summaryrefslogtreecommitdiff
path: root/Omni
AgeCommit message (Collapse)Author
2025-12-02Remove pyproject=true, use format=setuptools, add toggle_episode_publicBen Sima
2025-12-02Fix Admin.py imports for bild local dep detectionBen Sima
- Change 'from Biz.X import Y' to 'import Biz.X as X' style - bild only recognizes 'import X as Y' for local dep detection - Add setuptools to Python deps (required by newer nixpkgs) Amp-Thread-ID: https://ampcode.com/threads/T-fe8328a9-7709-4544-9d31-b099f04aa120 Co-authored-by: Amp <amp@ampcode.com>
2025-12-02Improve agent system prompt for better token efficiencyBen Sima
Add explicit guidance on: - Reading files with large ranges (500+ lines) instead of many small chunks - Using read_file directly when target file is known vs search_and_read - Cost awareness: planning refactors, avoiding redundant reads - Tool call limits for complex tasks
2025-12-02Remove duplicate prompt contentBen Sima
The task was being added to the prompt twice, once in the base prompt and once in the user prompt.
2025-12-02System prompt improvementsBen Sima
Worked with Gemini and Opus to improve the system prompt with learnings from the Amp prompt. Removed reference to Omni/Task/README.md because it is deprecated in favor of `jr task`.
2025-12-02jr: add 'prompt' command to inspect agent system promptBen Sima
jr prompt <task-id> constructs and prints the full system prompt that would be sent to the agent, including: - Agent configuration (model, cost budget) - Base instructions - AGENTS.md content - Relevant facts from knowledge base - Retry/progress context if applicable Useful for debugging agent behavior and token usage.
2025-12-01fix: lint Admin.py after refactorBen Sima
2025-12-01Compact amp-style timeline rendering and targeted file readingBen Sima
Timeline tool display: - Grep/search: ✓ Grep pattern in filepath - Read file: ✓ Read filepath @start-end - Edit file: ✓ Edit filepath - Bash: ϟ command (lightning bolt prompt) - Tool results only shown for meaningful output New search_and_read tool: - Combines search + read in one operation - Uses ripgrep --context for surrounding lines - More efficient than separate search then read Worker prompt updated to prefer search_and_read over separate search + read_file calls
2025-12-01Fix autoscroll toggle not being respectedBen Sima
timelineScrollScript now checks autoscrollEnabled before scrolling
2025-12-01Make complexity badge editable on task detail pageBen Sima
- Add ComplexityForm and ComplexityBadgePartial types - Add /tasks/:id/complexity POST endpoint - Add complexityBadgeWithForm component with dropdown - Add complexity dropdown JS for keyboard navigation - Add CSS styles for complexity dropdown - Always show complexity badge (Set Complexity if none) Task-Id: t-219
2025-12-01Fix lint and build errors in Web componentsBen Sima
- Components.hs: Fix escaped quotes in renderFormattedJson - Style.hs: Use overflowWrap instead of wordBreak for breakWord value
2025-12-01Fix build errors in Jr modulesBen Sima
- Fix Worker.hs to use EngineError instead of tuple - Fix Types.hs imports for LazyText.encodeUtf8 and dayOfWeek - Remove duplicate SortOrder from Components.hs (import from Types.hs) - Add orphan instance pragmas to Pages.hs and Partials.hs - Clean up unused imports
2025-12-01Fix timeline partial to include cost/token metrics and controlsBen Sima
The HTMX-refreshed AgentEventsPartial was missing: - Cost/token summary in header - Live toggle button - Autoscroll toggle button - Comment form Now matches the full page renderUnifiedTimeline output.
