Add Cycles to LLMOps tools — runtime budget and action enforcement

[amavashev]

Add Cycles

Name: Cycles
Website: https://runcycles.io
GitHub: https://github.com/runcycles
License: Apache 2.0
Languages: Python, TypeScript, Java/Spring, Rust

Category fit

Cycles fits squarely in LLMOps / Cost & Risk Management — it's the
runtime authority layer that enforces budget, risk, and action limits on
LLM-driven applications before they execute. It complements observability
tools (Helicone, Langfuse) and proxies (LiteLLM, OpenRouter) by sitting
in the pre-execution path rather than the after-the-fact telemetry path.

Suggested entry

For the "Cost" or "LLM Application Frameworks" section:

Cycles — Open protocol for runtime authority over autonomous agents.
Pre-execution enforcement of budget, risk, and action limits via reserve-commit lifecycle.
Multi-language SDKs (Python, TypeScript, Java/Spring), OpenAPI 3.1 spec,
integrates with LangChain, LangGraph, OpenAI Agents, CrewAI, AutoGen, MCP, and 20+ frameworks.
Self-hosted, no prompt storage. Apache 2.0.

Why include it

• Solves a category gap: most LLMOps tools observe what happened. Cycles
  prevents what shouldn't happen.
• 5,300+ package installs across PyPI, npm, Maven Central
• Active development with weekly releases
• Ships llms.txt and llms-full.txt for AI agent integration
• Used in production by design partners

Differentiation from existing entries

Tool	When you reach for it
Helicone / Langfuse	After execution — observability, traces, costs
LiteLLM / OpenRouter	Routing and provider abstraction
Guardrails AI	Output validation post-completion
Cycles	Before execution — block the call if budget/risk exhausted

Happy to open a PR directly with the formatting you prefer.