{
  "markdown": "# AI Token Optimizer\n\n> An MCP server that cuts AI coding-agent token usage by **60–80%** via framework-aware context optimization.\n\n[![CI](https://github.com/shivam-app-developers/ai-optimizer/actions/workflows/ci.yml/badge.svg)](https://github.com/shivam-app-developers/ai-optimizer/actions/workflows/ci.yml)\n[![CodeQL](https://github.com/shivam-app-developers/ai-optimizer/actions/workflows/codeql.yml/badge.svg)](https://github.com/shivam-app-developers/ai-optimizer/actions/workflows/codeql.yml)\n[![npm](https://img.shields.io/npm/v/@ai-optimizer/core.svg)](https://www.npmjs.com/package/@ai-optimizer/core)\n[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](./LICENSE)\n\n<!-- mcp-name: io.github.shivam-app-developers/ai-optimizer -->\n\nAI coding agents read whole files when they need 15 lines. They load `*.g.dart`,\n`R.java`, `node_modules/`, and `dist/` into context — then do it again next turn.\n**AI Token Optimizer** is a [Model Context Protocol](https://modelcontextprotocol.io)\nserver that intercepts those tool calls and strips framework noise _before_ it\nreaches the model.\n\nWorks with anything that speaks MCP: **Claude Code, Cursor, Cline, Continue, Zed,\nJetBrains AI, GitHub Copilot agent mode, Antigravity, Windsurf, OpenAI Codex CLI.**\n\n## Quickstart\n\n```bash\nnpm install -g @ai-optimizer/core\n```\n\nThen point your agent at the `ai-optimizer` stdio server. For Claude Code, add to\n`~/.claude/settings.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"ai-optimizer\": {\n      \"command\": \"ai-optimizer\"\n    }\n  }\n}\n```\n\nOr scaffold the config and preview your savings automatically:\n\n```bash\nnpx @ai-optimizer/init\n# \"we'd skip 47 files, ~14K tokens/session, ~$0.42/session at Sonnet pricing\"\n```\n\nFull per-agent setup (Cursor, others), LSP install, and verification steps live in\n[`packages/core/README.md`](./packages/core/README.md).\n\n## What it does\n\nIt registers tools the agent uses **instead of** its built-in file/dir/error reads:\n\n| Tool                    | What it saves                                                      |\n| ----------------------- | ------------------------------------------------------------------ |\n| `optimized_read_file`   | Framework-aware skip + optional line slice                         |\n| `optimized_list_files`  | Walks the tree applying ignore globs (+ `.gitignore`)              |\n| `optimized_grep`        | Content search scoped by ignore filters                            |\n| `optimized_diagnostics` | Spawns the right LSP, returns errors + a narrow code window        |\n| `read_symbol`           | LSP `workspace/symbol` slice instead of a full-file read (Pro/LSP) |\n| `strip_bash_noise`      | Strips ANSI, npm/maven/gradle progress, JVM warnings               |\n| `optimizer_status`      | Detected frameworks, active packs, cumulative tokens saved         |\n\nWhen a project type is detected (e.g. Python via `pyproject.toml`), the matching\npack's ignore rules activate automatically. `.gitignore` is respected on top.\n\n## Tiers\n\n|                                       | Free (MIT core) | Pro ($9/mo)                                                                      | Team ($29/seat)                  |\n| ------------------------------------- | --------------- | -------------------------------------------------------------------------------- | -------------------------------- |\n| Framework packs                       | Python, JS/TS   | + React, Flutter, Java, Kotlin, Go, Rust, C#, Swift, Ruby, Elixir, PHP, Solidity | all Pro packs                    |\n| History compaction                    | —               | ✓                                                                                | ✓                                |\n| Scheduler (`claude -p` cron)          | —               | ✓                                                                                | + work-stealing across providers |\n| Audit log + secret redaction + policy | —               | —                                                                                | ✓                                |\n\nSee [`PRICING.md`](./PRICING.md). Behavior-changing or data-sharing features\n(scheduler, work-stealing, telemetry) are **off by default** and require explicit\nopt-in. Display-only features (dollar-value savings) default on.\n\n## Monorepo layout\n\n| Package                                                                          | Purpose                                             | Published |\n| -------------------------------------------------------------------------------- | --------------------------------------------------- | --------- |\n| [`packages/core`](./packages/core)                                               | MCP server + free packs                             | npm (MIT) |\n| [`packages/init`](./packages/init)                                               | `npx @ai-optimizer/init` config + savings preview   | npm       |\n| [`packages/sdk`](./packages/sdk)                                                 | `definePack` plugin SDK                             | npm       |\n| [`packages/context-mode-import`](./packages/context-mode-import)                 | Import `.contextmode` configs                       | npm       |\n| `packages/pro`                                                                   | Pro/Team packs + scheduler + audit                  | private   |\n| `apps/landing`, `apps/dashboard`, `apps/vscode-extension`, `apps/license-server` | Site, local dashboard, IDE widget, license issuance | private   |\n| `tools/pack-bench`                                                               | Pack quality benchmark harness                      | private   |\n\nArchitecture details: [`ARCHITECTURE.md`](./ARCHITECTURE.md).\n\n## Development\n\n```bash\nnpm install            # at repo root (workspaces)\nnpm run build          # build all packages\nnpm test               # run all test suites\n```\n\n## Security\n\nLocal-only by design — license validation is the only network call. Report\nvulnerabilities per [`SECURITY.md`](./SECURITY.md).\n\n## License\n\nCore engine is MIT — see [`LICENSE`](./LICENSE). Pro/Team packs are proprietary.\n",
  "bytes": 6149,
  "sha": "94e67d8596937c35a8222212e35f7db178c2495d5084c52e55e0d81a950a10aa",
  "repo_slug": "shivam-app-developers/ai-optimizer",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_shivam_app_developers_ai_optim_b6a6b48b/readme"
}