2025-12-01Add lint warning for large files (>1000 lines)Ben Sima
Large files cause agent token bloat and edit_file failures. This adds a warning (not error) to encourage splitting. Task-Id: t-228
2025-12-01Refactor Web.hs into smaller modulesBen Sima
Split 3231-line Web.hs into focused submodules: - Types.hs (346 lines): Data types, forms, API definition - Components.hs (1464 lines): Reusable UI components - Pages.hs (866 lines): Full page ToHtml instances - Partials.hs (247 lines): HTMX partial ToHtml instances - Handlers.hs (642 lines): Servant handler implementations - Web.hs (39 lines): Main module with run function Reduces main file by 99%, preventing agent token bloat. Task-Id: t-226 Amp-Thread-ID: https://ampcode.com/threads/T-355fae3a-03e9-4bdb-a1c7-6132576bf601 Co-authored-by: Amp <amp@ampcode.com>
2025-12-01Strengthen prompt to stop immediately after tests passBen Sima
Jr was completing tasks but then going into verification loops, re-reading files and 'tracing through logic' after tests passed. This burned ~4 cents of extra cost on t-221. Made instructions more emphatic: - 'STOP IMMEDIATELY' with explicit list of what NOT to do - 'ANY further tool calls are wasted money' - Repeated in BUILD SYSTEM NOTES section Task-Id: t-227
2025-12-01Fix Recent Activity duplicating tasks on refreshBen Sima
Changed taskToUnixTs from round to ceiling to prevent sub-second timing issues where a task updated at e.g. 10:00:00.3 would round down to 10:00:00, causing the next poll with since=10:00:00 to include it again. Task-Id: t-221
2025-12-01Sort ready queue by priorityBen Sima
getReadyTasks now returns tasks sorted by: 1. Priority (P0 first, then P1, P2, etc.) 2. Creation time (oldest first within same priority) This ensures high-priority items are always at the top of both the CLI 'jr task ready' output and the web dashboard. Task-Id: t-220
2025-12-01Add prompt guidance for large file editingBen Sima
Instructs the agent to: - Use line ranges when reading large files (>500 lines) - Use minimal context for edit_file old_str matching - Re-read exact lines after failed edits - Stop after 2-3 failed edits to reconsider approach - Flag very large files (>2000 lines) for refactoring Task-Id: t-225
2025-12-01Add guardrail for repeated edit_file failuresBen Sima
Tracks 'old_str not found' errors from edit_file tool calls. After 5 consecutive failures, stops the agent to prevent burning tokens on impossible edits. This catches the pattern where the agent repeatedly tries to edit a large file with incorrect old_str matches, which was the root cause of t-222 exceeding its cost budget. Task-Id: t-224
2025-12-01Scale cost guardrail by task complexityBen Sima
Cost limits by complexity level: - Complexity 1: 50 cents - Complexity 2: 100 cents - Complexity 3: 200 cents (default) - Complexity 4: 400 cents - Complexity 5: 600 cents This prevents low-complexity tasks from burning budget while allowing complex tasks more room for iteration. Task-Id: t-223
2025-12-01Add --complexity flag to task update commandBen Sima
Allows updating complexity along with status in a single command: jr task update t-123 in-progress --complexity=3 Task-Id: t-222
2025-12-01Add autoscroll toggle button for timelineBen Sima
Adds a toggle button next to the LIVE indicator that controls whether the timeline auto-scrolls to new events. Default is ON. - renderAutoscrollToggle button with ⬇ icon - toggleAutoscroll() JS function tracks state - htmx:afterSettle checks autoscrollEnabled before scrolling - Blue styling to differentiate from green LIVE button Task-Id: t-222
2025-12-01Consolidate cost display in timeline header, remove Execution DetailsBen Sima
- Add cost/token summary to timeline header - Filter Cost events from timeline display - Remove Execution Details section (cost info now in header) - Remove unused renderExecutionDetails function Task-Id: t-216
2025-12-01Clicking LIVE label toggles live updates on/offBen Sima
- Add clickable LIVE toggle button that pauses/resumes timeline polling - Green pulsing when active, grey when paused - Uses htmx:beforeRequest event to cancel requests when paused - Increase duplicate tool call guardrail from 20 to 30 Task-Id: t-211
2025-12-01Revert "Agent log box auto-scrolls to bottom on updates"Ben Sima
This reverts commit 9c3034853c3cacea8a77358467acf46e75c982f5.
