Performance async image read 10974862260397376706 - #1486
Open
0m364 wants to merge 36 commits into
Open
Conversation
Instructions for unrooted android via termux 👀
corrected mistypes added termux-api (not needed for aichat for some reason but it does seem to make it faster) and added --locked install option as I ran into an issue on proot-distro of alpine .. fixed with --locked
makes a local API to use with a rust based llm (hopefully.. not tested yet)
- Add `pre_command` and `post_command` to global config to allow running scripts before/after chat completion. - Add `agents/android` with Termux tool definitions and wrapper scripts. - Add `create_tool` script to allow agentic tool creation. - Add `roles/jules.md` persona. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
…82055537 Add Android agent and hooks
Embeds a set of native developer tools directly into aichat to enhance its agent capabilities without relying on external scripts. This includes: - `fs_cat`: Read file contents. - `fs_ls`: List directory contents. - `fs_mkdir`: Create directories. - `fs_write`: Write file contents. - `fs_search`: Recursive substring search in files. - `command_run`: Execute shell commands. These tools are automatically registered and available for function calling, similar to OpenCode and Goose agents. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
Embed Agentic Toolkit (OpenCode/Goose style)
- Added `src/client/jules.rs` implementing JulesClient (OpenAI-compatible/Gemini-like API). - Registered `jules` in `src/client/mod.rs`. - Updated `src/config/mod.rs` to automatically enable all built-in tools (fs_*, command_run) for the `jules` client, making it "full powered" by default. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
- Removed blocking `std::thread::sleep` from `Spinner::set_message`. - Converted `Spinner::stop` to an `async` function and replaced `std::thread::sleep` with `tokio::time::sleep`. - Updated all call sites of `spinner.stop()` in `src/render/stream.rs` to `.await` the new async function. These changes prevent worker thread starvation in the tokio executor, especially during RAG synchronization and message streaming. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
Using `unwrap()` on `partial_cmp` is unsafe because floating point numbers can be NaN, which `partial_cmp` returns as `None`. This can cause a panic. This change replaces it with `total_cmp`, which provides a total ordering for f32 and safely handles NaN. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
Implemented a comprehensive suite of tests for the `strip_think_tag` function in `src/utils/mod.rs`. The tests cover: - Basic tag removal at the start of the string. - Handling of leading and trailing whitespace around the tag. - Multiline tags and content. - Strings without any tags. - Ensuring tags not at the start are preserved. - Verifying that only the first anchored tag is removed. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
This commit fixes a security vulnerability where filenames starting with a dash ('-')
could be interpreted as command-line options when substituted into a loader command.
Changes:
- Added `sanitize_path` to prepend `./` to relative paths starting with '-'.
- Refactored `run_loader_command` to use a more robust placeholder replacement
logic that prevents double-expansion vulnerabilities.
- Added unit tests for path sanitization and placeholder replacement.
- Extracted logic into separate functions for better testability.
Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
…5988954876939 🔒 fix: argument injection via filename in loader command
…96288677195082 🧪 Add unit tests for `strip_think_tag`
…6243707556956 🧹 replace unsafe unwrap() on partial_cmp with total_cmp
…1211793 Add Google Labs Jules client with full powered tools
…47469 ⚡ Optimize Spinner by removing blocking sleeps in async context
…event XSS - Introduce DOMPurify to sanitize HTML generated from Markdown. - Implement event delegation for 'Copy code' buttons to maintain functionality after sanitization. - Sanitize error messages to prevent reflected XSS. - Apply fixes to both assets/playground.html and assets/arena.html. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
Added a comprehensive test suite for `estimate_token_length` in `src/utils/mod.rs` covering: - Empty strings - Single and multiple ASCII words - Non-ASCII (CJK) characters - Mixed content and punctuation handling This increases the reliability and coverage of the token length estimation utility. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
This commit optimizes the performance of the web crawler by avoiding redundant parsing of CSS selectors. 1. Added a static LINK_SELECTOR using LazyLock for the common "a" tag selector. 2. Updated CrawlOptions to include a cached extract_selector (Arc<Selector>). 3. Added a prepare() method to CrawlOptions to pre-parse the extraction selector once per crawl. 4. Updated crawl_website to call options.prepare() and crawl_page to use the pre-parsed selectors. These changes significantly reduce CPU overhead and allocations during website crawling, especially for large sites with many pages. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
…398699382699 🧪 Add unit tests for estimate_token_length utility
…9400824048618 ⚡ [performance] Cache CSS selector parsing
…98204911 🔒 [security fix] Fix XSS in Playground and Arena via Markdown Rendering
Convert `read_media_to_data_url` to an asynchronous function using `tokio::fs::read` to avoid blocking the Tokio worker threads in `load_documents`. This improves the overall efficiency and responsiveness of the application by yielding the thread back to the runtime during disk I/O. Co-authored-by: 0m364 <13804150+0m364@users.noreply.github.com>
ei-grad
added a commit
to ei-grad/aichat
that referenced
this pull request
Jul 11, 2026
Add an explicit --files list grammar for shell-expanded inputs while preserving the established single and repeated -f behavior. This avoids classifying positional prompt tokens by filesystem existence and makes the file/prompt boundary explicit. Apply command preludes before --info with defined REPL fallback semantics, parse editor commands into an executable plus arguments without invoking a shell, and reuse that parser for REPL buffer editing. Move local image reads to Tokio, use the configured warning color for generated shell commands, and document installation and local-agent layouts. Adapt the bounded fixes discussed in: sigoden#1451 sigoden#1509 sigoden#1534 sigoden#1533 sigoden#1512 sigoden#1525 sigoden#1383 sigoden#1437 sigoden#1486 Co-Authored-By: Codex CLI <noreply@openai.com>
ei-grad
added a commit
to ei-grad/aichat
that referenced
this pull request
Jul 11, 2026
Salvage only the async media-read intent from sigoden#1486 without importing its polluted ancestry. Enable Tokio filesystem support, await local image reads, separate byte-to-data-URL conversion for deterministic tests, and include the source path in unsupported-type and read failures. Source PR: sigoden#1486 Source commit: sigoden@0cfa32f Reviewed behavior: 542f02a Co-Authored-By: Codex CLI <noreply@openai.com>
ei-grad
added a commit
to ei-grad/aichat
that referenced
this pull request
Jul 11, 2026
Integrate the narrow Tokio filesystem branch without importing upstream PR sigoden#1486 or its unrelated history. The second parent records only the media read, error-context, helper, test, and Cargo feature delta. Source: sigoden#1486
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.