fix(research): Remove address truncation and add continuous pipes to horizontal pipeline

BREAKING CHANGES:
- Full 44-char Solana addresses now displayed (never truncated)
- Continuous vertical pipes with NO gaps between metadata
- Increased depth support from 4 to 7 levels
- Wider canvas (400x200) for better horizontal layout

Changes:
1. Removed address truncation in render_horizontal_pipeline() (line 749)
2. Set address_truncate_length: 0 in render config (line 1744)
3. Expanded canvas width to 400 and box width to 50 chars
4. Rewrote pipe rendering logic to draw pipe on EVERY line (line 771)
5. Increased column count from 4 to 7 for deeper graphs

Compliance:
- CLAUDE.md rule: "NEVER truncate wallet addresses" - now compliant
- Blockchain forensics requirement: full addresses for verification

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

Author: 0xrinegade

plan_fix.md

@@ -0,0 +1,130 @@
+# Plan to Fix OSVM Research Agent Output
+
+## User's Requirements (Re-read Analysis):
+
+1. **"where is beatiful chart like this?"**
+   - User wants the HORIZONTAL PIPELINE renderer (from `/tmp/pipeline_with_metadata.rs`)
+   - NOT the simple depth-based sections
+   - Want the ASCII art with arrows flowing LEFT→RIGHT like the prototype showed
+
+2. **"why you expand only surface level?"**
+   - Currently showing depth-based sections separately
+   - User wants TRUE horizontal expansion with wallets side-by-side
+   - Like columns in a table, not separate sections
+
+3. **"stop printing out ovsm out"**
+   - Don't show raw OVSM LISP code execution in output
+   - Save it to file instead
+
+4. **"save it to file, not to the output"**
+   - All OVSM execution details → file
+   - Only show final results to user
+
+5. **"in ai call specify maxTokens: 128"**
+   - Make AI responses more concise
+   - Add maxTokens parameter to AI service calls
+
+6. **"it gives us markdown - you must render markdown"**
+   - AI responses come as markdown
+   - Need to render markdown to terminal (colors, formatting)
+
+7. **"all osvm output must be written to the log file"**
+   - Path: `~/.osvm/logs/log-{unixtimestamp}.log`
+   - ALL events, output, execution details
+   - Keep logging until program exits
+
+## Tasks to Implement:
+
+### Task 1: Fix Horizontal Pipeline Renderer ✅ HIGH PRIORITY
+**Problem**: Current implementation shows depth sections VERTICALLY, not true horizontal pipeline
+**Solution**:
+- Modify `render_horizontal_pipeline()` to render wallets SIDE-BY-SIDE at same depth
+- Create true LEFT→RIGHT flow like the prototype
+- Use columnar layout with proper ASCII art arrows
+
+**Files to modify**:
+- `src/services/research_agent.rs` - Fix renderer to be truly horizontal
+
+### Task 2: Add Logging System ✅ HIGH PRIORITY
+**Problem**: No centralized logging to file
+**Solution**:
+- Create `~/.osvm/logs/` directory
+- Generate log filename with Unix timestamp
+- Redirect all output to log file
+- Use `tracing` or `log` crate with file appender
+
+**Files to modify**:
+- `src/main.rs` - Initialize logging on startup
+- `src/services/research_agent.rs` - Log all OVSM executions
+
+### Task 3: Hide OVSM Execution Output ✅ MEDIUM PRIORITY
+**Problem**: OVSM LISP code shown in terminal
+**Solution**:
+- Don't print OVSM execution to stdout
+- Only log to file
+- Show only final results to user
+
+**Files to modify**:
+- `src/services/ovsm_service.rs` or wherever OVSM is executed
+- Capture stdout/stderr and redirect to log
+
+### Task 4: Add maxTokens to AI Calls ✅ MEDIUM PRIORITY
+**Problem**: AI responses too verbose
+**Solution**:
+- Add `max_tokens: 128` parameter to AI API calls
+- Make responses more concise
+
+**Files to modify**:
+- `src/services/ai_service.rs` - Add maxTokens parameter
+- `src/utils/ai_client.rs` - Pass parameter to API
+
+### Task 5: Render Markdown in Terminal ✅ MEDIUM PRIORITY
+**Problem**: AI returns markdown, we show raw text
+**Solution**:
+- Use `termimad` or `bat` crate to render markdown
+- Convert markdown to ANSI-colored terminal output
+
+**Files to modify**:
+- `src/services/research_agent.rs` - Render AI responses as markdown
+- Add markdown rendering utility
+
+## Implementation Order:
+
+1. **FIRST**: Fix horizontal pipeline renderer (user's main complaint)
+2. **SECOND**: Add logging system (infrastructure for other changes)
+3. **THIRD**: Hide OVSM output + redirect to logs
+4. **FOURTH**: Add maxTokens parameter
+5. **FIFTH**: Render markdown
+
+## Expected Result:
+
+When user runs `osvm research --agent ADDRESS`:
+- ✅ Beautiful horizontal ASCII chart with LEFT→RIGHT flow
+- ✅ No OVSM execution output visible
+- ✅ All details logged to `~/.osvm/logs/log-{timestamp}.log`
+- ✅ Concise AI responses (max 128 tokens)
+- ✅ Markdown rendered with colors/formatting
+- ✅ Clean, professional terminal output
+
+## Self-Check Questions:
+
+- ❓ Do I understand what "horizontal" means? → YES: Side-by-side wallets, not vertical sections
+- ❓ Do I know what output to hide? → YES: OVSM LISP execution details
+- ❓ Do I know where to log? → YES: `~/.osvm/logs/log-{unixtimestamp}.log`
+- ❓ Do I know what maxTokens does? → YES: Limits AI response length
+- ❓ Do I know what markdown rendering means? → YES: Convert `**bold**` to ANSI colors
+
+## Files to Read/Modify:
+
+1. `/home/larp/larpdevs/osvm-cli/src/services/research_agent.rs` - Fix renderer
+2. `/home/larp/larpdevs/osvm-cli/src/main.rs` - Add logging init
+3. `/home/larp/larpdevs/osvm-cli/src/services/ai_service.rs` - Add maxTokens
+4. `/home/larp/larpdevs/osvm-cli/Cargo.toml` - Add logging/markdown deps
+5. `/tmp/pipeline_with_metadata.rs` - Reference for beautiful chart
+
+## Notes:
+
+- User emphasized "read this message again" - did that ✅
+- User wants self-asking - doing that now ✅
+- User wants plan in plan_fix.md - writing this ✅
+- User's tone suggests frustration with current output - prioritize visual fixes