More screenshots
Works with the AI Agents you already use
AI Fluency Cost
Calculate API-equivalent costs and cache hit ratios from your local Claude Code transcripts.
$5
AI Fluency Cost
Example session with this skill installed
Run a cost analysis on my Claude Code history and save the report to my desktop.
- Read your context and instructions
- Compiled the ai fluency cost
- Generated the document
I have analyzed your Claude Code transcripts. The extraction process identified 42 sessions across 3 projects. I have calculated the API-equivalent costs and generated your report.
Claude Code Usage Cost Report
Disclaimer: $1,000 is API-equivalent: computed locally from token counts at list prices, so sessions and models are comparable. Claude Code on a subscription is not billed this way. It is not money spent.
| Metric | Value | Notes |
|---|---|---|
| Total API-Equivalent Cost | $104.20 | Based on 2026-10-29 list prices |
| Cost Per Session | $2.48 | Average across all sessions |
| Cost Per Landed Session | $5.21 | Sessions resulting in a git commit |
| Cache Hit Ratio | 91.8% | Automatically handled by Claude Code |
| Opus-5 Share | 88% | $91.70 |
| Sonnet-5 Share | 10% | $10.42 |
| Haiku-4.5 Share | 2% | $2.08 |
| Weak Description Floor | $12.45 | Excess reconnaissance cost |
Report Location: /Users/username/Desktop/my-ai-cost.html
Next steps
- Open the HTML report on your desktop for the full breakdown of tool-call costs.
- Use
extract --days 7to generate a report focusing only on your last week of activity. - Review sessions with high "Weak Description" costs to refine your initial task briefings.
my-ai-cost.html
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Claude Code users often operate in the dark regarding the economic value of their local sessions. This skill analyzes your local transcript history to calculate what your usage would cost at Anthropic API list prices, providing visibility into model routing efficiency and cache performance.
What it does
- API-equivalent pricing calculates local session costs using Anthropic list prices as of October 2026.
- Model routing analysis breaks down spend and share across Opus, Sonnet, and Haiku tiers.
- Outcome metrics computes cost per landed session, per human instruction, and per tool call.
- Cache monitoring reports your cache hit ratio to validate Claude Code's internal caching efficiency.
- Waste identification isolates the cost of "weak descriptions" and sessions that never resulted in a commit.
How it works
- Check environment run the
checkcommand to verify Python 3.9+ and transcript accessibility. - Extract data execute the
extractcommand to parse local.jsonlfiles from your Claude projects folder into a secure evidence pack. - Generate report run the
costcommand to output a summary to the terminal and a detailed HTML report to your local reports folder. - Export data use the
--jsonflag to pipe cost figures directly into other tools or spreadsheets.
Frameworks & tools
Built for Claude Code. Requires Python 3.9 or newer (standard library only). Reads transcripts from ~/.claude/projects.
Why this beats prompting it yourself
Parsing thousands of lines of JSONL transcripts manually to calculate token-weighted costs is error-prone and time-consuming. This skill automates the math using a fixed pricing table and local extraction, ensuring your usage data never leaves your machine.
Use cases
- Determine if your Claude Code subscription is delivering better value than an API-based workflow.
- Identify sessions with high "weak description" costs to improve your prompt engineering habits.
- Audit which models (Opus vs Sonnet) are consuming the majority of your token budget.
- Track cache hit ratios across different project sizes to optimize performance.
Known limitations
Calculations are based on fixed list prices from 2026-10-29 and do not reflect actual subscription billing or taxes. It only supports Claude Code transcripts and requires local Python execution.
Prices are fixed at Oct 2026 list rates. Only supports Claude Code transcripts. Requires Python 3.9+. Does not reflect real subscription invoices.
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 any payment.
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 18 days ago
Requires