weave output truncated halfway: How to fix
Updated 9/24/2026
Figma Weave generates layouts, code snippets, and SVG code directly into your Figma workspace. However, if the target output is highly complex or requires long-form code blocks, Weave may abruptly stop generating. This results in cut-off code, incomplete component designs, unclosed styling tags, or broken vector nodes on your canvas.
This behavior is caused by system-level output token ceilings or regional server-side execution timeouts (usually capped at 30 seconds per stream request). Use the following troubleshooting steps to handle and complete these truncated outputs.
1. Execute a Targeted 'Resume' Prompt Do not simply type "continue" or "finish" into the Weave console when generation halts. This often causes the model to restart from the beginning, wasting tokens and hitting the same timeout limits again.
- Locate the exact spot where the code or layout construction was cut off.
- Copy the final three lines of completed code or identify the last successfully rendered layer name.
- In the chat input, write: "Resume generation exactly from this line: [paste final 3 lines here]. Do not rewrite the preceding code. Output only the remaining components."
- If generating vector layers, write: "Resume building the component group, starting directly with the [name of last layer] layer. Do not regenerate existing parent elements."
2. Deconstruct Generation Requests into Modular Blocks Large code outputs (such as a full React component with complex embedded inline SVGs) easily trigger Weave's maximum output limits. Deconstructing your requests avoids reaching these limits.
- Break your UI designs into modular blocks before prompting.
- Ask Weave to generate the primary layout structure first, using standard CSS layout skeletons.
- In a separate prompt, request the specific, dense CSS classes or SVG asset paths separately.
- Merge the output modules manually rather than forcing Weave to write them in a single generation cycle.
3. Force Lightweight Styling Rules Inline CSS declarations, repeated Tailwind utility classes, and complex gradients inflate the size of Weave's generation payloads. Forcing lightweight outputs helps keep the generation under the token ceiling.
- Instruct Weave to avoid inline CSS or complex SVG nodes where possible.
- Use prompts like: "Generate this component layout using semantic HTML tags and CSS variable references only. Do not inline SVGs or write redundant styling declarations."
- This keeps the text payload small, allowing the entire structural code to complete without running into timeout errors.
- Apply complex styling variables locally using your Figma variables library after Weave finishes compiling the structure.
4. Clear Figma App Cache and Enable Hardware Acceleration If your local Figma desktop app is experiencing high memory pressure, it may shut down the background web workers responsible for parsing and rendering Weave's streamed outputs, cutting off the process mid-way.
- Go to the Figma menu and navigate to Preferences.
- Ensure Use Hardware Acceleration is checked. If it is already enabled, toggle it off and back on.
- Clear the Figma desktop cache files:
- macOS: Delete files in ~/Library/Application Support/Figma/ (specifically the Cache and GPUCache folders).
- Windows: Delete files in %APPDATA%\Figma\ (specifically the Cache and GPUCache folders).
- Restart your system and try running the generation again.
---
Mitigating Output Constraints Avoid prompting Weave when you have multiple heavy browser tabs open or when running resource-intensive design plugins in parallel. This ensures your local rendering thread has enough memory overhead to handle long-running layout streams without dropping WebSocket frames.
---