2025-12-01Show tool call arguments inline instead of JSON blobBen Sima
- Add formatToolCallSummary to extract key argument from JSON - Shows run_bash command, file paths for read/edit/write, patterns for search - Display summary inline in tool call header (e.g., run_bash: `ls -la`) - Increase token guardrail from 1M to 2M to prevent premature stops Task-Id: t-212
2025-12-01Agent log box auto-scrolls to bottom on updatesBen Sima
Simplify auto-scroll behavior to always scroll to bottom after HTMX swap, rather than checking if user is near bottom. This ensures latest messages are always visible as they come in. Task-Id: t-208
2025-12-01Replace HumanTask type with NeedsHelp statusBen Sima
- Remove HumanTask from TaskType enum (now Epic | WorkTask only) - Add NeedsHelp to Status enum for tasks requiring human guidance - Update getReadyTasks to filter NeedsHelp instead of HumanTask - Rename humanTasks to tasksNeedingHelp in HumanActionItems - Add CLI parsing for needs-help status in list/update commands - Add badge styling for NeedsHelp (amber/yellow theme) - Update all status pattern matches in tree view and print functions - Update tests to verify NeedsHelp exclusion from ready queue Task-Id: t-210
2025-12-01Fix task links in Recent Activity inheriting wrong hx-targetBen Sima
Links inside the recent-activity div were inheriting hx-target=#activity-list from the parent, causing page content to be swapped into the list item instead of navigating. Explicitly set hx-target=body on list group items.
2025-12-01Remove editing of commit message in hookBen Sima
This was cutting off jr's commit messages so now I have a bunch of commit messages that are just cut off in the middle. Ugh.
2025-12-01Bug: Recent Activity links and Load More button not working on homepageBen Sima
Good! The tests are passing. My fix addresses the JavaScript error by ma However, this might not fully solve the problem of links not working. Th The issue was that the JavaScript code in the `hx-vals` attribute used ` **Fix:** Changed the JavaScript expression from: ```javascript hx-vals="js:{since: this.dataset.newestTs}" ``` to: ```javascript hx-vals="js:{since: document.getElementById('recent-activity')?.dataset? ``` This uses optional chaining (`?.`) to safely access the data attribute a The tests pass, confirming the fix doesn't break existing functionality. Task-Id: t-176
2025-12-01Show task complexity on /tasks/<id> pageBen Sima
Perfect! The implementation is complete. Let me create a summary of what I have successfully implemented the task complexity indicator on the `/t 1. **Added `complexityBadge` function** in `Omni/Jr/Web.hs` (lines 685-6 - Creates a badge displaying the complexity with the ℂ symbol (comple - Format: "ℂ {1-5}" - Includes a tooltip "Task Complexity (1-5)" - Uses CSS classes for styling with different colors for each complex 2. **Integrated complexity badge into task detail page** in `Omni/Jr/Web - Added conditional rendering that only shows the badge if `taskCompl - Positioned after the priority badge with a separator (·) - Follows the same pattern as other metadata badges 3. **Added CSS styles** in `Omni/Jr/Web/Style.hs` (lines 696-713): - Base `.badge-complexity` style - Individual styles for each complexity level (1-5) with appropriate - Complexity 1: Green (easy task) - Complexity 2: Blue (moderate) - Complexity 3: Amber (medium complexity) - Complexity 4: Darker amber (high complexity) - Complexity 5: Red (very complex) - Used the ℂ (complex numbers) symbol as suggested in the task descripti - Made it non-editable (read-only badge) since complexity is set during - Only displays when complexity is set (handles `Maybe Int` gracefully) - Color scheme follows a gradient from green (easy) to red (complex) - Consistent with existing badge styling patterns in the UI ✅ All tests pass successfully with `bild --test Omni/Jr.hs` The feature is now complete and ready for use. Tasks with a complexity r Task-Id: t-207
2025-12-01Always bild on run.shBen Sima
Otherwise we end up running stale artifacts and that's no good.
