Save 46% on Claude Code Tokens

Block-level editing and persistent cache. Read and modify only the sections you need — backed by automated benchmarks across 9 real-world scenarios.

Get Pro — $49 Lifetime →

How It Works

Three simple steps. No config needed — install and start saving tokens immediately.

1

List Sections

Claude Code sees all sections in your Markdown file with their IDs, titles, and line ranges.

2

Read One Section

Read only the section you need — ~300 tokens instead of ~3,000 for a full file.

3

Replace One Section

Modify only that section — ~200 output tokens instead of rewriting the entire file.

Features

Everything you need to edit files efficiently in Claude Code.

Token Savings

  • ✓ Block-level editing for Markdown files
  • ✓ Persistent file cache (survives session restarts)
  • ✓ 46% average savings, up to 62% for power users

Zero Compromise

  • ✓ All processing local — no code leaves your machine
  • ✓ Auto-backup before every modification
  • ✓ Graceful fallback if cache is unavailable

Power Tools

  • ✓ Semantic search across documents (Pro)
  • ✓ English & Chinese interface
  • ✓ 9 pricing models tracked (Claude, GPT-4, DeepSeek...)

Backed by Real Data

9 automated scenarios, 3 user types, fully reproducible. No manual scoring.

46%

Average token savings across all scenarios

42-62%

Light Users
(docs, README)

26-54%

Medium Users
(single project)

42-62%

Heavy Users
(multi-project)

Best case: 62.1% (multi-project) · Worst case: 26.4% (cross-file bugfix) · View full results

Free vs Pro

Both versions get the full token-saving experience. Pro removes limits.

Feature Free Pro
Markdown block editing
Safety engine + backups
File cache (read_file)50 hits/dayUnlimited
ProjectsUp to 3Unlimited
Semantic search
Priority supportCommunity
Lifetime updates

Pricing

One-time payment. No subscriptions. No hidden fees.

Free

$0

No credit card required

  • Full block-level editing
  • 50 file cache hits/day
  • Up to 3 projects
  • Community support
Install Free

Install

One command to install. One command to activate.

# Install

pip install tokslash --extra-index-url https://tokslash.stratsapien.com/pypi/ --trusted-host tokslash.stratsapien.com

# Activate (after purchase)

tokslash activate --key YOUR-LICENSE-KEY

FAQ

How does TokSlash save tokens?

Instead of reading entire files, TokSlash lets Claude Code read and modify only specific blocks (sections) of a Markdown file. A 1,000-line README might have a 60-line "Installation" section — reading just that block saves ~94% of input tokens. Block-level writes save output tokens too.

What file formats are supported?

Block-level editing currently supports Markdown (.md) files only. Other text files (.py, .js, .json, .yaml, etc.) work with full-file read/write and benefit from the persistent cache. Binary files and files over 5MB are excluded from caching.

What happens when I use up 50 free cache hits?

TokSlash automatically falls back to full-file reads and writes. Nothing breaks. Your files are still editable — you just won't get the token savings until the next day (quota resets at midnight UTC).

Is my code sent anywhere?

No. TokSlash runs entirely locally as an MCP server. Your files, cache, and index never leave your machine. The only network call is a one-time license activation check (optional — works offline too).

What if a file gets modified incorrectly?

TokSlash automatically creates a backup in .tokslash/backups/ before every modification. You can restore any version at any time. The safety engine also verifies file integrity before and after each change.

Does it support Chinese Markdown?

Yes. Block editing is based on the CommonMark + GFM standard and works with Chinese content, CJK characters, and mixed-language documents.

Start Saving Tokens Today

Install free in 30 seconds. Upgrade to Pro when you're ready.

Get Pro — $49 Lifetime →