ahel is live on Product Hunt today. Upvote

Compress - TokenStack Proxy Manager

SkillDev tools

This skill makes an AI rewrite its own text in a short, simplified style so its replies use far fewer tokens. Once added, your AI can cut its token usage by 65% while still getting its point across. It changes how your AI writes, not what it can do.

Available today. Use it from your connected AI after setup.

After adding it, ask your AI to compress its responses. It will rewrite its text in the shorter style so each reply costs fewer tokens.

Then ask your AI: use the Compress - TokenStack Proxy Manager skill

What your AI can do with it

  • Rewrite its replies in a short, simplified style
  • Cut token usage by 65%
  • Say the same thing with fewer words
  • Keep responses brief and easy to follow

What this skill tells your AI

The instructions your AI receives, as published by cwinvestments/memstack in skills/compress/SKILL.md and read by ahel’s review.

Monitor and troubleshoot the built-in TokenStack compression proxy for CC sessions.

Activation

When this skill activates, output:

Compress - Checking TokenStack status...

Then execute the protocol below.

  • Keywords: tokenstack, compression stats, token savings, proxy status, check proxy
  • Contextual: When user asks about token usage, context window limits, or session cost
  • Level: 2 (explicit trigger only)

Context Guard

ContextStatus
User says "tokenstack", "compression stats", "check proxy"ACTIVE - run status check
User asks about token savings or context windowACTIVE - point to dashboard report
Proxy errors or API connection failures appearACTIVE - run health diagnostics
General discussion about CC featuresDORMANT - do not activate
User is actively coding (no proxy issues)DORMANT - do not activate

What It Does

TokenStack is a transparent proxy between Claude Code and the Anthropic API. It compresses tool output before it reaches the API, extending effective context and lowering token cost.

This skill checks that the proxy is running and routing, and troubleshoots connection issues. For the full feature overview, enable steps, and transform tables, use the Token Optimization skill.

Prerequisites

TokenStack ships inside the memstack-skill-loader package. There is no separate install.

  • Start with the dashboard (recommended): python -m memstack_skill_loader dashboard --with-proxy
    • Starts the proxy on 127.0.0.1:8787 and sets ANTHROPIC_BASE_URL automatically.
  • Run only the proxy: python -m memstack_skill_loader proxy

Workflow

1. Health Check

curl http://127.0.0.1:8787/health

A healthy proxy responds on /health. If there is no response, the proxy is not running.

Note: the proxy has no /stats endpoint. Live savings are reported in the dashboard, not over curl. See "Reading Savings" below.

2. Confirm Routing

Open the dashboard and look at the proxy indicator. A live PRO or FREE badge means Claude Code traffic is routing through TokenStack. If the badge reads "not detected," the dashboard was started without --with-proxy; restart it with the flag.

3. Diagnose "not running"

Check the port:

netstat -ano | findstr 8787

If nothing is listening, start it:

python -m memstack_skill_loader dashboard --with-proxy

4. Reading Savings

Savings live in the dashboard (default http://localhost:3333):

  • Proxy indicator: PRO or FREE badge with session and 30-day percentages
  • Overview header: session and lifetime tokens saved
  • Burn Report: per-transform breakdown, estimated cost, time filters (all-time, daily, weekly, monthly), and a per-agent split for the Manager, Builder, and Reviewer

Tier Behavior

TierTransforms
FreeSix lossless text reductions, always applied
ProAdds seven transforms including AST truncation (license-gated)

A valid Pro license switches the proxy to Pro tier automatically. The dashboard badge shows the active tier.

Troubleshooting

SymptomFix
No response on /healthProxy not running. Start with python -m memstack_skill_loader dashboard --with-proxy.
Proxy indicator shows "not detected"Dashboard was started without --with-proxy. Restart with the flag.
Badge shows FREE but you hold ProLicense cache may be stale. The proxy revalidates the license on next start.
Cost figures differ from Anthropic ConsoleEstimates use list token prices and do not model server-side prompt caching. For billed cost, check console.anthropic.com.

Output Format

TokenStack Status
  Proxy: Running on :8787
  Tier: PRO
  Savings: read in the dashboard (Overview header and Burn Report)

Integration

  • Dashboard: proxy indicator, Overview header, and Burn Report
  • CC sessions: auto-routed when the dashboard is started with --with-proxy
  • Standalone proxy: python -m memstack_skill_loader proxy

Relationship to Other Skills

SkillScopeWhen to Use
Token OptimizationFull TokenStack overview, enable steps, transform tablesUnderstanding or turning on compression
Compress (this)Health, routing, troubleshootingProxy not running or not routing

Level History

  • Lv.1 - Base: health check and stats reporting for the legacy Headroom proxy. (Origin: MemStack, Feb 2026)
  • Lv.2 - Rewrite: retargeted to the built-in TokenStack proxy. Removed the external Headroom install and the /stats curl; savings are now read from the dashboard. (Jun 2026)

Signals

GitHub stars
421
Forks
44
Last commit
Sep 2026
Hacker News mentions
20
Advanced
Catalog kind
skill
Gateway key
compress
Source
github.com/cwinvestments/memstack