2025-12-01Increase duplicate tool call guardrail limit from 5 to 20Ben Sima
The limit of 5 was too aggressive - reading 5 different files while exploring a codebase would trigger the guardrail. 20 allows for legitimate exploration while still catching infinite loops.
2025-12-01Add explicit dark text colors for comment and thought bubblesBen Sima
- Comment bubble: #1f2937 (dark gray) on #f3f4f6 background - Thought bubble: #78350f (dark amber) on #fef3c7 background
2025-12-01Fix timeline dark mode styling and render comments/thoughts as markdownBen Sima
- Add dark mode styles for comment-bubble and thought-bubble - Render comments with renderMarkdown instead of plain text - Render thoughts with renderMarkdown instead of renderTextWithNewlines - Remove unused renderTextWithNewlines function
2025-12-01Fix timeline and metrics HTMX partial issuesBen Sima
- Timeline partial now uses getAllEventsForTask instead of getEventsForSession so all events show after HTMX refresh, not just the latest session - Move Execution Details header outside HTMX swap target so it persists Task-Id: t-214, t-215
2025-12-01Add actor tracking for status changes and use unified timelineBen Sima
- updateTaskStatusWithActor logs status_change events to agent_events - Worker uses Junior actor for status changes - Jr review uses System/Human actors appropriately - CLI task update uses Human actor - Remove task_activity table schema (migrated to agent_events) - addComment now inserts into agent_events with event_type='comment' Task-Id: t-213
2025-12-01Remove separate Agent Log page, consolidate timeline stylesBen Sima
- Rename agentLogScrollScript to timelineScrollScript - Target .timeline-events instead of obsolete .agent-log class - Rename agentLogStyles to timelineEventStyles - Remove obsolete container styles (.agent-log-section, .agent-log-live, .agent-log) - Remove dark mode styles for obsolete classes Task-Id: t-213.6
2025-12-01Add actor column to agent_events tableBen Sima
- Add 'actor' column to agent_events table (human/junior/system) - Add System to CommentAuthor type (reused for actor) - Add SQL FromField/ToField instances for CommentAuthor - Update insertAgentEvent to accept actor parameter - Update all SELECT queries to include actor column - Update Worker.hs to pass actor for all event types - Guardrail events logged with System actor Migration: ALTER TABLE adds column with default 'junior' for existing rows. Task-Id: t-213.1
2025-12-01Fix code block display in task descriptions for light modeBen Sima
Light mode: light gray background (#f8f8f8) with dark text and subtle border Dark mode: dark background (#1e1e1e) with light text Previously used dark theme for both modes which had poor contrast in light mode. Task-Id: t-206
2025-12-01Make Result sections collapsible in Agent Log (collapsed by default)Ben Sima
Wrap entire tool result in a <details> element so it starts collapsed. User can click to expand and see full output. Task-Id: t-205
2025-12-01Add guardrails and progress tracking to Jr agentBen Sima
Implement runtime guardrails in Engine.hs: - Cost budget limit (default 200 cents) - Token budget limit (default 1M tokens) - Duplicate tool call detection (same tool called N times) - Test failure counting (bild --test failures) Add database-backed progress tracking: - Checkpoint events stored in agent_events table - Progress summary retrieved on retry attempts - Improved prompts emphasizing efficiency and autonomous operation Worker.hs improvements: - Uses guardrails configuration - Reports guardrail violations via callbacks - Better prompt structure for autonomous operation Task-Id: t-203
2025-12-01Render task comments as markdown in web viewBen Sima
Use renderMarkdown for comment text instead of plain text rendering. Comments now support formatting, code blocks, lists, etc. Task-Id: t-204
2025-12-01Jr -> Junior headerBen Sima
2025-12-01Improve Jr agent structure with progress file and incremental workflowBen Sima
Perfect! All changes are in place and working correctly. Let me create a I have successfully implemented the improvements to Jr Worker agent stru 1. **Progress File Tracking** - Added `readProgressFile` function to read `_/llm/${taskId}-progress - Added `buildProgressPrompt` function to include progress context in - Modified `runWithEngine` to load and include progress at the start 2. **Incremental Workflow Enforcement** - Updated base prompt to explicitly instruct: "Pick ONE specific chan - Added "INCREMENTAL WORKFLOW (IMPORTANT)" section with clear guidanc - Added instruction to write progress after each change - Emphasized that tasks may be run multiple times to complete all cha 3. **Explicit Verification** - Maintained existing requirement to run `bild --test` before complet - Added instruction to save progress only after tests pass - Clarified that code must be left in clean, testable state 4. **Avoid Redundant Testing** - Updated BUILD SYSTEM NOTES to clarify running `bild --test` on name - Added explicit instruction not to re-run tests unless more changes - Explained that bild handles dependencies transitively - `bild --test Omni/Agent/Worker.hs` - **PASSED** ✓ - `lint Omni/Agent/Worker.hs` - **NO ISSUES** ✓ - `_/llm/t-203-progress.md` - Progress file documenting this implementat - `_/llm/t-203-implementation-summary.md` - Detailed summary of changes The implementation follows industry best practices from Anthropic, OpenA - Reduced token usage through focused, incremental changes - Better code quality with isolated, tested changes - Improved reliability with progress tracking across sessions - Clear workflow preventing "declaring victory" too early Task-Id: t-203
2025-12-01Improve Worker.hs prompt to avoid redundant test/lint runsBen Sima
Perfect! The changes are exactly what we need. The diff shows that I've 1. ✅ `bild --test` on a namespace tests all its dependencies - no need t 2. ✅ Don't re-run tests after they already passed 3. ✅ Lint runs automatically via git hooks, so manual runs are optional I have successfully completed task t-202. The changes made to `Omni/Agen 1. **Explaining dependency handling**: Added explicit note that `bild -- 2. **Preventing redundant testing**: Added instruction to not re-run tes 3. **Clarifying lint automation**: Explained that lint runs automaticall All tests pass (`bild --test Omni/Agent/Worker.hs`) and all lint checks Task-Id: t-202
2025-12-01Fix agent event content double-encoding in web UIBen Sima
Excellent! The changes have been successfully applied. Let me create a s The issue was that agent event content was being double-encoded in the w 1. **ToolResult events** showed raw JSON like `{"output":"Replaced 1 occ 2. **Assistant messages** showed literal `\n` instead of actual newlines - In `Omni/Agent/Engine.hs` (line 600), tool results are JSON-encoded wh - These JSON strings are stored as-is in the database via `insertAgentEv - The Web UI was displaying these JSON strings directly without decoding - Assistant messages contained literal `\n` escape sequences that weren' I modified `Omni/Jr/Web.hs` with the following changes: 1. **Added import**: `Data.Aeson.KeyMap` to work with JSON objects 2. **Created helper function `renderTextWithNewlines`** (line 2545-2553) - Splits text on literal `\n` sequences - Renders each part with `<br>` tags between them - Used in `renderAssistantEvent` to properly display newlines 3. **Created helper function `renderDecodedToolResult`** (line 2555-2563 - Attempts to decode JSON content - Extracts the `output` field from the JSON object - Falls back to raw content if parsing fails - Used in `renderToolResultEvent` to show clean output instead of raw 4. **Updated `renderAssistantEvent`** (line 2473): - Changed from `Lucid.toHtml truncated` to `renderTextWithNewlines tr 5. **Updated `renderToolResultEvent`** (lines 2502-2503): - Changed both occurrences from `Lucid.toHtml content` to `renderDeco The build now passes successfully with `bild --test Omni/Jr/Web.hs`. Task-Id: t-200
2025-12-01Add author field to task comments (Human vs Junior)Ben Sima
Comments now track whether they were made by a Human or by Junior (the agent). The CommentAuthor type is stored in the database and displayed in the web UI with styled badges. Task-Id: t-201
generated by cgit v1.2.3 (git 2.25.1) at 2025-12-19 14:36:12 +0000