Build a local Claude Code atlas

    1

    One offline HTML page of your Claude Code history - no server, no scripts, no network.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Build a local Claude Code atlas

    Example session with this skill installed

    Give me one offline report of my Claude Code history that I can open without running a server.

    • Read your context and instructions
    • Compiled the build a local

    Result

    Wrote atlas.html — one self-contained file. No server, no scripts, no fonts, no network.

    MetricValue
    Sessions128
    Days covered41
    Model substitutions3
    Compaction events17
    File size1.2 MB

    What is inside

    • Per-session token and cache accounting
    • Model requested next to model that answered
    • Every compaction with the tokens it discarded
    • Upstream errors with their retry counts

    Open the file directly in a browser. The data is embedded; nothing is fetched at view time.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    The three CLI reports answer three questions. Most people want one page.

    This writes a single self-contained HTML file - inline CSS, no JavaScript, no remote fonts or images - that joins session history, model attribution, substitutions, compaction losses, upstream errors and install checks into one view, with a day-by-day strip showing when the pattern changed.

    It is built to be honest rather than flattering:

    • A figure resting on an unknown-model price is labelled a placeholder.
    • Substitution is reported unmeasured when no record carries the requested model.
    • An unrecognised log format stops the page with a banner instead of printing zeroes.
    • The footer states the date the undocumented format was last verified, and the command to re-verify it.

    Open the file from disk in five years and it still renders. That is the point of no scripts, no fonts and no network.

    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.

    Requires Node.js 20+. Cannot price unlisted models. Read-only access to config directory. No remote assets or network access supported in output.

    How to use

    1. 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. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 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

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 2 days ago

    • Free with an Agensi account

    Software
    Node.js

    Permissions

    Terminal / Shell
    Read Files
    Listed2 days ago

    What's inside

    Frequently Asked Questions

    This skill solves the issue of inaccurate cost reporting caused by session resumptions, which often lead to double-counting. It also provides a clear, visual dashboard for complex terminal-based metrics like context window compaction and model substitutions.

    Build Atlas is specifically designed for environments running Claude Code and requires Node.js version 20 or higher. It operates as a local utility and does not require an active internet connection to generate the HTML report once installed.

    No, the skill operates with read-only access to your Claude Code configuration directory. It parses and consolidates your session data into a separate HTML file without modifying or deleting your original local log files.

    The skill uses hard-coded logic to de-duplicate unique session IDs and calculate 'delta-based' token loss. This ensures that even if you resume the same session multiple times, the final report accurately reflects actual usage rather than inflated cumulative totals.

    The generated report is a single, self-contained HTML file with CSS inlined. This makes it completely portable, allowing you to open it in any browser or share it with stakeholders as a standalone document without needing external dependencies.

    New in LLM Ops

    Free