{"data":{"slug":"juliusbrussee-caveman","name":"caveman","tagline":"Reduce token usage with concise 'caveman'-style prompts.","github_url":"https://github.com/JuliusBrussee/caveman","owner":"JuliusBrussee","repo":"caveman","owner_avatar_url":"https://avatars.githubusercontent.com/u/104168679?v=4","primary_language":"Go","stars":98423,"forks":5690,"topics":["ai","anthropic","caveman","claude","claude-code","llm","meme","prompt-engineering","skill","tokens"],"archived":false,"github_pushed_at":"2026-08-15T14:07:37+00:00","maintenance_label":"Very active","stars_delta_30d":8329,"url":"https://www.graphcanon.com/tools/juliusbrussee-caveman","markdown_url":"https://www.graphcanon.com/tools/juliusbrussee-caveman.md","api_url":"https://www.graphcanon.com/api/graphcanon/tools/juliusbrussee-caveman","graph_url":"https://www.graphcanon.com/api/graphcanon/graph?tool=juliusbrussee-caveman","description":"🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman","homepage_url":"https://caveman.so/","license":"MIT","open_issues":485,"watchers":237,"ai_summary":"A JavaScript-based repository offering a Claude Code skill that minimizes the number of tokens used in AI interactions by employing simplified language structures, reducing token count by up to 65%.","readme_excerpt":"## Install\n\nCaveman is three separate installs. Each works alone. Not sure? Start with the skill — it needs no account, proxy, Go toolchain, or code changes.\n\n**Shorter answers — the skill (MIT).** The caveman skill + slash commands in every supported agent found on your machine. Nothing else.\n\n```bash\nnpx skills add JuliusBrussee/caveman\n```\n\n**Smaller inputs — Caveman Proxy (BSL-1.1 runtime; MIT CLI).** The `caveman` CLI plus signed local binaries: proxy, engine, MCP recovery, memory — and the optional browse + shrink tools. Then `caveman claude` wraps your agent.\n\n```bash\nnpm install -g @caveman-ai/cli && caveman setup --install\n```\n\n**Compressed browsing — Caveman Browse (BSL-1.1).** Included in `caveman setup --install` (needs Chrome) — a local Chrome driver your agent reaches as MCP tools.\n\n```bash\ncaveman browse <url>\n```\n\n**A new agent — Agent SDK.** A TypeScript agent project on the native Caveman runtime. Client SDKs: `npm i @caveman-ai/sdk` · `pip install caveman-sdk`.\n\n```bash\nnpm create @caveman-ai/agent@latest my-agent\n```\n\nOn Windows, the skill installs with `irm https://raw.githubusercontent.com/JuliusBrussee/caveman/v1.10.0/install.ps1 | iex` (PowerShell 5.1+).\n\nThe skill installer needs Node.js 18+, finds supported agents already on your machine, skips the rest, and is safe to rerun. Prefer one agent only?\n\n```bash\n\n---\n\n### Install the CLI\n\n```bash\nnpm install -g @caveman-ai/cli\ncaveman setup --install   # downloads the signed runtime binaries\n```\n\n`setup --install` verifies the signed checksum manifest and the SHA-256 of\nevery binary before an atomic install. Prefer building from source? A clone\nplus `scripts/install-local-cli.sh` (macOS/Linux) or\n`pwsh -File scripts/install-local-cli.ps1` (Windows) still works — that path\nneeds Go and `pnpm`. SDK users can point provider base URLs at the local\nProxy directly (`ANTHROPIC_BASE_URL=http://127.0.0.1:8787/anthropic`).\n\n```bash\ncaveman claude                  # full stack (default): S4 compress + TOON best-of + caveman & browse MCP tools + output shrink\ncaveman wrap --off codex        # byte-safe pass-through metering only\ncaveman wrap --pixel claude     # lossy text → PNG pixel mode (model-gated)\n```\n\nSubscription logins work — see the [note](#wrap-any-agent) below.\n\n| Mode | What it does | Bytes the model sees |\n|---|---|---|\n| default stack *(`caveman claude`)* | Structural compression routed per content type (table below), plus uniform JSON tool results re-encoded as TOON only when measured smaller; config `toon: false` turns it off. | Changed, recoverable |\n| `--off` | Counts tokens and cost. Changes nothing. | **Byte-identical** |\n| `--pixel` | Dense text slabs rendered to PNG pages for vision models. | Changed, recoverable |\n\nSafety gates stay explicit:\n\n- **CCR first.