token-budget

Code intelligence for agents: find the code that matters and keep your context window and tokens lean.

Run-aware token governance for multi-agent systems.

A budget-aware context compiler for coding agents - scan, grade, spark-test for secrets, and pour a hard-budget context pack with a stamped manifest.

让 Agent 高效又守纪律 — 不止省 token：ZeroToken 压缩无效上下文/推理/输出；尉缭子十原则约束权限边界、单一指令、先谋后动、验证先于结束；附 Unicode 编码规范、搜索规范、六种任务模式。More than token savings: ZeroToken efficiency + AI coding discipline for Reasonix / Codex / OpenCode / Hermes

Gobstopper makes long Claude Code and Codex sessions smaller. A local proxy compacts live requests, and saved sessions get a smaller copy beside the original.

GenPark AI Agent Skill - Multi-tenant token budget tracker, sliding-window rate limiter, and model pricing ledger.

GenPark AI Agent Skill - Multi-tenant token budget tracker, sliding-window rate limiter, and model pricing ledger.

Open-source platform for deterministic, token-aware context selection for AI agents and LLMs

Self-hosted model routing gateway and agent control plane for OpenAI, Anthropic, Gemini and A2A agents behind one OpenAI-compatible API, protocol-first. MCP server security, governed tools, prompt enhancement, virtual-key budgets, cache, observability, audit chain and agent governance. Zero-key local first run; public preview.

Multi-modal vision token cost and tile decomposition calculator for OpenAI, Anthropic, and Google architectures

Self-hosted spend firewall and gateway for LLM ( OpenAI / Anthropic / Gemini ). Hard per-user & per-project budget caps that block runaway costs before the API call, plus cost-per-customer tracking, semantic caching, and failover. One line of code, single Go binary.

Multi-modal vision token cost and tile decomposition calculator for OpenAI, Anthropic, and Google architectures

Local-first MCP server that tells coding agents which files to read, and how sure it is. 97% Hit@10 on Java (SWE-bench Multilingual), 89% on Python (SWE-bench Lite), zero LLM cost. mcp-brain calibrate re-measures it on your own Git history. Plus team memory and conflict checks.

Runtime containment kernel for LLM agents. Enforces budget, step, retry, and circuit-breaker limits before the model call.

Embeddable, zero-dependency durable execution for agents and NHEs. Deterministic replay, retries, cycle detection, a token budget, and a multi-agent task board, with no sidecar service.

Don't go into production without these - 3 auto-triggering Claude Code skills for cost and drift prevention. 6 months of practitioner notes.

Read, filter, redact and export the on-disk sessions of ten AI coding agents (Claude Code, Codex CLI, Gemini CLI, Antigravity, Copilot Chat, Cursor, Cline, Roo Code, Continue, Aider) and three chat apps (LM Studio, Open WebUI, SillyTavern), or copy one back into its agent and resume it. Command line and a local web UI.

A drop-in SKILL that forces AI coding agents (Claude Code, Codex, Cursor, Cline, Roo, Windsurf, Copilot, Augment, Aider, …) to deliver exactly what was asked — minimum diff, zero unsolicited files, terse output by default.

Enforce real-time token budgets and spending limits for OpenAI, Anthropic Claude, and Google Gemini API calls in Node.js

TokenSched 给 Claude Code 的 token 预算装上了一个 CPU 调度器：它按子任务期望值预分配预算、预测超支，并在 5 小时窗口耗尽前自动把低价值工作降级到 Haiku 或抢占——把硬截断变成可调度的软退让。