OscoCoding agents & MCP
Coding agents that read your docs before they write code.
An agent is only as good as the context it can reach. Osco runs a hosted MCP server, so Claude Code, Cursor, VS Code and other MCP clients can read and update your docs, endpoints, features and tasks with your permissions.
The problem
What gets in the way today
- Agents guess at conventions and invent API endpoints that do not exist.
- Engineers paste the same context into prompts again and again.
- Docs fall behind the code as agents make changes nobody documents.
How Osco helps
Context that travels with the work
One connection, many agents
Create a personal access token in settings and paste one command or config snippet into your agent.
Your permissions, not more
Every MCP call runs with the role and project access of the person who connected it.
Agents that keep docs current
Ask an agent to arrange your repo's Markdown into folders, update docs after a change or build the feature graph.
Revocable at any time
Tokens are named and listed with their last use, and can be revoked individually.
How it works
From setup to everyday use
- 1
Connect
Create a token and add the Osco MCP server to Claude Code, Cursor or VS Code.
- 2
Read
The agent pulls the task, its acceptance criteria and the linked docs and endpoints.
- 3
Work
It makes the change in your repo with the right context.
- 4
Update
It updates the docs and tasks in Osco, and you review the result in the activity feed.
Frequently asked questions
Which agents work with Osco?
Any MCP-capable client. We provide setup snippets for Claude Code, Cursor and VS Code.
Can an agent delete our docs?
Only if the person who connected it has permission to delete. Deleted docs go to trash and can be restored.
Do you train AI models on our content?
No. Your content is not used to train AI models.
More use cases
Give every task the context to finish it.
Start free with your team. Bring your Markdown, connect your coding agents and invite people when you are ready.