Conversation
…d Tauri API feat(review-platform): link pull requests with review chat sessions feat(review-platform): support inner git platform fix(review-platform): fix: improve PR review panel reliability and launch context feat(review-platform): open PR links in dedicated detail panel feat: add review platform agent tool feat(review-platform): refine pull request detail context actions fix(review-platform): propagate remoteId through PR context and auto-resolve write actions feat(review-platform): unify PR review/comment timeline perf(review-platform): load PR detail sections on demand feat(review-platform): change User-Agent to ReviewPlatform feat(review-platform): add lazy CI log details for pull requests feat(review-platform): add auth challenge flow for token setup feat(review-platform): replace inner CI data source with MR-level APIs feat(review-platform): integrate PR panel APIs into ReviewPlatform tool
refactor(core): externalize multitask reminders and clarify subagent delegation
feat(review-platform): add review platform panel with core service and Tauri API
…d-preview fix: render mermaid in markdown editor preview
- add AIConfig.write_tool_mode with plaintext_followup default - support inline_content and plaintext_followup under the same Write tool name - switch Write schema, validation, and execution by context mode - gate round-executor plaintext follow-up generation by write tool mode - propagate write tool mode through execution and tool pipeline contexts - preserve ACP compatibility by forcing inline content mode
…t handler - Extend document-level Shift+Tab exclusion to cover INPUT, SELECT, and contentEditable elements, not just TEXTAREA, preventing unintended mode cycling when typing in any form field - Remove redundant Shift+Tab handler from textarea React onKeyDown since the document listener handles it with proper exclusion - Remove unnecessary comment explaining what the code already expresses
…ling The textarea handler and document listener serve different purposes: - Textarea handler: Shift+Tab in the editable chat input → cycle mode - Document listener: Shift+Tab when input is collapsed → cycle mode (with TEXTAREA exclusion to prevent double-fire) Removing the textarea handler meant Shift+Tab in the expanded textarea would be consumed by the document listener instead, which is less direct and breaks the expected UX of Tab-based mode switching while typing.
…tions - generalize background result delivery beyond subagents - auto-inject completed background bash command results via steering - persist full background bash output to runtime files instead of chat - clarify no-poll guidance for background subagents and commands
feat(cli): Improve non interactive command
…exclusion fix(flow-chat): improve Shift+Tab input exclusion and remove redundant handler
Move product/runtime-only direct dependencies behind existing feature gates while keeping product-full as the default complete runtime assembly. Document that service-integrations is still product-full-bound, add no-default dependency checks, enforce explicit feature owners for optional runtime dependencies, guard product entry crates so their bitfun-core dependency keeps explicit product-full assembly, scan product entry manifests for new direct core dependency coverage, and keep owner crate default/product-full feature graphs explicit.
feat(acp): display session context usage
A floating button appears when the user scrolls up from the message list. Tapping it smoothly scrolls back to the latest messages. Uses programmaticScrollRef guard and ref-based change detection to suppress button flicker during auto-scroll (new messages, streaming updates).
fix(assistant-config): correct assistant template config saving to wrong mode
Move ToolUseContext runtime/service helpers out of the tool framework into a core-owned runtime binding module. Centralize pipeline, manifest description, and write preflight context materialization there as well, while preserving existing custom_data, cancellation, path policy, runtime URI, remote containment, and Deep Review hook behavior behind focused regressions and boundary checks.
…CWing#843) Group models by provider_instance_id instead of display name alone, migrate legacy configs sharing a base URL, preserve custom provider names, support multi-select custom model entry, and tolerate missing created in OpenAI SSE chunks.
…CWing#844) The collapsible section used grid 0fr collapse but kept overflow visible on the inner wrapper, so hidden session rows still painted over workspace items. Clip the inner content and disable pointer events when collapsed.
- Add shared persistence for oversized tool results. - Exempt Read and image outputs from persistence. - Enforce per-tool and per-round result budgets. - Persist raw Bash output and add remote stdout/stderr budgeting. - Include line counts in persisted-output previews.
…ce file transfer improvements (GCWing#848) * feat(goal) * fix(core) * fix(model) * fix
…CWing#849) Save flow previously cleared requirement probes and raced with acp-clients-changed reloads, which could mark already-enabled agents as failed. Refresh probes after save, suppress self-triggered reloads, and ignore transient probe timeouts for enabled agents.
GCWing#850) Insert a local goal_pending turn immediately after /goal is sent so users see loading feedback during AI goal synthesis instead of a frozen chat UI.
…ile tool guidance (GCWing#852) Emit goal verification lifecycle events for /goal mode UI feedback, cancel orphaned subagent executions when parent turns are dropped or cancelled, and surface structured guidance hints for file read/edit/write guardrails.
…read budget. (GCWing#853) Raise the default Read output cap and align tool-result storage so agents see enough context before Edit, and tighten prompts plus not-found diagnostics to require verbatim copy from Read results.
- move workspace context and dynamic tool catalogs out of env info and tool descriptions - centralize request context section rendering in the prompt builder, including collapsed tool guidance - simplify request context policy defaults and require agents to declare sections explicitly - refactor request context construction toward chainable, easier-to-maintain composition - update tool and prompt tests to cover the new request context layout
- persist user-managed related directories for local and remote workspaces - add workspace UI for creating, editing, removing, and displaying related directories - inject related directories into agent request context and prompt workspace context - propagate workspace id through agentic session creation to resolve related context reliably - support optional related directory descriptions and improve remote picker stacking behavior
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
9 participants
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.