# SwiftCNS Docs

> SwiftCNS is decision infrastructure for teams building with AI: test an idea against success criteria set up front, decide on evidence, and keep every decision on record (the Decision Record).

- MCP server: https://api.swiftcns.ai/mcp (streamable HTTP, OAuth). Scopes: cns:read, cns:write, cns:decide, cns:sources:read.
- Start: https://docs.swiftcns.ai/docs/get-started/install, then https://docs.swiftcns.ai/docs/get-started/first-test.
- Plans, data use and access: https://docs.swiftcns.ai/docs/use-the-app/plans-and-usage and https://docs.swiftcns.ai/docs/reference/security-and-data.
- Agents: read https://docs.swiftcns.ai/docs/agents/mcp-best-practices and https://docs.swiftcns.ai/docs/reference/mcp-server first. SwiftCNS proposes; the person calls the result, writes the Learning and records the Decision. Never pre-select a result or recommend a decision unless the person explicitly delegates it. Every page has a Markdown version at /llms.mdx/docs/<slug>/content.md; everything in one file at /llms-full.txt.


- [SwiftCNS Docs](/docs): Install SwiftCNS, take an idea to a recorded decision from the AI tool you already use, and learn the method behind it.

- **Get started**
- [Install SwiftCNS](/docs/get-started/install): Create your workspace, connect the AI tool you already use, and check the connection. About five minutes.
- [Your first test](/docs/get-started/first-test): Take one idea from a plain question to a recorded Decision, with SwiftCNS doing the drafting and you making every call.

- **The method**
- [How to test an idea](/docs/method): The four steps SwiftCNS gives every test, so your team sets success criteria up front, sees what's running and remembers why it decided.
- [1. Find the assumption](/docs/method/1-find-the-assumption): Name the one belief that would sink the idea if it were false, and turn it into a prediction you can test.
- [2. Set the success criteria](/docs/method/2-set-success-criteria): Choose the experiment and agree, before anything runs, what result counts as success. Approving it fixes the criteria.
- [3. Run the smallest experiment](/docs/method/3-run-the-experiment): Collect the evidence in one tracker while the experiment runs, then complete it so the result can't be quietly changed.
- [4. Decide, and record why](/docs/method/4-decide-and-record): Call the result against the success criteria, write what you learned, and record the decision so the next one starts from it.
- [Best practices](/docs/method/best-practices): What separates a test that changes your mind from one that only confirms it. Habits for each step, drawn from real runs.

- **Use the app**
- [Workspace and projects](/docs/use-the-app/projects-and-home): Find your way around SwiftCNS, from the workspace dashboard to a project's home, and see what's waiting on you.
- [The workbench](/docs/use-the-app/workbench): The screen where a question moves from idea to decision. The thread on the left, the current step and the record on the right.
- [Track experiments](/docs/use-the-app/track-experiments): See every experiment in a project on a board or a list, keep owners and due dates current, and log evidence in the tracker.
- [Learnings and Decisions](/docs/use-the-app/learnings-and-decisions): Write and publish a Learning, record a Decision, and find both again later in the project's registers.
- [Documents and plugins](/docs/use-the-app/documents-and-plugins): Bring the evidence you already keep into SwiftCNS by uploading files or importing from Google Drive, GitLab, Jira and Confluence, then use it in a conversation.
- [Team and roles](/docs/use-the-app/team-and-roles): Add people to your workspace, or share a project or conversation, and know who can do what.
- [Plans and usage](/docs/use-the-app/plans-and-usage): What each plan includes, how usage works, what you see when it runs out, and how to change plan.

- **Working with your AI tool**
- [Best practices with MCP and the app](/docs/agents/mcp-best-practices): Ask in your AI tool, watch and decide in SwiftCNS beside it. What to say at each step, and what to keep for yourself.
- [Permissions](/docs/agents/permissions): What your AI tool may read and do in SwiftCNS, which steps stay yours, and how to disconnect it.
- [Teach your AI tool when to use SwiftCNS](/docs/agents/teach-your-agent): Paste one rule into your AI tool's instructions so it reaches for SwiftCNS before a consequential decision, without being asked.
- [Pick up where you left off](/docs/agents/pick-up-later): Continue an experiment days later, start the next idea from your last decision, or replace a decision when the facts change.

- **Concepts**
- [The Decision Record](/docs/concepts/decision-record): What a Decision Record holds, why each part is there, and who reads it later.
- [Evidence and the snapshot](/docs/concepts/evidence-snapshot): Why SwiftCNS fixes the success criteria and the evidence at two moments, and why Not met and Unclear still count.

- **Reference**
- [MCP server reference](/docs/reference/mcp-server): Endpoint, authentication, permissions, and every SwiftCNS MCP tool grouped by method step.
- [Troubleshooting and FAQ](/docs/reference/troubleshooting-faq): Fixes for the problems people hit most, and short answers to common questions.
- [Security and data](/docs/reference/security-and-data): Who can see your work, what your AI tool and connected tools can reach, how AI uses your content, and how to export or delete it.
- [Glossary](/docs/reference/glossary): The method's words, and what SwiftCNS calls the same thing in the app and in your agent's replies.