Open-source edge engine to control API request budgets and enforce fair usage.

Budget-aware context compilation and context firewall for tool-heavy AI agents.

OpenClaw Financial Guard is a financial guardrail layer for agent systems, preventing runaway token spend with circuit breakers, budget limits, infinite-loop protection, and an auditable usage ledger

Deterministic context budgeting for LLM prompts, demonstrating stable prompt packing within fixed token limits.

Experiments in analyzing the correlation and effects of token budgets on a language model's ability to reason and generate accurate responses.

Automated pipeline to transcribe lecture audio with Whisper, generate structured Markdown notes using LLMs, and create multiple-choice quizzes. Model-agnostic, with automatic token budgeting for long recordings.