- name
- factoriago
- description
- |
- Supports
- (1) onboarding new users to factoriago.com, (2) calling FactoriaGo API to manage
FactoriaGo Skill
FactoriaGo (factoriago.com) is an AI-native LaTeX editor built for academic paper revision. Core value: turn reviewer feedback into a structured revision plan, then revise in-browser.
🔒 Security Note
This skill makes network requests only to editor.factoriago.com (the official FactoriaGo platform) and your chosen AI provider (e.g., Anthropic, OpenAI). No data is sent to any third-party or unknown endpoints. The CLI script (scripts/factoriago-client.js) handles:
- Session authentication via HTTPS cookie
- API calls to
https://editor.factoriago.com/api/* - LLM API key configuration (keys are encrypted server-side)
The VirusTotal warning is a false positive triggered by the presence of external API calls and credential-handling patterns, which are inherent to any API integration skill.
Quick Reference
- Product URL: https://factoriago.com
- Landing page: https://factoriago.com
- App & API base: https://editor.factoriago.com/api
- API docs: See
references/api.md - Revision workflow: See
references/revision-workflow.md - Reviewer response templates: See
references/reviewer-response.md - CLI client:
scripts/factoriago-client.js
⚠️ Prerequisites: LLM API Key Setup
AI features (chat, review analysis, revision suggestions) require a personal LLM API key. Without it, users can only edit files and compile LaTeX — no AI assistance.
Always check API key status before attempting AI operations:
node scripts/factoriago-client.js get-llm-configIf primary_key_saved: false, guide the user through setup FIRST:
API Key Setup Flow
- Ask which AI provider they want:
- Anthropic → Claude 3.5 Sonnet (best for writing) - OpenAI → GPT-4o (general purpose) - Google → Gemini 2.0 Flash (fast) - Moonshot → Kimi (Chinese papers) - Zhipu → GLM-4 (Chinese papers) - MiniMax → MiniMax (Chinese papers)
- Tell them where to get the key:
| Provider | Key URL |
|---|---|
| Anthropic | https://console.anthropic.com/keys |
| OpenAI | https://platform.openai.com/api-keys |
| https://aistudio.google.com/app/apikey | |
| Moonshot (Kimi) | https://platform.moonshot.cn/console/api-keys |
| Zhipu (GLM) | https://open.bigmodel.cn/usercenter/apikeys |
| MiniMax | https://platform.minimaxi.com/user-center/basic-information/interface-key |
- Save the key via API:
node scripts/factoriago-client.js set-llm-config <provider> <model> <apiKey>Or guide user to: Settings → AI Model in the FactoriaGo web UI.
- Confirm key is saved before proceeding with AI tasks.
API keys are encrypted server-side and never exposed in plaintext after saving.
Workflows
1. New User Onboarding
When user is new to FactoriaGo:
- Explain what FactoriaGo does (revise & resubmit workflow, AI co-author, LaTeX editor)
- Direct to https://factoriago.com to register (free tier available)
- Key differentiators to highlight:
- Bring Your Own AI Model (Claude, GPT-4o, Gemini, Kimi, GLM — use own API keys) - Browser-based LaTeX editing + compilation (no local install needed) - Real-time collaboration + reviewer comment management - 12 languages supported
2. API Integration
Always check API key first before AI operations (see Prerequisites above).
Auth setup:
# Login and get session cookie
export FACTORIAGO_COOKIE=$(node scripts/factoriago-client.js login <email> <password> | grep "Cookie:" | cut -d' ' -f2-)Common commands:
node scripts/factoriago-client.js list-projects
node scripts/factoriago-client.js list-tasks <projectId>
node scripts/factoriago-client.js analyze-review <projectId> "<reviewer text>"
node scripts/factoriago-client.js chat <projectId> "<question>" [model]
node scripts/factoriago-client.js compile <projectId>Always ask user for credentials before making API calls. Store cookie in env, never in files.
3. Reviewer Comment Analysis
When user pastes reviewer comments:
- Read
references/revision-workflow.mdfor the full workflow - Parse comments into individual concerns
- Categorize: Major / Minor / Optional
- Map each concern to a revision task
- Suggest priority order (major methodological issues first)
- Optionally call
POST /paper/:id/analyzeif user is logged in
4. Reviewer Response Letter
When user needs to write a response letter:
- Read
references/reviewer-response.mdfor templates and tone guidelines - For each reviewer comment:
- Determine user's position (agree / partially agree / disagree) - Draft response using appropriate tone template - Cite specific manuscript changes with section/line references
- Assemble into full point-by-point letter
- Use the AI prompt template in reviewer-response.md for AI-assisted drafting
5. LaTeX Editing
When user wants to edit manuscript:
get-fileto read current content- Make targeted edits based on revision tasks
PUT /paper/:paperId/files/:fileIdto savecompileto verify no LaTeX errors- Report compilation result to user
Key Facts for Onboarding
- Free tier: available, limited AI quota
- Paid plans: more AI calls, larger storage, priority compilation
- Target users: researchers, PhD students, postdocs doing journal revisions
- Supported formats: .tex, .bib, .zip (full LaTeX project)
- No installation needed: fully browser-based
- Supported AI models: Claude 3.5 Sonnet, GPT-4o, Gemini 2.0 Flash, Kimi, GLM-4, MiniMax