Context Bloat Finder

    1

    Finds context items irrelevant to your current task and tells you exactly how to remove them.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Context Bloat Finder

    Example session with this skill installed

    Context is almost full and replies are getting cut off. /find-bloat — what can I drop?

    • Read your context and instructions
    • Compiled the context bloat finder
    🧹 Found ~41,200 recoverable tokens across 4 items
    
    🔴 Remove first — high impact
    - node_modules/express/index.js (~9,800 tokens) — loaded while debugging a README, not needed. Close the file tab.
    - Turns 1–7 (~18,000 tokens) — initial onboarding discussion, unrelated to current refactor. Run: /clear 7
    
    🟡 Consider removing — medium impact
    - mcp: slack (~4,800 tokens) — no Slack actions in the current task, but you may need it later.
    - package-lock.json (~8,600 tokens) — only the dependency name matters, not the full lockfile.
    
    🟢 Keep — necessary context
    - src/api/auth.ts, CLAUDE.md, turns 8–now
    
    💡 Optimization moves
    - Run /clear 7 for the biggest single gain
    - Consider disconnecting the slack MCP server for this session: /mcp disconnect slack
    

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

    About this skill

    Identifies the current task, scores every context item for relevance (0–10), and surfaces everything scoring 3 or below as a bloat candidate — with the exact command to remove each one.

    When to use

    • Responses are getting truncated or feel degraded
    • You've loaded many files and aren't sure what's still needed
    • You want a prioritized list of what to clean up

    How it works

    Invoke it by name in your agent (in Claude Code: /find-bloat). It reads only what is already in the current session's context — no files are written, no network calls, no external services.

    Part of Claude Token Inspector, an open-source (MIT) suite of 8 context-window skills by the original author. Source: https://github.com/msishariful/claude-token-inspector

    Token estimates are approximations. The skill identifies bloat but requires the user to execute the suggested removal commands.

    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

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 23 days ago

    • Free with an Agensi account

    Permissions

    No special permissions declared or detected

    Listed23 days ago

    What's inside

    Frequently Asked Questions

    This skill analyzes your active session to identify redundant files, stale history, and unnecessary instructions that consume your token budget, preventing model truncation and performance degradation.

    It's a standard SKILL.md skill, built and tested on Claude Code. It also works in other SKILL.md-compatible agents (Cursor, Codex CLI, Gemini CLI, etc.), though suggested cleanup commands like /clear and /mcp are Claude Code conventions.

    No setup beyond installing the skill. It's user-invoked: run it by name when you want it (in Claude Code: /find-bloat). It won't trigger on its own.

    The SKILL.md instructions file. It's free and open source (MIT); the same skill is part of the Claude Token Inspector repo on GitHub.

    It gives an estimate, not an exact count (about 1 token per 4 characters of prose, slightly fewer for dense code). That's enough to see which items are worth removing first.

    It never deletes anything. It only reports what's worth removing and suggests commands; you decide what to run.

    New in LLM Ops