A few weeks ago we released the open-source Stigg Agent SDK, which governs every LLM call your agents make. Today, Stigg runs inside your agent orchestration layer, giving your orchestrator the monetization context it was missing: entitlements, credits, and budgets, at every step of a run.

🆕 What's new?

Complex agentic workflows run on an orchestrator. That layer decides which model to call, which tool to run, and when to hand a step to a sub-agent. Every one of those decisions has a cost, and until now the orchestrator had no idea what the customer is entitled to or how much budget is left.

The Stigg Agent SDK now natively supports the following orchestrators: Vercel AI SDK, LangChain & LangGraph, and Google ADK.

💡 Why it matters

• Decide during the run, not after it: Billing usually counts spend after the run, when it's too late to change what the agent did. Stigg moves the decision into the run, so you no longer have to choose between letting agents overspend (leaking revenue) and blocking the whole run up front (losing the outcome).
• Fall back to a cheaper model instead of failing: Give an agent a budget per model. When a customer's premium-model allowance runs dry, the next step answers with a cheaper model instead of stopping the run.
• Charge for premium tools: Monetize tools such as web search, code execution, or data enrichment per use. Each governed tool call is checked before it runs and metered after it succeeds.
• Run agents in parallel on a shared credit pool: Each agent gets its own allocation from the customer's pool, so one runaway agent can't drain the budget for the rest of the workflow.
• Track usage per customer, per agent, and per tool: Usage is attributed to the agent that made each call, so you can track cost and adoption without stitching logs together, and price from what runs actually cost.
• Price from real runs: Stigg meters each run end to end, so you know what a workflow actually costs. Use that data to make pricing and packaging decisions, whether you use token-based or outcome-based pricing, and keep iterating as your costs change or as your agents deliver more value.
• See Stigg in your existing traces. Every check and usage report shows up as an OpenTelemetry span, so Stigg appears in Langfuse, LangSmith, and similar tools with no extra configuration.

⚒️ Get started

1. Install the latest version of the Agent SDK: npm install @stigg/agent-sdk-js
2. Read the orchestrator guides: Vercel AI SDK, LangChain & LangGraph, Google ADK

📦 Availability

Agent orchestrator support is available in the Stigg Agent SDK v0.7.7 or later.