---
name: brycewang-stanford-monitor-experiment
description: "Monitor running experiments, check progress, collect results. Use when user says \"check results\", \"is it done\", \"monitor\", or wants experiment output."
license: "NOASSERTION"
metadata:
  decimalai-skill-id: "08688940-932a-442c-b70d-543c9ad452f5"
  decimalai-version-id: "cc94b05e-b215-499c-8020-f57bf1a07f19"
  decimalai-version: "2"
  decimalai-body-sha256: "ad93b08eba15043b092bc652352050cf60c46d4dfb08e2897f55304a585547be"
  decimalai-name: "brycewang-stanford/monitor-experiment"
  decimalai-format: "1"
  decimalai-invocation: "model"
  decimalai-category: "ai-ml"
  decimalai-source: "https://app.decimal.ai/s/brycewang-stanford-monitor-experiment@2/SKILL.md"
---
# Monitor Experiment Results

Monitor: $ARGUMENTS

## Workflow

### Step 1: Check What's Running
```bash
ssh <server> "screen -ls"
```

### Step 2: Collect Output from Each Screen
For each screen session, capture the last N lines:
```bash
ssh <server> "screen -S <name> -X hardcopy /tmp/screen_<name>.txt && tail -50 /tmp/screen_<name>.txt"
```

If hardcopy fails, check for log files or tee output.

### Step 3: Check for JSON Result Files
```bash
ssh <server> "ls -lt <results_dir>/*.json 2>/dev/null | head -20"
```

If JSON results exist, fetch and parse them:
```bash
ssh <server> "cat <results_dir>/<latest>.json"
```

### Step 4: Summarize Results

Present results in a comparison table:
```
| Experiment | Metric | Delta vs Baseline | Status |
|-----------|--------|-------------------|--------|
| Baseline  | X.XX   | —                 | done   |
| Method A  | X.XX   | +Y.Y              | done   |
```

### Step 5: Interpret
- Compare against known baselines
- Flag unexpected results (negative delta, NaN, divergence)
- Suggest next steps based on findings

### Step 6: Feishu Notification (if configured)

After results are collected, check `~/.codex/feishu.json`:
- Send `experiment_done` notification: results summary table, delta vs baseline
- If config absent or mode `"off"`: skip entirely (no-op)

## Key Rules
- Always show raw numbers before interpretation
- Compare against the correct baseline (same config)
- Note if experiments are still running (check progress bars, iteration counts)
- If results look wrong, check training logs for errors before concluding