TOKENMETER

Token and cost counter / budgeter for LLM apps, CI-ready

Developer Tools — fast, single-purpose, CI- and agent-friendly.

pip install cognis-tokenmeter
tokenmeter count -f prompt.txt -m claude-sonnet -o 500   # tokens + cost in ms

Usage — step by step

1. Install (Python 3.9+):

   pip install tokenmeter

2. Count tokens and estimate cost for some text or a file, against a pricing model:

   tokenmeter count -f prompt.txt -m claude-sonnet --output-tokens 500
   tokenmeter count -t "hello world" -m claude-sonnet

3. List known models and their pricing:

   tokenmeter models --format json | jq .

4. Batch-estimate many files and roll them up:

   tokenmeter batch prompts/*.txt -m claude-sonnet

5. Gate a budget in CI. budget exits 1 when the cost/token cap is exceeded:

   tokenmeter budget -f prompt.txt -m claude-sonnet --max-cost 0.05 || echo "Over budget"

6. Compare every model for one workload, cheapest first, and export it:

   tokenmeter compare -f prompt.txt -o 500                 # ranked table
   tokenmeter compare -f prompt.txt -o 500 --format csv    # for a spreadsheet
   tokenmeter models  --format csv                         # the whole price book

   count, models, batch, and compare all support --format csv for FinOps spreadsheets and dashboards, alongside table and json.

Worked demos

Each folder under tokenmeter/demos/ is a real-use-case scenario with an input file in the tool's real input format, a SCENARIO.md explaining where the data came from and how to act, and exact run commands:

Demo	Scenario
01-basic	Budget an LLM system prompt in CI
02-rag-context	Size an assembled RAG prompt before sending
03-model-selection	Pick the cheapest model with compare
04-batch-prompt-library	Roll up a whole prompt library with batch
05-chat-transcript	Cost of continuing a multi-turn chat
06-context-window-guard	Catch a context-window overflow before the API does
07-fewshot-vs-zeroshot	Quantify the recurring cost of few-shot examples
08-csv-finops	Export per-step agent cost as CSV for finance
09-stdin-pipeline	Measure anything piped on stdin (e.g. a git diff)

Contents

• Why tokenmeter? · Features · Quick start · Example · Architecture · AI stack · How it compares · Integrations · Install anywhere · Related · Contributing

Why tokenmeter?

AI cost control

tokenmeter is single-purpose, scriptable, and self-hostable: point it at a target, get prioritized results in the format your workflow already speaks (table · JSON · SARIF), gate CI on it, and let agents drive it over MCP.

↑ back to top

Features

• ✅ Add Model
• ✅ Get Pricing
• ✅ List Models
• ✅ Count Tokens
• ✅ Estimate
• ✅ Check Budget
• ✅ Aggregate
• ✅ Compare models (rank one workload by cost, cheapest first)
• ✅ Output as table · JSON · CSV (CSV for FinOps spreadsheets)
• ✅ Runs on Linux/macOS/Windows · Docker · devcontainer
• ✅ Ports in Python, JavaScript, Go, and Rust (ports/)

↑ back to top

Quick start

pip install cognis-tokenmeter
tokenmeter --version
tokenmeter count -t "hello world" -m claude-sonnet      # count + cost
tokenmeter count -f prompt.txt -m gpt-4o --format json  # machine-readable
tokenmeter budget -f prompt.txt -m claude-opus --max-cost 0.01   # CI gate (exit 1)
tokenmeter compare -f prompt.txt -o 500                 # rank models by cost

↑ back to top

Example

$ tokenmeter compare -f prompt.txt -o 500
model          in_tok  out_tok  total_cost_usd  ctx_used_%
gpt-4o-mini    476     500      0.000371        0.76
claude-haiku   476     500      0.002381        0.49
gpt-4o         476     500      0.006190        0.76
claude-sonnet  476     500      0.008928        0.49
claude-opus    476     500      0.044640        0.49

↑ back to top

Architecture

flowchart LR
  IN[input] --> P[tokenmeter<br/>analyze + score]
  P --> OUT[report]

Loading

↑ back to top

Use it from any AI stack

tokenmeter is interoperable with every popular way of using AI:

• MCP server — tokenmeter mcp (Claude Desktop, Cursor, Cognis.Studio, uncensored-fleet)
• OpenAI-compatible / JSON — pipe tokenmeter scan . --format json into any agent or LLM
• LangChain · CrewAI · AutoGen · LlamaIndex — wrap the CLI/JSON as a tool in one line
• CI / scripts — exit codes + SARIF for non-AI pipelines

↑ back to top

How it compares

	Cognis tokenmeter	tiktoken
Self-hostable, no account	✅	varies
Single command, zero config	✅	⚠️
JSON + SARIF for CI	✅	varies
MCP-native (AI agents)	✅	❌
Polyglot ports (JS/Go/Rust)	✅	❌
Open license	✅ COCL	varies

Built in the spirit of tiktoken, re-framed the Cognis way. Missing a credit? Open a PR.

↑ back to top

Integrations

Pipes into your stack: SARIF for code-scanning, JSON for anything, an MCP server (tokenmeter mcp) for AI agents, and a webhook forwarder for SIEM/Slack/Jira. See docs/INTEGRATIONS.md.

↑ back to top

Install — every way, every platform

pip install "git+https://github.com/cognis-digital/tokenmeter.git"    # pip (works today)
pipx install "git+https://github.com/cognis-digital/tokenmeter.git"   # isolated CLI
uv tool install "git+https://github.com/cognis-digital/tokenmeter.git" # uv
pip install cognis-tokenmeter                                          # PyPI (when published)
docker run --rm ghcr.io/cognis-digital/tokenmeter:latest --help        # Docker
brew install cognis-digital/tap/tokenmeter                             # Homebrew tap
curl -fsSL https://raw.githubusercontent.com/cognis-digital/tokenmeter/main/install.sh | sh

Linux	macOS	Windows	Docker	Cloud
scripts/setup-linux.sh	scripts/setup-macos.sh	scripts/setup-windows.ps1	docker run ghcr.io/cognis-digital/tokenmeter	DEPLOY.md (AWS/Azure/GCP/k8s)

↑ back to top

Related Cognis tools

• mcpforge — Scaffold, test, and publish MCP servers in minutes
• promptlint — Lint, version, and test prompts as code with a CI gate
• envdoctor — .env validator, secret-presence and config-drift checker
• apidiff — Breaking-change detector for OpenAPI / GraphQL across commits
• codeglance — Repo onboarding map — architecture + hotspots for humans and agents
• flakefinder — Flaky-test detector from CI history with quarantine suggestions

Explore the suite → 🗂️ all 170+ tools · ⭐ awesome-cognis · 🔗 cognis-sources · 🤖 uncensored-fleet · 🧠 engram

↑ back to top

Contributing

PRs, new rules, and demo scenarios are welcome under the collaboration-pull model — see CONTRIBUTING.md and SECURITY.md.

⭐ If tokenmeter saved you time, star it — it genuinely helps others find it.

Interoperability

{} composes with the 300+ tool Cognis suite — JSON in/out and a shared OpenAI-compatible /v1 backbone. See INTEROP.md for the suite map, composition patterns, and reference stacks.

License

Source-available under the Cognis Open Collaboration License (COCL) v1.0 — free for personal, internal-evaluation, research, and educational use; commercial / production use requires a license (licensing@cognis.digital). See LICENSE.

---

_{Cognis Digital · one of 170+ tools in the Cognis Neural Suite · Making Tomorrow Better Today}