anomalyco/opencode
1
0
Fork 0
mirror of https://github.com/anomalyco/opencode.git synced 2026-07-22 02:05:26 -04:00
Code Issues 5.9k Packages Projects Releases 19 Wiki Activity

Unsure if image analysis is supported in OpenCode (Windows 11, WezTerm & PowerShell) #1927

New Issue
Closed
opened 2026-02-16 17:33:16 -05:00 by yindo · 4 comments
yindo commented 2026-02-16 17:33:16 -05:00
Owner
Copy Link
Copy Source

Originally created by @thecuriousair598 on GitHub (Oct 2, 2025).

I am new to OpenCode and I’m not sure whether screenshot/image analysis is actually supported or if I need to configure something locally. When I try, I consistently get a message saying the assistant cannot analyze images. I tried with different models and in both WezTerm and native PowerShell on Windows 11. Reporting here for clarification, as I could not find anyone else mentioning this issue, and maybe other new users may face the same question.

Environment

OS: Windows 11 (fully updated)

Terminal(s): WezTerm (latest) and Windows Terminal / native PowerShell

OpenCode Models: Tested with multiple (including Claude Sonnet 4.5)

Location: Australia/Sydney (UTC+10)

Plan: ChatGPT Plus

Steps to Reproduce

Start an OpenCode session on Windows 11.

Attach a JPG/PNG screenshot for analysis.

Prompt the assistant to “analyze the screenshot.”

Observe the reply.

Actual Result
Instead of analyzing the screenshot, the assistant replies:

I apologize, but I cannot directly view or analyze image files like JPG screenshots. My tools only allow me to read text-based files, not binary image files.

Expected (Not sure)

If image analysis is supported: I expected the assistant to process the screenshot directly.

If it requires setup: Clear documentation or instructions to enable the feature.

If not supported: A clearer message confirming the limitation.

Impact
Blocks me from using screenshots for troubleshooting/debugging, which I thought was part of the feature set.

Notes

  • The reply also included a line like: Plan claude-sonnet-4-5
  • I’m reporting mainly for clarification — not sure if it’s a bug, misconfiguration, or expected behaviour.
Image
Originally created by @thecuriousair598 on GitHub (Oct 2, 2025). I am new to OpenCode and I’m not sure whether screenshot/image analysis is actually supported or if I need to configure something locally. When I try, I consistently get a message saying the assistant cannot analyze images. I tried with different models and in both WezTerm and native PowerShell on Windows 11. Reporting here for clarification, as I could not find anyone else mentioning this issue, and maybe other new users may face the same question. **Environment** OS: Windows 11 (fully updated) Terminal(s): WezTerm (latest) and Windows Terminal / native PowerShell OpenCode Models: Tested with multiple (including Claude Sonnet 4.5) Location: Australia/Sydney (UTC+10) Plan: ChatGPT Plus Steps to Reproduce Start an OpenCode session on Windows 11. Attach a JPG/PNG screenshot for analysis. Prompt the assistant to “analyze the screenshot.” Observe the reply. **Actual Result** Instead of analyzing the screenshot, the assistant replies: I apologize, but I cannot directly view or analyze image files like JPG screenshots. My tools only allow me to read text-based files, not binary image files. **Expected (Not sure)** If image analysis is supported: I expected the assistant to process the screenshot directly. If it requires setup: Clear documentation or instructions to enable the feature. If not supported: A clearer message confirming the limitation. **Impact** Blocks me from using screenshots for troubleshooting/debugging, which I thought was part of the feature set. **Notes** - The reply also included a line like: Plan claude-sonnet-4-5 - I’m reporting mainly for clarification — not sure if it’s a bug, misconfiguration, or expected behaviour. <img width="3815" height="1916" alt="Image" src="https://github.com/user-attachments/assets/22f9e53c-4c2f-40ba-9773-43f77e505f8e" />
yindo closed this issue 2026-02-16 17:33:16 -05:00
yindo commented 2026-02-16 17:33:16 -05:00
Author
Owner
Copy Link
Copy Source

@github-actions[bot] commented on GitHub (Oct 2, 2025):

This issue might be a duplicate of existing issues. Please check:

  • #2657: Similar question about reading images - user asking what tool can read images in OpenCode, experiencing the same "Read tool fails to read any images" issue
  • #1351: Unable to read/view PNG images - user requesting ability to view/read PNG files, similar limitation
  • #2825: Same error message "I cannot analyze or describe image files" and "The read tool cannot process binary files like images" when trying to analyze generated image files

Feel free to ignore if none of these address your specific case.

@github-actions[bot] commented on GitHub (Oct 2, 2025): This issue might be a duplicate of existing issues. Please check: - #2657: Similar question about reading images - user asking what tool can read images in OpenCode, experiencing the same "Read tool fails to read any images" issue - #1351: Unable to read/view PNG images - user requesting ability to view/read PNG files, similar limitation - #2825: Same error message "I cannot analyze or describe image files" and "The read tool cannot process binary files like images" when trying to analyze generated image files Feel free to ignore if none of these address your specific case.
yindo commented 2026-02-16 17:33:16 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Oct 2, 2025):

@thecuriousair598 so there is some better support we need to do, we may even have something we need to fix that is specific to windows but @itsrainingmani would know status of it, put currently:

only way to give images to the llm is via:
A) @ referencing it your prompt and selecting it as an "attachment"
B) drag & drop / paste the image into input and it should be linked as attachment

I have a branch where i allow LLM to use the read tool to read images but I need to clean it up a bit before we can merge it

@rekram1-node commented on GitHub (Oct 2, 2025): @thecuriousair598 so there is some better support we need to do, we may even have something we need to fix that is specific to windows but @itsrainingmani would know status of it, put currently: only way to give images to the llm is via: A) @ referencing it your prompt and selecting it as an "attachment" B) drag & drop / paste the image into input and it should be linked as attachment I have a branch where i allow LLM to use the read tool to read images but I need to clean it up a bit before we can merge it
yindo commented 2026-02-16 17:33:16 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Oct 2, 2025):

We actually have issues tracking both of these things, I will link them here but I will close this as a duplicate in meantime

@rekram1-node commented on GitHub (Oct 2, 2025): We actually have issues tracking both of these things, I will link them here but I will close this as a duplicate in meantime
yindo commented 2026-02-16 17:33:16 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Oct 2, 2025):

reading images via read tool: #2657

window image pasting issues: https://github.com/sst/opencode/issues/2878

@rekram1-node commented on GitHub (Oct 2, 2025): reading images via read tool: #2657 window image pasting issues: https://github.com/sst/opencode/issues/2878
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
acp
bad
bug
discussion
docs
documentation
enhancement
extension
good first issue
growth
help-wanted
mcp
model-problem
needs:compliance
nix
opentui
perf
pull-request
Mirrored from GitHub Pull Request
 tui
web
windows
zen
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: anomalyco/opencode#1927
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?