scbe-colab-bridge
Control Google Colab notebooks from Claude Code via Chrome extension. Execute cells, run terminal commands, read outputs, and manage GPU compute remotely.
Best use case
scbe-colab-bridge is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Control Google Colab notebooks from Claude Code via Chrome extension. Execute cells, run terminal commands, read outputs, and manage GPU compute remotely.
Teams using scbe-colab-bridge should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/scbe-colab-bridge/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How scbe-colab-bridge Compares
| Feature / Agent | scbe-colab-bridge | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Control Google Colab notebooks from Claude Code via Chrome extension. Execute cells, run terminal commands, read outputs, and manage GPU compute remotely.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# SCBE Colab Bridge
Execute code on Google Colab from the Claude Code terminal via the Chrome extension bridge.
## Prerequisites
1. Chrome is open with Claude-in-Chrome extension connected
2. A Colab notebook is open in a Chrome tab
3. The Colab runtime is connected (green checkmark visible)
## Capabilities
### Terminal Commands (PREFERRED — handles multi-line code)
1. Click into the Colab terminal panel (right sidebar)
2. Type any bash/python command
3. Press Enter
4. Read output via screenshot
```
# Simple command
python3 -c "import torch; print(torch.__version__)"
# Write a script file then run it (avoids indentation issues)
cat > /content/my_script.py << 'EOF'
import torch
# ... multi-line Python with proper indentation
EOF
python3 /content/my_script.py
```
### Cell Execution (single-line only)
1. Click `+ Code` button at (170, 84) in toolbar
2. Click into the new cell editor
3. Type single-line Python with semicolons (NO indentation)
4. Press Ctrl+Enter
5. Read output via JS: `document.querySelectorAll('.output_text')`
### Cell Search (viewport-limited)
Inject ColabBridge then search:
```javascript
window.ColabBridge.findCell('search text') // only finds rendered cells
window.ColabBridge.readOutput(cellIndex) // read cell output
window.ColabBridge.runtimeStatus() // check runtime
```
### Full Notebook Index
Use local copy at `notebooks/spiralverse_protocol_training_generator.ipynb`
Cell index at `artifacts/colab_bridge/notebook_cell_index.json`
## Known Limitations
| Issue | Workaround |
|-------|-----------|
| Multi-line code in cells gets wrong indentation | Use terminal or write to .py file first |
| Cell search only finds viewport-rendered cells | Use local notebook index for full search |
| Focus can shift from terminal to notebook editor | Always click terminal panel before typing |
| Colab lazy-loads cells | Scroll to cell before reading via JS |
| No GPU on free tier by default | Use Runtime > Change runtime type > GPU |
## Proven Bridge Pattern
```
1. mcp__claude-in-chrome__tabs_context_mcp (get tab ID)
2. mcp__claude-in-chrome__navigate (open Colab URL)
3. mcp__claude-in-chrome__computer screenshot (verify state)
4. mcp__claude-in-chrome__computer left_click (click terminal)
5. mcp__claude-in-chrome__computer type (enter command)
6. mcp__claude-in-chrome__computer key Enter (execute)
7. mcp__claude-in-chrome__computer wait 5 (wait for output)
8. mcp__claude-in-chrome__computer screenshot (read result)
```
## Key Cells in Current Notebook
| Cell | Content |
|------|---------|
| 82 | H1-B: Raw Q/K/V weight FFT (BREAKTHROUGH) |
| 83 | Mirror Differential Telemetry |
| 84 | Thermodynamic Mirage Spectral Mapping |
| 17 | Spiralverse SDK patent validation |
| 18 | Harmonic Cryptography d^2 test |
| 10 | SCBE Agent System demo |
## Training Data Generation
Every Colab bridge interaction generates potential SFT data:
- Command sent -> output received = (instruction, response) pair
- Error encountered -> fix applied = (problem, solution) pair
- Cell executed -> result analyzed = (task, completion) pair
These can be exported to `training/sft_records/` for model training.Related Skills
scbe-training-pair-authoring
Create prompt and response and metadata training pairs from SCBE documents, repair traces, terminal sessions, and operational workflows using the repository's canonical dataset contract and provenance rules.
scbe-spin-conversation-engine
Generate SFT training data via radial matrix conversation pivots with D&D-style combat research mode. Produces diverse, cost-effective training pairs with Sacred Tongue encoding, golden spiral problem distribution, and harmonic re-attunement.
scbe-research-training-bridge
Stage arXiv evidence and Obsidian markdown into source-grounded Hugging Face training bundles for research, review, and later SFT runs.
scbe-document-management
Consolidate overlapping docs, classify files by authority, and keep SCBE repo documents aligned with runtime truth. Use when the repo has drift between canonical docs, public docs, proposal notes, research branches, and generated evidence.
scbe-claim-to-code-evidence
Map SCBE Notion technical claims, proof pages, and patent-facing architecture notes to concrete repository evidence such as code paths, tests, demos, and docs. Use when Codex needs to build a due-diligence packet, claim-to-code audit, implementation crosswalk, patent support note, or proof summary from local Notion exports and repo artifacts.
scbe-autonomous-worker-productizer
Turn SCBE automation, autonomous worker, and revenue-system notes into concrete offers, workflow packs, pilot plans, or SaaS-facing product packets. Use when Codex needs to package Notion automation pages into buyer-ready offerings, n8n/Zapier workflow designs, flock-backed worker systems, or implementation roadmaps tied to existing SCBE repo surfaces.
scbe-code-scanning-ops
Operate GitHub code scanning and CodeQL remediation for SCBE repositories. Use when triaging code-scanning alerts, mapping alert classes to fix patterns, validating targeted regressions, or wiring dedicated CodeQL workflows and runbooks into the repo.
scbe-world-anvil-lore-rag-7th-tongue
Build and operate a lore-focused RAG system using World Anvil exports and SCBE docs, with deterministic Claude/Codex cross-talk packets for handoff. Use when users ask to structure lore canon retrieval, sync worldbuilding data, enforce citation-grounded generation, or coordinate a 7th Tongue overseer lane across multiple AI agents.
scbe-webtoon-book-conversion
Convert The Six Tongues Protocol and related manuscript sections into webtoon/manhwa storyboard packets, episode roadmaps, panel expansion plans, and image-generation-ready prompt lanes. Use when extending the series storyboard, adapting book chapters into vertical scroll episodes, or keeping art generation tied to canon instead of drifting into generic fantasy panels.
scbe-voice-render-verification
Govern and verify SCBE voice rendering work that maps Langues weighting into breath, phase, and Layer 14 audio-axis packets. Use when implementing or reviewing `scripts/voice_gen_hf.py`, emitting sidecar voice packets, validating canonical tongue ordering, tuning breath planning or phase timing, or keeping voice docs and code aligned with `docs/LANGUES_WEIGHTING_SYSTEM.md` and `docs/specs/SCBE_VOICE_EMOTIONAL_TIMBRE_SYSTEM.md`.
scbe-universal-synthesis
Orchestrate all installed Codex skills through an auto-updating synthesis matrix with Sacred Tongues routing, emotion/intent metadata, and decodable lexicon packets tied to established SCBE characters. Use when the user asks for cross-skill coordination, auto skill updates, multi-skill routing, or Sacred Tongues intent mapping.
scbe-system-engine
Coordinate SCBE-AETHERMOORE math, automation, and service connectors so multi-agent systems can execute work end-to-end. Use when tasks require SCBE dimensional validation, AI-to-AI workflows, browser automation planning, Hugging Face or GitHub/Linear/Notion/Zapier integration, or self-improving skill updates.