TealfabricLibrary

Platform Limits (Current Defaults)

This page summarizes the most relevant default limits for end users.
Values reflect current platform configuration and can be adjusted by operators per environment.

File Upload Limits

  • Platform file API uploads: 1 GiB per file
  • Webapp process uploads: 1 GiB per file

Process and Step Execution Limits

  • Default step timeout: 1 hour
  • Maximum step timeout: 24 hours
  • Default step memory input: 64 MB
    • Enforced for ProcessFlow step execution.
    • JS/TS VM runner currently does not hard-enforce this as a process memory cap.
  • Synchronous process concurrency: 20 concurrent sync executions per API instance (WebApp form posts and async=false API calls share this pool)

WebApp Runtime Rate Limits

Published WebApp traffic (/webapp-runtime/...) is limited per client IP per WebApp:

  • Burst: 1000 requests per 10 minutes
  • Hourly: 10 000 requests per hour

Exceeded limits return HTTP 429 with { "success": false, "message": "Rate limit exceeded" }.

Rate-limited requests are rejected before ProcessFlow starts, so they are not recorded in the process execution journal. For batch or scripted workloads, use client-side rate limiting, slow ramp-up, and cool-down after 429, and track job state in your application so rate_limited items are not mistaken for missing executions.

See WebApps API documentation — Rate limiting for patterns and examples.

Chat and Agent Limits

  • Chat agent turn timeout: 4 hours default
  • LLM request timeout: 30 minutes default
  • MCP tool execution timeout: 1 hour default
  • Integration sync timeout: 30 minutes default

Agent Wait Limits

  • Per-call max wait: 2 hours
  • Per-turn total wait: 4 hours default
  • Rate limit: 45 wait invocations per user per minute

Conversation Summary Cadence

  • Long-memory summarization cadence: every 5 messages by default

If your organization needs different limits, ask your platform administrator to adjust the relevant platform settings.