** Before a lossy transform goes upstream, original bytes land in **CCR**, a content-addressed store on your disk. The agent retrieves them through `caveman_retrieve` or `caveman retrieve <handle>`. Parse problem, store failure, or larger result sends original bytes unchanged.\n- **Visible declines.** Pixel refuses sparse code. Convert refuses skills when PNG pages do not beat text. TOON runs only when its output measures smaller. Each decline includes its reason.\n- **Labeled evidence.** Local results report `inferred`: estimates for choosing what to try. `verified` requires real traffic and eval gates. Offline caveman never reports it.\n\n---\n\n## License\n\nSplit license. Skill and adoption surfaces are [MIT](./LICENSE). Engine-linked runtime is BSL-1.1 source-available, not OSI Open Source before Change Date.\n\n**MIT** — the skill, Agent SDK and initializer, the CLI, both client SDKs (TS + Python), kit, evals/graders, contracts, provider catalog, the extension shell, and the thin cavemem clients.\n\n**BSL-1.1** — Engine, Proxy, Cache Engine, rewriter, Browse, MCP server, `shrink`, cavemem Go core, and shared Go platform. New Engine-linked runtime modules default to BSL-1.1. Source-available: read","github_created_at":"2026-04-04T10:03:00+00:00","created_at":"2026-07-07T17:30:34.309883+00:00","updated_at":"2026-08-16T06:01:36.46234+00:00","categories":[{"slug":"developer-tools","name":"Developer Tools","url":"https://www.graphcanon.com/categories/developer-tools","markdown_url":"https://www.graphcanon.com/categories/developer-tools.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/developer-tools"},{"slug":"llm-frameworks","name":"LLM Frameworks","url":"https://www.graphcanon.com/categories/llm-frameworks","markdown_url":"https://www.graphcanon.com/categories/llm-frameworks.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/llm-frameworks"}],"tags":[{"slug":"ai","name":"ai"},{"slug":"anthropic","name":"anthropic"},{"slug":"caveman","name":"caveman"},{"slug":"claude-code","name":"claude-code"},{"slug":"prompt-engineering","name":"prompt-engineering"},{"slug":"tokens","name":"tokens"}],"trust":{"provenance":{"is_fork":false,"github_id":1201173969,"owner_type":"User","methodology":"github_public_v1","parent_repo":null,"near_duplicate_slugs":[]},"computed_at":"2026-08-16T06:01:35.759Z","maintenance":{"label":"Very active","score":96,"methodology":"github_public_v1","releases_90d":5,"days_since_push":0,"last_release_at":"2026-08-11T15:55:03Z","stars_delta_30d":8329,"open_issues_delta_30d":84},"security_summary":{"status":"no_lockfile","scanner":null,"low_count":0,"high_count":0,"last_scan_at":"2026-07-11T10:56:03.697Z","medium_count":0,"scan_profile":"none","critical_count":0}},"capability_facts":{"mcp":{"source":"repo_scan","observed_at":"2026-08-16T06:01:36.197Z","server_manifest":false},"scan":{"source":"repo_scan","observed_at":"2026-08-16T06:01:36.197Z"},"has_cli":{"value":true,"source":"package.json:bin|scripts","observed_at":"2026-08-16T06:01:36.197Z"},"languages":{"value":["go","javascript"],"source":"github.language+package.json","observed_at":"2026-08-16T06:01:36.197Z"},"license_spdx":{"value":"Other","source":"github.license","observed_at":"2026-08-16T06:01:36.197Z"}},"decision_facts":{"hosting":null,"pricing":null,"requirements":null,"constraints":null,"when_to_use":["When you need to significantly cut down on token usage in AI interactions, up to 65%, without losing essential information content.","If your project can benefit from a simplified or direct language style that still conveys necessary instructions effectively."],"when_not_to_use":["When requiring complex and detailed prompts that necessitate more nuanced expression beyond simple, 'caveman'-style sentences.","For situations where adherence to formal or specific linguistic structures is mandatory for the task's success."],"source":"enrich:decision_facts","observed_at":"2026-07-11T11:52:43.508Z"},"constraint_facets":null,"decision_summary":[{"label":"Adopt for","value":"The **caveman** tool is designed for developers and AI users who aim to optimize their token usage through the generation of more concise prompts, thereby potentially reducing costs and improving efficiency. However, it犺"}]}}