Your coding agent is reading junk files.
That’s why it keeps looping.
When Cursor rewrites the same component for the fourth time, the model usually isn’t the problem. It’s re-reading 900KB of lockfiles and dist folders on every prompt, and your task is getting crowded out. ContextPruner writes the config files that tell it what to skip.
No signup needed. Your file list stays in your browser — we never see it.
Paste manifest
Paste your repo’s file manifest to watch the triage run — and see what the junk is costing you.
Works with your whole toolchain
One paste generates the exact instruction file every major coding agent reads — including the cross-agent AGENTS.md standard that 19 of them support natively.
Claude Code
CLAUDE.mdDoes not read AGENTS.md; bridge with an `@AGENTS.md` import inside CLAUDE.md or a symlink.
Docs ↗Gemini CLI
GEMINI.mdReads AGENTS.md only via `{"context": {"fileName": "AGENTS.md"}}` in .gemini/settings.json.
Docs ↗Aider
CONVENTIONS.mdAdd `read: AGENTS.md` to .aider.conf.yml to auto-load rules.
Docs ↗Lovable
AGENTS.mdAlso reads CLAUDE.md and in-app Knowledge settings.
Docs ↗Replit Agent
replit.mdReplit Agent maintains replit.md itself — paste rules into it or use in-app instructions.
Docs ↗v0
(in-app Instructions)No repo instruction file — paste rules into v0's project Instructions.
Docs ↗Pricing shown is indicative, as of July 2026 — check each vendor's docs for current tiers.
The momentum killer
Agents go in circles when they spend their attention on lockfiles, build output, and vendored dependencies instead of your task. Tell them which files to skip and the loop usually breaks on the next session.
The $33.75-a-month lockfile
You are paying $3.00 per million tokens to show a reasoning model your pnpm-lock.yaml. A 900KB lockfile is about 225,000 tokens. Even leaking into 5% of each prompt’s context, that one file costs $33.75 a month at 50 prompts a day, and the model learns nothing about your code from it.
Week 6 repo entropy
A repo that started at 40 files is 4,000 by week six, with three abandoned components-old/ folders the agent still reads. You ask for a button color change and it refactors your auth. Paste your file tree once and the config files point it back at your real code.
Free where it builds trust. Paid where it saves time.
Free
The full triage intelligence, in your browser.
- Unlimited manifest triage — 100% client-side
- Five config files: AGENTS.md, CLAUDE.md, GEMINI.md, Cursor .mdc rules, copilot-instructions.md
- Estimated monthly savings counter with live model pricing
- No signup, no upload, no telemetry on your file list
Pro
Set-and-forget sync on every commit.
- Drop-in GitHub Action + git pre-commit hook
- Config files auto-regenerate on every push
- Zero-trust: we never hold repo write tokens — your runner commits
- API key management + lifetime savings stats