Engineering Setup Guides

This directory contains setup guides for CLI tools, APIs, and integrations used by Brainforge engineering teams and Cursor AI agents.

Vendor / product index: To browse the same tools by vendor name and find Platform-owned integration packs (e.g. HubSpot), see knowledge/platform/integrations/README.md.


🚀 Start Here: Local Development Environment

New team member? Start with the comprehensive setup guide:

📖 Local Development Environment Setup — Complete macOS setup for Platform, OpenWork, data pipelines, and all engineering workflows. Includes one-liner install script and verification checklist.



Quick Reference

Essential Setup

GuidePurposeRequired For
Local Development EnvironmentComplete macOS CLI toolchain setupAll engineers — Platform, OpenWork, data pipelines
Brainforge Setup SkillAuto-diagnostic + install for MCPs, browser automation, CLIsRun /brainforge-setup to check & install missing tools

Active Integrations (In Use)

GuidePurposeCursor MCP?
1Password CLISecure credential managementNo
Gitleaks pre-commitSecret detection before commitNo
HubSpot access + APIHubSpot app first; MCP read fallback; API for agent writes/notesRead-only MCP fallback; writes via API
Langfuse APIPrompt management & observabilityNo
Cursor skills (submodules)External Cursor skills (e.g. visual-explainer) as submodulesNo
AI Legal Assistant (Cursor + PDF)Contract-review and drafting skills; optional ReportLab PDFNo
Loophole CLIAdversarial ethics / rule stress-test tool (tools/loophole, Anthropic API)No
Linear APITicket/issue managementYes
Calendar Clockify SyncConfigure Calendar → Clockify sync tool env and APIsNo
Google Workspace CLI (gws)Sheets/Drive/Docs/Gmail/Calendar from shell; GTM sheet and OKR workflowsNo
Email agent workflowsAgent SOP for Gmail/Calendar client comms (CLI first, MCP fallback) + standards linkageNo
Supabase Query AccessQuery Internal AI Core, Zoom, and Slack Supabase (JS client, direct Postgres, or Supabase MCP with project_id)Yes
Optional PostgREST audit loggingAppend-only audit rows from Node via PostgREST; env gating, RLS, truncation, non-blocking insertsNo
Turbopuffer APIVector search & full-text searchNo
Railway CLIDeployment & hostingYes
Slack ticket approvalApprove/Reject/Reassign dry-run Linear tickets from SlackNo
GitHub CLIRepository & PR managementYes
GitHub + Railway without UIRepo variables, secrets, and Railway services via CLI only (no dashboard)No
Snowflake CLIData warehouse accessNo
dbt Fusion + Hedra local devLocal dbt runs for Hedra (Fusion CLI, isolate failing models)No
dbt Brainforge internal data platform setupLocal dbt runs for internal data platform models (Clockify/Operating delivery marts)No
OpenWork local build setupArchived monorepo note; current Work local setup lives in brainforge-workNo
OpenWork Labs access policyInternal-only access gate + usage logging for hosted Labs (labs.brainforge.ai)No
OpenWork hosted ops runbookHosted Work operational baseline; active automation and app code now live in brainforge-workNo
OpenWork Labs Railway deployHosted Work Railway deploy notes; source repo is brainforge-workYes
files.brainforge.ai Railway migrationMove files.brainforge.ai from Heroku to Railway (DNS + custom domain)No
OpenWork hosted runtime contractStorage/env contract for hosted Work runtimeNo
Azure OpenAIBase URL, deployments, Cursor configNo
CodexCodex app/CLI/IDE, Azure East US 2, Local EnvironmentsNo
OpenAI admin usage & cleanupPlatform usage by user, projects/keys report; ChatGPT Team CSV analysis for seat cleanupNo
Snowflake CLIData warehouse accessNo
RillDelivery & financial dashboards (rill start)No
Omni CLI (Cursor + Blobby)OmniSync / Model Local Editor; Cursor + Blobby for Omni chart build and migrationNo
Rill production credentials (service_user_report)Configure/rotate Snowflake DSN in Rill Cloud projectsNo
Cursor Cloud Agent (Snowflake + Rill)Service account and env vars for Cursor Cloud AgentsNo
Granola MCPMeeting transcript retrieval and queryingYes
MotherDuck CLIDuckDB CLI install, MOTHERDUCK_TOKEN, connect to MotherDuck (prereq for all MD CLI work)No
MotherDuck — Default export loadLand Default (Calendly-style) CSV exports into my_db + upsert raw_exportNo
Hex MCPQuery Hex apps (e.g. GTM Daily Metrics Tracker) for ARR/metricsYes
Notion MCPRead/write Notion pages and databases (OAuth per user)Yes
Google Cloud CLIGCP access, project for Google AI (Vertex/Gemini), startup creditsNo
FigmaDesign context from Figma files; asset map in apps/platform/docs/figma-assets-map.mdYes
Webflow APIbrainforge.ai site + CMS for migrationNo
Linear MCPTicket/issue management (in .cursor/mcp.json)Yes
Operating MCP + APIOperating.app integration — MCP for common ops, Direct API for project/client managementYes

