token-budget-claude — Calibration Workflow
Integrate the offline Claude count approximation, calibrate representative samples, and report uncertainty without claiming exact counts.
$1
Works with the AI Agents you already use
token-budget-claude — Calibration Workflow
Example session with this skill installed
Use token-budget-claude-workflow to integrate this package and run its local fixture, keeping unverified provider behavior explicit.
- Read your context and instructions
- Compiled the token-budget-claude
Local synthetic fixture result from the published package: {"baseCount":2,"syntheticRatio":2}
Interpretation: Approximate text count only. No baked-in empirical Claude accuracy, true Claude token ids, live host context or billing certification. calibrate does not fetch usage. Empty/zero-base and malformed sample handling require explicit validation.
Next validation: Count an empty string and representative prose/code/multilingual inputs. Use synthetic calibration only as an algorithm test; never label its targets real provider usage. Check encode is absent and invalid calibration inputs do not enter your application pipeline.
Evidence scope: source version 0.1.5 suite passed 15 tests; published 0.1.4 was separately exercised with the included fixture. These are distinct baselines. Live provider/SDK/backend behavior was not tested, and no network service credential was needed for the example.
Connects securely to your tools. The creator never sees your data.
About this skill
Integrate the offline Claude count approximation, calibrate representative samples, and report uncertainty without claiming exact counts.
Original application-integration playbook covering setup, exact supported exports, migration, failure handling and local verification. This is a curated workflow package, not a new MCP server or automatic host-context control.
Verified source 0.1.5: 15 tests passed. The included local fixture also passed against published npm 0.1.4 tarball code using peer/dependency resolution from the reviewed monorepo, not an independent clean-consumer install. No live provider, embedding service or telemetry backend was contacted.
Limitations: Approximate text count only. No baked-in empirical Claude accuracy, true Claude token ids, live host context or billing certification. calibrate does not fetch usage. Empty/zero-base and malformed sample handling require explicit validation.
The separate npm runtime remains free/open-source under MIT at https://github.com/shivam039/token-budget/tree/main/packages/token-budget-claude. MIT notice is preserved; this $1 purchase is for original curated workflow instructions, not exclusive runtime access. Provider/exporter services, if selected, are separately authorized and may charge. No runtime dependencies are bundled and no blanket dependency-security clearance is promised.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.
Approximate text count only. No baked-in empirical Claude accuracy, true Claude token ids, live host context or billing certification. calibrate does not fetch usage. Empty/zero-base and malformed sample handling require explicit validation.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 19 days ago
- Accounts
GitHub@shivam.dixit/token-budget@shivam.dixit/token-budget-claude@0.1.4
- Software
Node.js 18 or newerTerminal/project-edit capable skill host
Git
Permissions
Allowed Hosts
Project-scoped setup, edits and synthetic local fixtures. No provider/exporter endpoint is configured by this skill. Any live data transmission or optional service use requires the authorized destination and scope; never include credentials in evidence.