BrowseFull catalogOutcomesSolve a specific problemRolesStack by teamTrustFilter by risk tier
← Back to the Claude Observatory

tokensafe

Skill Development Usable
Works inClaude Code
Usable Scanned — metadata only

Reduces token waste and hallucination rates by filtering irrelevant context from LLM prompts.

Your coding agent isn't getting dumber. Your context is just full of trash. Local context firewall & zero-loss runtime for Cursor, Claude Code, and autonomous agents.

46 starsMIT (commercial OK)FreeQuick setup
Usable rating — This tool is functional but has notable gaps. Review the evaluation notes below before deploying.

Reduces token waste and hallucination rates by filtering irrelevant context from LLM prompts. It optimizes inference costs and improves agentic reasoning accuracy by maintaining high-signal input for autonomous coding workflows.

Engineering teams using autonomous coding agents or IDE extensions to minimize context noise and operational costs.

Claude Code Claude Cowork Claude Chat

https://github.com/tokensafe-dev/tokensafe

By tokensafe-dev

How to Get It

Option 1: Claude Desktop App (Code Mode)Click the + button next to the prompt box → Plugins → Add plugin. Search and click Install. Skills work in Claude Code only.
Option 2: Paste into Claude CodeCopy the command below and paste it into your conversation. Claude will install it.
Command
claude plugins install tokensafe-dev/tokensafe

Tip: Paste this into a Claude Code conversation. Verify command matches your Claude Code version.

Auto-generated from the tool's public listing — not hands-on verified. Cross-check against the source repo's README before running.

First thing to try

After installing, paste this into Claude:

Help me clean up irrelevant code files before sending them to an AI agent
CostFree

Trust Signals Auto-scanned

Stars46Last updated2026-11-17LicenseMIT (OK for commercial use)Known CVEsNone foundSources: GitHub Advisory Database + OSV.dev · Scanned 2026-11-17 · scanner v1

Reviewer notes

Auto-scanned review. These are observations, not a security certification.

Scored from trust signals (evidence-eval-v1): 46 GitHub stars; contributors unknown; last commit 0d ago; license MIT.

Things to check

  • Scanned, not hands-on tested — this entry was auto-scanned from public metadata (GitHub metrics, license, security flags). No reviewer has run it, and no tool-specific limitations have been documented yet.

How to evaluate tools before deploying →

Data shown here comes from public APIs and automated scanning. Reviewer notes reflect one person's experience. This is not a security certification or legal recommendation. Always evaluate tools according to your own organization's policies.

Evaluation

Ease of Use
3/5
Versatility
2/5
Reliability
3/5
Security
3/5
Overall score2.75 / 5.00 UsableEvaluatedNov 2026
Scored from trust signals (evidence-eval-v1): 46 GitHub stars; contributors unknown; last commit 0d ago; license MIT.

← Back to the Claude Observatory

Rolling Claude out in your org? Let's talk.

Start a conversation →