ctx autopsy

    1

    Audit local Claude Code logs to identify oversized tool outputs and redundant file re-reads.

    $15

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    ctx autopsy

    Example session with this skill installed

    Run an autopsy on my Claude Code logs for the last 7 days to see why my context window is filling up so fast.

    • Read your context and instructions
    • Compiled the ctx autopsy

    I have analyzed your local Claude Code session logs from the last 7 days. The autopsy identified 4 flagged events contributing to 72% of your tool-output context, primarily driven by large bash outputs and redundant file reads.

    Context Autopsy Report (Last 7 Days)

    CategoryItemSize (Chars)Flag
    Tool Resultbash (npm test)84,200OVERSIZED
    Re-readsrc/types/schema.ts42,150REDUNDANT
    Subagent Re-readdocs/api-spec.md31,800REDUNDANT
    Tool Resultmcp:github (get_contents)22,400OVERSIZED
    First RequestSystem Prompt + Tools12,100NORMAL

    Summary Statistics:

    • Total tool-output text: 192,650 characters
    • Redundant re-reads: 73,950 characters (38.4%)
    • Largest single output: npm test (84.2k chars)
    • Claude Code version detected: 2.1.310 (Split available)

    First-Request Breakdown:

    • System Prompt: 4,200 chars
    • Tool Definitions (Built-in): 3,500 chars
    • Tool Definitions (MCP): 4,400 chars

    Next steps

    • Review your npm test script and consider piping output to a file or using a reporter that produces less stdout.
    • Consolidate subagent tasks to prevent them from re-reading docs/api-spec.md which is already in the main session context.
    • Check if src/types/schema.ts is being modified; if not, avoid re-reading the full range in subsequent turns.

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

    About this skill

    Find out what ate your Claude Code context window — ranks the biggest tool results, file re-reads, and subagent re-reads across your local sessions, offline. Run it when context fills up fast or sessions compact early.

    Requires Node 22+. Token counts are rough estimates (chars / 3). Does not measure content of images, PDFs, or tool search results. Redacts potentially sensitive file names.

    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 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 10 days ago

    Requires

    ctx-autopsy
    Node.js

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access
    Listed10 days ago

    What's inside

    Frequently Asked Questions

    This skill identifies 'context bloat' by auditing your local logs to find oversized bash outputs, redundant file re-reads, and excessive subagent overhead that standard Claude Code commands like /usage cannot see.

    The skill requires Node.js 22 or higher and is specifically designed for Claude Code users, reading logs directly from the local project directories.

    No, this is a local-first CLI tool that reads your existing logs on your machine; it does not require an active API connection or send your private log data to external servers.

    You receive the full Node-based CLI tool, including the logic for log deconstruction, redundant read detection, and the ability to export forensic reports in JSON format.

    Yes, it is compatible with Claude Code 2.1.310 and above, including specific features to deconstruct the new system prompt and tool definition structures in the latest versions.

    New in LLM Ops