Legacy/Migration (Being Phased Out)

GuidePurposeStatus
n8n APIWorkflow automationMigrating to Mastra
Windmill CLIPipeline orchestrationMigrating to Railway
Dagster CLIData orchestrationMigrating to inline processing
Dagster inline migration trackerPipeline-by-pipeline migration statusActive tracker
Dagster transition deploy strategyHosting/cutover approach during migrationActive strategy

Cursor AI Agent Access

MCP Integrations (Built-in)

These services have native Cursor MCP support - agents can call them directly:

  • Linear - user-linear-* tools (in repo .cursor/mcp.json; OAuth on first use)
  • Railway - user-Railway-* tools
  • GitHub - user-github-* tools
  • Figma - design context from Figma files (in repo .cursor/mcp.json; OAuth on first use)
  • Granola - mcp_brainforge-platform-granola_* tools (meeting transcripts; OAuth required; see granola-mcp-setup.md for rate limit guidance)
  • Hex - query Hex apps for ARR/metrics; OAuth required; see hex-mcp-setup.md
  • Notion - read/write Notion; OAuth on first use, per user; see notion-mcp-setup.md
  • Operating - Operating.app integration via MCP (in repo .cursor/mcp.json). For project/client management (create, archive, rename), use the Direct API.
  • Supabase - user-supabase tools (execute_sql, list_tables) with project_id for Zoom (viqeppmsqvwpslpvttkk) and Slack (oqtkgsndvitzyfzwcdoz); see supabase-query-access.md

API Integrations (Manual)

These services require API calls via shell commands:

  • HubSpot - use the HubSpot app first for normal review/manual work, HubSpot MCP as read-only fallback in Cursor, and REST API for agent-performed writes such as notes/activities. See hubspot-api-setup.md.
  • Langfuse - REST API with Basic Auth
  • Turbopuffer - REST API with Bearer token
  • 1Password - CLI tool (op)
  • Google Cloud (gcloud) - CLI for GCP access; use when user asks about creating a GCP project, Google AI/Vertex/Gemini, or startup credits. See google-cloud-cli-setup.md.
  • Google Workspace (MCP + gws) - For Cursor/agent workflows, prefer the Brainforge Google Workspace MCP first (see setup-google-workspace-mcp.md), then fall back to the gws CLI for bulk/iterative work or when MCP is unavailable. See google-workspace-cli-setup.md.

Credential Management

All credentials are stored in 1Password. Common vaults:

VaultPurpose
Brainforge AI TeamShared team credentials
EmployeePersonal API keys

Access credentials:

# List items in vault
op item list --vault "Brainforge AI Team"
 
# Get specific item
op item get "Item Name" --vault "Vault Name"

Adding New Guides

When documenting a new CLI/API integration:

  1. Use the existing format - Follow structure in existing guides
  2. Include 1Password location - Where to find credentials
  3. Show authentication pattern - How to authenticate
  4. Provide common operations - Practical examples
  5. Note Cursor integration - MCP support or manual API calls

Questions?

For questions about these integrations, contact the AI Team.