-
Total Skill Fires
-
Unique Skills Used
-
Sessions Tracked
-
Skill Discovery
-
Active Skills
Project:
Last Session:
Tokens Saved:

Top Skills

No data yet

Categories

No data yet

Most Active Project

No activity

Recent Projects

No data yet

Session Diary

Loading diary entries...

Projects

Loading projects...
-
Total Tokens Burned
-
Estimated Cost
-
Agent Runs
-
Avg Tokens / Run
Loading compression data...

API Costs

-
Actual API Spend
-
API Input Tokens
-
API Output Tokens

Cost by Model

Model Calls Input Tokens Output Tokens Cost
No API cost data yet. Costs are tracked when agents run in API mode.

Cost by Agent

Agent Calls Input Tokens Output Tokens Cost
No API cost data yet. Costs are tracked when agents run in API mode.

API Spend Over Time

Tokens Over Time

Per-Agent Token Cost

Agent Fires Tokens Saved Est. Cost
Loading...

Recent Agent Runs

Loading...
📚

Memory Browser

Coming in v4.1

Search across sessions, lessons, and facts from your development history. Memory capture pipelines are being refined for the next release.

Launch Agent Task

Model Selection
Builder MCP Tools
Enter a working directory to discover MCP servers

Or from CLI: python -m memstack_skill_loader run "Your task"

Your Profile

Agent Names

Agent Runner Configuration

Agent Runner Setup

API key is optional. Required only when agents are set to API execution mode. Configure agent modes in the Agent Execution Mode section below.

API Key Headroom Proxy

Your API key is stored locally at ~/.memstack/api_key and never sent anywhere except Anthropic's API.

Injected into agent subprocesses for subscription billing. Stored at ~/.memstack/oauth_token.

Default MCP Tools

Set which MCP servers are blocked by default across all projects. Per-project settings override these defaults.

Saved
Plugin Manager

Plugins load globally across all Claude Code sessions. Changes take effect on the next CC session.

Click Refresh to load installed plugins
Git Workflow

Controls how the agent pushes code after commits. Per-project overrides the global default.

Agent Execution Mode

Controls how each agent is invoked. Subscription uses your Claude Code plan. API uses your Anthropic API key and is billed separately.

Important: If you have both a Claude subscription and an active API key in the Anthropic Console, Claude Code may bill API credits even in Subscription mode. To avoid unexpected API charges, disable your API key in the Console (console.anthropic.com > API keys) when using Subscription mode. Users with only a subscription or only an API key are not affected.

Dashboard Info

MemStack Version Dashboard URL localhost:3333 Python Version Platform

Paths

Skills Directory Pro Skills Directory Stats Database Agent Runner Sessions