Case studyMaintainer

Praxis

A Go library for orchestrating LLM agents with the guardrails built into the kernel — typed invocation state machine, policy hooks at every sensitive boundary, budget enforcement and mandatory telemetry.

As of September 2026.

4-Dbudget enforcement: time, tokens, tool calls, costYou can check this: praxis.
OTelmandatory, not optionalYou can check this: praxis.
2.0Apache licence, published on pkg.go.devYou can check this: praxis.

How it works

Diagram generated from the knowledge base
The Praxis invocation state machine An invocation passes a policy pre-hook, which can deny it, then calls the model. A tool decision follows. Tool calls run, their output passes a filter chain, and the model continues; more tool calls loop back to the decision. At end of turn a post-hook runs and the invocation completes. A breach of any budget dimension ends it as BudgetExceeded instead. more tools PreHookpolicy allows or denies LLMCall BudgetExceededany of four dimensions breached ToolDecision ToolCall PostToolFilterfilter chain on tool output LLMContinuation PostHookthen Completed

Context

Most agent libraries are thin wrappers over a provider's HTTP API. That is the right shape for calling a model, and the wrong shape for a team that has to answer for what the agent did, what it cost and who authorised it.

Problem

Governance added on top of an orchestration library is governance that can be bypassed. If policy checks, budgets and telemetry live in application code, every new call site is a chance to forget one, and there is no framework-level guarantee to point at during an audit.

Role

Maintainer.

Decisions

01 · Guardrails and policy enforcementGuardrails in the kernel, not in the applicationGuardrails and policy enforcement

A typed invocation state machine with policy hooks and filter chains at every security-sensitive boundary, plus four-dimensional budget enforcement and a typed error taxonomy. The cost is overhead that a simple API call does not need — the README says so plainly and points at the thin SDKs for anyone who wants that instead.

Ask: Why put the guardrails in the framework, not the app?
02 · Agent orchestrationProvider-agnostic LLM interfaceAgent orchestration

Model access sits behind an interface, so the runtime is not bound to one vendor.

Ask: Can I swap the model provider?
03 · LLM observability and tracingTelemetry is mandatoryLLM observability and tracing

OpenTelemetry is not an optional integration but a framework guarantee, so every invocation is traceable without the application opting in. Optional per-call identity signing sits alongside it.

Ask: Why can't telemetry be turned off?

Details

Skills
GoAgent orchestrationLLM observability and tracingGuardrails and policy enforcement

Results

  • Apache 2.0, published on pkg.go.dev with CI on the repository.
  • The README states what the library is not for, which is the honest way to ship an opinionated runtime.
  • The kernel enforces budgets in four dimensions — wall-clock time, LLM tokens, tool-call count and estimated cost — and a breach ends the invocation.

    you can check this: praxis

  • OpenTelemetry spans are a framework guarantee, emitted for every invocation without the application opting in.

    you can check this: praxis

  • Released under Apache 2.0 and published on pkg.go.dev, with CI on the repository.

    you can check this: praxis