Team Knowledge Base

Scenario: Your team wants to capture and maintain institutional knowledge — architecture decisions, debugging tips, onboarding notes, and reusable procedures.

The Problem

Team knowledge lives in scattered places: Slack threads, meeting notes, individual memory, and outdated wiki pages. When someone leaves or a new person joins, critical context is lost.

The Workflow

1. Capture Decisions as Notes

When the team makes an architectural choice, record it:

notes_create({
  title: "Database: PostgreSQL over MongoDB",
  content: "We chose PostgreSQL because:\n- Strong consistency for financial transactions\n- JSON columns cover our semi-structured needs\n- Team has deep SQL experience",
  tags: ["architecture", "database"]
})

2. Connect Related Concepts

Build a knowledge web by creating relations between notes:

notes_create_link({
  fromId: "database-postgresql-over-mongodb",
  toId: "src/db/connection.ts::createPool",
  targetGraph: "code",
  kind: "documents"
})

Link notes to each other:

notes_create_link({
  fromId: "database-postgresql-over-mongodb",
  toId: "migration-strategy",
  targetGraph: "knowledge",
  kind: "related_to"
})

3. Save Procedures as Skills

Capture reusable procedures so they are discoverable:

skills_create({
  title: "Run database migrations",
  description: "How to create and apply database migrations",
  steps: [
    "1. Create migration: npm run migrate:create -- --name describe-change",
    "2. Write up/down SQL in the generated file",
    "3. Test locally: npm run migrate:up",
    "4. Verify: npm run migrate:status",
    "5. Commit the migration file"
  ],
  triggers: ["database migration", "schema change", "add column", "create table"],
  tags: ["database", "ops"]
})

Skills have triggers — short phrases that help the AI recall the skill when a matching task comes up.

4. Edit in Your IDE

Graph Memory mirrors notes, tasks, and skills to markdown files:

your-project/
  .notes/
    database-postgresql-over-mongodb/
      note.md
  .skills/
    run-database-migrations/
      skill.md

Team members can edit these files directly in their IDE or text editor. Changes are automatically imported back into the graph. This means the knowledge base is always editable as plain text — no vendor lock-in.

5. Search and Discover

The knowledge base is fully searchable:

notes_search({ query: "why did we choose PostgreSQL" })
skills_recall({ context: "how to deploy" })

Semantic search means you do not need to remember exact titles — searching by concept works.

Key Tools

Tool	Purpose
notes_create	Capture a decision, fact, or observation
notes_create_link	Connect notes to each other or to code/docs/tasks
notes_list_links	See all connections for a note
notes_search	Find notes by concept
skills_create	Save a reusable procedure
skills_recall	Find relevant skills for a task
skills_bump_usage	Track which skills get used
skills_link	Connect skills to each other
skills_create_link	Link a skill to code, docs, or tasks

Tips

• Use tags consistently across the team — they make filtering reliable.
• Add triggers to skills — they significantly improve recall accuracy.
• Link notes to code — it makes knowledge discoverable when working on related files.
• Review .notes/ and .skills/ in PRs — treat knowledge changes as part of the code review process.