agentsclimarketplace
Sign in

Gobbler setup

Skill Enablement-Engineering/gobbler/skills/gobbler-setup

Installs, configures, and troubleshoots Gobbler. Triggers on install gobbler, setup, not working, connection refused, Docker/service failures, ffmpeg problems, browser relay failures, or diagnostic requests.From its SKILL.md

Install
npx -y skills add Enablement-Engineering/gobbler --skill gobbler-setup

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.0 KB, as published. Nobody here has run it

Gobbler setup and troubleshooting

Use this Skill for installation, configuration, dependency checks, Docker services, the SQLite worker, and browser-relay problems.

Start with diagnostics

gobbler --version
gobbler doctor --json
gobbler status --json
  • doctor --json is the broad setup report: Python, ffmpeg, Docker, config, and conversion services.
  • status --json focuses on provider readiness and YouTube fallback/proxy state.
  • status --json can emit a useful JSON body and exit nonzero when degraded.

Do not assume every degraded service blocks every workflow. YouTube needs no local service; local audio needs ffmpeg; documents need Docling; web pages need Crawl4AI.

Install

Requirements: Python 3.11+ and uv. ffmpeg is needed for audio/video. Docker is needed for documents/web pages.

git clone https://github.com/Enablement-Engineering/gobbler.git
cd gobbler
uv sync
uv run gobbler --version
uv run gobbler doctor --json

Optional isolated global tool:

uv tool install .
gobbler --version

After pulling updates:

uv tool install . --force

Docker services

The Compose stack has Docling and Crawl4AI only:

make start-docker
docker compose ps
gobbler doctor --json

Default endpoints:

  • Docling: http://localhost:5001
  • Crawl4AI: http://localhost:11235

Inspect failures:

docker compose logs --tail 100 docling
docker compose logs --tail 100 crawl4ai

Do not tell users Redis or RQ is required. The current queue is SQLite-backed.

Configuration

Path:

~/.config/gobbler/config.yml

Commands:

gobbler config path
gobbler config show --format json
gobbler config init

Edit the YAML file directly; the current CLI has no config set or config validate command. There is no universal environment override system and GOBBLER_CONFIG does not select another config file. Current integration variables include:

  • TRANSCRIPTAPI_KEY
  • WEBSHARE_USER, WEBSHARE_PASS, YOUTUBE_PROXY
  • CRAWL4AI_PROXY
  • OPENAI_API_KEY
  • CRAWL4AI_API_TOKEN for Compose, paired with config
  • GOBBLER_MODELS_PATH for the Docling cache mount

Do not print secrets. doctor --json redacts sensitive config values, but inspect any report before publishing it.

Common recovery paths

Docker unavailable

docker info
# macOS/Colima
colima start --cpu 5 --memory 10
make start-docker

Docling disconnect or OCR memory pressure

gobbler document digital.pdf --no-ocr -o output.md
docker compose logs --tail 100 docling

Crawl4AI unavailable

docker compose up -d crawl4ai
gobbler doctor --json
gobbler webpage "https://example.com" --no-proxy -o /tmp/example.md

Local audio failure

ffmpeg -version
gobbler audio recording.mp3 --model tiny -o recording.md

Current faster-whisper integration requests CPU execution through CTranslate2. Do not promise Metal, CoreML, or CUDA selection from the CLI.

YouTube rate limit or IP block

gobbler status --json
gobbler youtube "URL" --language auto

Then wait/retry, configure a proxy, or set TRANSCRIPTAPI_KEY for the documented fallback. Never preserve credential-bearing URLs in notes or issue reports.

Queue worker

gobbler jobs worker status
gobbler jobs worker start
gobbler jobs list

Queueing is explicit, for example:

gobbler batch webpages urls.txt -o ./pages --queue --json

Do not claim queue.auto_queue_threshold currently auto-queues commands. gobbler daemon is a browser-relay compatibility wrapper, not the job worker.

Browser extension

gobbler relay restart
gobbler browser status
gobbler browser list --json
gobbler browser inject

Command guards target tabs matching the extension's stored group ID. A different group with the same title does not match, while manually moving a tab into the existing managed group makes it eligible. Origin permissions gate extraction/page-API scripting, but debugger-based browser exec only checks group eligibility. Third-party AI chat adapters are DOM-dependent; generic browser success does not prove NotebookLM/Claude/ChatGPT/Gemini selectors remain current.

Verification before reporting success

Run the smallest real workflow relevant to the user's issue:

# YouTube
# gobbler youtube "USER_URL" -o /tmp/test.md

# Document
# gobbler document USER_FILE --no-ocr -o /tmp/test.md

# Webpage
# gobbler webpage "https://example.com" --no-proxy -o /tmp/test.md

Verify the command exit code and output file. Do not infer readiness from a container being listed alone.

Full reference: https://enablement-engineering.github.io/gobbler/setup-troubleshooting/

Keep looking

Skills are one crate of 627,801. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.