Skip to content

Merge dev to main for 0.2.0 - #17

Merged
Natuworkguy merged 8 commits into
mainfrom
dev
Aug 21, 2026
Merged

Merge dev to main for 0.2.0#17
Natuworkguy merged 8 commits into
mainfrom
dev

Conversation

@Natuworkguy

Copy link
Copy Markdown
Owner

No description provided.

@Natuworkguy Natuworkguy added the release Official release pull request. Allows merging directly into main. label Aug 21, 2026
@Natuworkguy
Natuworkguy merged commit 1e54359 into main Aug 21, 2026
19 checks passed

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: bcd11df325

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread flash/ai.py
Comment on lines 829 to +831
res, err = _chat_with_status(
console, client, [system_message] + messages, tools
console, client, [system_message] + messages, tools,
is_image=bool(pending_images),

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Disable tool calling for image-only models

When /image sets pending_images, this call still sends the full tools list to Ollama. Common vision models such as the documented llama3.2-vision are image-capable but not tool-capable, so Ollama rejects the request before the model can answer; image recognition will fail for the recommended setup unless tools are omitted or gated for image turns.

Useful? React with 👍 / 👎.

Comment thread flash/ai.py
warn("Usage: /image <path> [prompt]")
continue
try:
parts = shlex.split(arg)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Preserve Windows backslashes when parsing image paths

On Windows, unquoted paths normally contain backslashes, but shlex.split defaults to POSIX escaping, so an input like /image C:\Users\me\Pictures\cat.png describe is parsed as C:UsersmePicturescat.png and then fails the is_file() check. This makes the new image command unusable for ordinary Windows paths unless users quote or double-escape them; parse with Windows-aware rules or split the first argument without consuming backslashes.

Useful? React with 👍 / 👎.

Comment thread flash/tools.py
Comment on lines +202 to +203
if count > MAX_SEARCH_FILES:
return

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Surface incomplete grep/glob scans

For repositories with more than 5,000 non-excluded files, _iter_files silently returns before scanning the rest of the tree. Because glob_tool/grep_tool can then report No matches even when later files contain matches, and the system prompt tells the model to prefer these tools for codebase search, users can get incorrect search results with no truncation warning.

Useful? React with 👍 / 👎.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

release Official release pull request. Allows merging directly into main.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant