Skip to content

Commit 4f5a5dc

Browse files
author
Brendan Gray
committed
v1.7.3: planning tools, animated spinner, auto-apply code, TodoPanel polish
1 parent 9693efe commit 4f5a5dc

273 files changed

Lines changed: 3314 additions & 45 deletions

File tree

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

‎.continue/rules/new-rule.md‎

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,5 @@
1+
---
2+
description: A description of your rule
3+
---
4+
5+
Your rule content

‎.stfolder/syncthing-folder-066761.txt‎

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,5 @@
1+
# This directory is a Syncthing folder marker.
2+
# Do not delete.
3+
4+
folderID: guide-ide
5+
created: 2026-03-03T18:58:51-05:00

‎.stignore‎

Lines changed: 55 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,55 @@
1+
// guIDE IDE workspace — Syncthing ignore patterns
2+
// Excludes build artifacts, dependencies, and large binary files
3+
// that have no business being synced to the server.
4+
5+
// Node dependencies (every package in the workspace)
6+
node_modules
7+
8+
// Python virtualenv
9+
.venv
10+
11+
// Electron build output — .exe files, installers, unpacked dirs
12+
dist-electron
13+
dist-electron-new
14+
15+
// App build output
16+
build
17+
18+
// Next.js build artifacts (website)
19+
website/.next
20+
website/.next-ready
21+
22+
// LLM model files — can be 2GB–100GB each, NEVER sync these
23+
models
24+
25+
// Log files
26+
*.log
27+
fresh_logs.txt
28+
helper-log.txt
29+
stress-test-log*.txt
30+
31+
// Large text dumps / scratch files
32+
build-output*.txt
33+
build-website*.txt
34+
builder-output*.txt
35+
d2l_*.txt
36+
test-results-*.txt
37+
vitest-results.txt
38+
ts_errors.txt
39+
40+
// Electron-builder cache
41+
.cache
42+
43+
// VS Code local state (not project config)
44+
.vscode/settings.json
45+
.vscode/launch.json
46+
47+
// Python cache
48+
__pycache__
49+
*.pyc
50+
*.pyo
51+
52+
// Misc
53+
New folder
54+
output
55+
set

‎BUG_REPORT.md‎

Lines changed: 46 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,46 @@
1+
# BUG REPORT guIDE Session Audit
2+
3+
> **WORKING ORDER: Chronological, 1 through N. No skipping. No scattering. Fix in order, verify in order.**
4+
5+
***CHECK OFF AS YOU GO!!!!**** CRITICAL CHECK OFF AS YOU FUCKING FIX THEM!>
6+
7+
**Log file:** `%APPDATA%\guide-ide\logs\guide-main.log`
8+
**Session started:** fresh previous bugs cleared, log wiped
9+
**Log reviewed by agent:** not yet (run a fresh test first)
10+
11+
---
12+
13+
## RULES FOR USING THIS FILE
14+
15+
1. Every bug gets a number. Numbers are assigned IN ORDER based on severity/discovery.
16+
2. Agent reads the log FIRST before assigning bugs or discussing anything.
17+
3. User provides front-end observations (screenshots, behavior descriptions).
18+
4. Agent provides backend evidence (log lines, function names, exact cause).
19+
5. No fix is "done" until it is checked off HERE.
20+
6. No skipping ahead. No fixing #5 before #3.
21+
7. If a fix is uncertain (not verified), mark as NEEDS VERIFY, not FIXED.
22+
8. Agent NEVER says "done" without proof. A feature is real and functional, or it is not done.
23+
24+
---
25+
26+
## HOW THIS SESSION WORKS
27+
28+
1. User rebuilds the app
29+
2. User runs through GAUNTLET.md test script
30+
3. User reports front-end observations (what they see, screenshots)
31+
4. Agent reads the fresh log from `%APPDATA%\guide-ide\logs\guide-main.log`
32+
5. Agent populates this file with numbered bugs, evidence, root cause, fix plan
33+
6. User approves fix plan
34+
7. Agent implements, checks off, moves to next IN ORDER
35+
36+
---
37+
38+
## BUG LIST
39+
40+
*(empty awaiting fresh test session)*
41+
42+
---
43+
44+
## FIX LOG
45+
46+
*(empty no fixes this session yet)*

‎Bug Report 24.md‎

Lines changed: 85 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,85 @@
1+
# Bug Report 24
2+
3+
## Agent Rules (Read Before Any Bug Fixes)
4+
- From AGENT_RULES.md and .github/copilot-instructions.md
5+
- Must read these rules before attempting any bug fixes
6+
- Must follow tripwire format: `[Task: <what you're doing> | Last: <what was just completed>]`
7+
- Must test before shipping, no placeholder data, no guessing, no lying
8+
- Must get approval before implementing, always present exact plan
9+
- Never build the app - say "Ready to build" instead
10+
- Never kill all node processes - only kill specific PIDs
11+
- Must check PENDING_FIXES.md at start of every session
12+
13+
## Issues Reported
14+
15+
### 1. JSON Terminal Bubbles Still Appearing (Image 2)
16+
- **Status**: NOT FIXED despite previous claims
17+
- **Previous Fix Attempt**: Tool feedback display modification
18+
- **Fallback Fix**: Additional tool feedback handling
19+
- **Result**: Failed - JSON bubbles still appearing in terminal
20+
- **Evidence**: Multiple JSON bubbles with "Copy" and "Apply" buttons visible in Image 2
21+
22+
### 2. Model Not Acknowledging Requests Before Tool Calls (Image 2)
23+
- **Issue**: Model runs tool calls without acknowledging user request first
24+
- **Issue**: Model doesn't follow up after tool call completion
25+
- **Expected Behavior**:
26+
1. Acknowledge request ("I'll help you find houses in Houston under $200k")
27+
2. Run tool call
28+
3. Follow up with results
29+
- **Actual Behavior**: Direct tool execution without acknowledgment or follow-up
30+
- **Evidence**: Image 2 shows tool call without proper acknowledgment
31+
32+
### 3. Browser Viewport Still Too Narrow (Image 1)
33+
- **Status**: NOT FIXED despite previous viewport fixes
34+
- **Issue**: Browser can be shrunk to very narrow width beyond expected limits
35+
- **Previous Fix Attempt**: Viewport width constraints implementation
36+
- **Result**: Failed - viewport still shrinkable to unusable width
37+
- **Evidence**: Image 1 shows browser content cut off at very narrow width
38+
39+
### 4. New Model Loading Error (Image 4)
40+
- **Error**: "Failed to load 'Llama-3.2-3B-Instruct-Q4_K_S'"
41+
- **Error Details**: "Error invoking remote method 'llm-load-model': reply was never sent"
42+
- **Status**: New error, never seen before
43+
- **Context**: Same model works fine in LM Studio (not a model issue)
44+
45+
### 5. Unexpected Response in New Session (Image 3)
46+
- **Issue**: Strange response when user said "hello" in new ICE session
47+
- **Context**: Same model works fine in LM Studio
48+
- **Status**: Needs investigation
49+
50+
## Previous Failed Fixes Documentation
51+
52+
### JSON Bubbles Fix Attempts:
53+
1. **Primary Fix**: Modified tool feedback display system
54+
- Implementation: Changed how tool outputs are rendered
55+
- Result: Failed - bubbles still appear
56+
57+
2. **Fallback Fix**: Additional tool feedback handling
58+
- Implementation: Added secondary feedback processing
59+
- Result: Failed - bubbles still appear
60+
61+
### Browser Viewport Fix Attempts:
62+
1. **Primary Fix**: Added viewport width constraints
63+
- Implementation: Set minimum width limits in browser component
64+
- Result: Failed - viewport still shrinkable beyond limits
65+
66+
## Required Actions
67+
1. Fix JSON terminal bubbles permanently
68+
2. Implement proper request acknowledgment before tool calls
69+
3. Fix browser viewport minimum width constraints
70+
4. Investigate and fix model loading error
71+
5. Investigate unexpected session responses
72+
6. Test all fixes thoroughly before claiming completion
73+
74+
## Testing Requirements
75+
- Must verify JSON bubbles are completely eliminated
76+
- Must verify proper acknowledgment flow: acknowledge → tool call → follow-up
77+
- Must verify browser viewport cannot be shrunk below usable width
78+
- Must verify model loading works without errors
79+
- Must verify session responses are appropriate
80+
81+
## Notes
82+
- User explicitly stated model works fine in LM Studio
83+
- Do not blame model for issues
84+
- All fixes must work for all hardware configurations (4GB to 128GB GPUs)
85+
- Must follow all agent rules during fix implementation

0 commit comments

Comments
 (0)