Fixes prompt cache regression in Claude Code that causes up to 20x cost increase on resumed sessions
Enhanced prompt cache and sticky session management plugin for OpenCode, especially effective for AI API relay/gateway setups with stable SHA256-based cache keys.
45% cost reduction measured. The only Claude Code plugin built from CC source analysis — cache expiry prevention, SubTask auto-delegation, zero-cost context restoration, real-time dashboard. Max Plan + API pay-per-use.
See your AI coding costs in real time. Tokenline adds context usage, prompt-cache savings, TTL countdown and rate-limit pacing to Claude Code and Antigravity CLI.
fak — the Fused Agent Kernel: one Go binary that turns a tool-using agent (Claude Code, Codex, Cursor, any OpenAI/Anthropic/MCP client) into a managed agent: cache-stable model traffic, context compaction + crash resume, nanosecond tool-call policy, local GGUF serving with SSD expert offload.
Freeze Claude Code's prompt prefix so DeepSeek's automatic cache always hits — alignment proxy + coalescing + keepalive, installable as a CC plugin. Measured 64% cheaper on real Claude Code traffic.
Prompt caching for Pi's OpenCode Go provider (kimi, deepseek, mimo, qwen, minimax). Stamps prompt_cache_key, 24h retention, and cache_control breakpoints on every request — beyond what pi-ai or opencode CLI do by default.
Request-level image lifecycle middleware for AI coding agents: keep current-turn images, downgrade historical ones to stable placeholders.
Context-aware caching proxy for Anthropic and OpenAI — maximizes prompt-cache hit rates by anchoring long-lived context into stable boundaries. Drop-in for Claude Code, Codex, Cursor, and custom harnesses.
Local-first memory and prompt-cache layer for Claude Code
Keep your Claude Code prompt cache warm during breaks. Saves up to 92% on cache rebuild costs.
OpenCode sidebar plugin for cache hit rate, token usage, and session cost—with sub-agent (child session) aggregation and optional per-call JSONL timeline.
Autonomous loop runner + Beads-powered methodology for AI coding agents. Carve code, not chaos.
Chinese-first project baseline handoff kit for AI-assisted development.
Local-first spend & cache-efficiency dashboard for Claude Code. Reads ~/.claude/projects, tells you exactly what's costing you. No telemetry, no API key.
Zero-dependency OpenAI-compatible API bridge for OpenAI, Anthropic, GLM (Z.ai) & Kimi (Moonshot AI), with prompt injection and an Anthropic prompt-cache keepalive. Bring your own key.
Cache-aware context surgery proxy for Claude Code — cut dead weight from history without breaking the prompt cache
Local facade and install scripts for OpenClaw third-party OpenAI-compatible Responses providers, with allowlisted routing and reversible setup.
Unified AI API gateway for developers — token-based access to Claude, GPT, and domestic models (glm, kimi, minimax, qwen) with simple integration and routing via https://api.lmuai.com
Diagnose your Anthropic prompt cache hit rate and find the exact reason you're missing it.
Add a description, image, and links to the prompt-cache topic page so that developers can more easily learn about it.
To associate your repository with the prompt-cache topic, visit your repo's landing page and select "manage topics."