Skip to content

faq.en.md

Latest commit

 

History

History
70 lines (44 loc) · 2.01 KB

faq.en.md

File metadata and controls

70 lines (44 loc) · 2.01 KB
Raw

FAQ

This FAQ reflects the current repository state.

What is already production-like today?

The strongest parts today are the runtime platform pieces and the stateful research foundation:

  • app bootstrapping and control-plane APIs
  • channels, providers, skills, MCP, automation, and the console
  • project/workflow/task state
  • claim/evidence graph
  • experiment tracking, result-bundle ingestion, and blocker remediation

What is still incomplete is the higher-level research quality layer: evidence matrix scoring, stronger claim validators, richer external execution adapters, and submission packaging.

Which providers are supported?

Provider types in code today are openai, anthropic, gemini, ollama, dashscope, deepseek, minimax, other, and custom.

Which channels are built in?

console, telegram, discord, dingtalk, feishu, imessage, qq, and voice.

Where is my data stored?

  • working data: ~/.researchclaw
  • secrets: ~/.researchclaw.secret

The secret dir stores envs.json and providers.json.

The research workflow state is stored under ~/.researchclaw/research/state.json unless you override the research path.

Why does / show Console not found?

The backend can start without prebuilt console assets. Build them once:

cd console
npm install
npm run build

How do I secure automation triggers?

Set RESEARCHCLAW_AUTOMATION_TOKEN on the server. Requests must send the same token via Authorization: Bearer <token>, x-researchclaw-token, or x-researchclaw-automation-token.

Why can plain pytest fail in a fresh checkout?

If the package is not installed editable yet, Python may not find researchclaw. Either run:

pip install -e ".[dev]"

or:

PYTHONPATH=src pytest -q

Does the console include research workflow controls?

Yes. The current console includes a dedicated Research page with:

  • project dashboards
  • workflow execution
  • execution health
  • recent blockers
  • remediation drill-down and batch actions