state-trace
Bounded working memory for coding agents — typed nodes, current-vs-stale state, MCP-mountable.
Open the source (opens github.com) https://github.com/razroo/state-trace
Before you use it
- Runs on your computer with your permissions. Read what it can do before connecting it to an agent.
What it is
MCP servers from io.github.razroo.
The problem it solves and who it suits haven't been reviewed by InnerVault. The description above is the source's own.
How to get it
pypi package (stdio)
state-trace@0.3.1
From the server's entry in the official MCP Registry
Settings it asks for
STATE_TRACE_STORAGE_PATH· Path to the SQLite or JSON file backing the working-memory graph. .db/.sqlite/.sqlite3 = SQLite (recommended); other = JSON. Defaults to…STATE_TRACE_NAMESPACE· Namespace to scope reads and writes (e.g. the repo slug). Nodes without a namespace remain visible to all namespaces.STATE_TRACE_CAPACITY_LIMIT· Working-memory budget in capacity units. Default 256.
Related
-
GitHub MCP server
MCP serversGitHub's official MCP server. It lets an agent read and work with repositories, issues and pull requests.
https://api.githubcopilot.com/mcp/ -
Apify MCP server
MCP serversLets an agent run Apify's ready-made scrapers to pull data from websites, search engines, maps and social platforms.
-
Official MCP Registry
MCP serversThe MCP project's own registry of public MCP servers, so you can look for a connector before building one.
-
MCP reference servers
MCP serversSmall example servers from the MCP project (filesystem, git, fetch, memory and more) that show how MCP works.
-
Connect MCP servers to Claude Code
MCP serversAnthropic's guide to adding MCP servers to Claude Code, with examples of local and remote servers.
-
Playwright MCP
MCP serversLets an agent drive a real browser: open pages, click, fill in forms and read what's on screen.
claude mcp add playwright npx @playwright/mcp@latest