The product engineering brain — persistent decisions, hypotheses, and artifacts for teams building with AI agents.

Website · CLI · Agents · llms.txt

Decisions end up in Slack threads, Notion pages, and Linear tickets — without the reasoning attached. Agent sessions reset. CLAUDE.md goes stale. Notiky keeps product context in one graph and pushes it back to your tools.

  Inputs                    Engine                         Outputs
  -------                   ------                         -------
  customer calls      -->   capture & reason        -->    decisions
  meeting transcripts       compose                 -->    CLAUDE.md / .cursor/rules
  code repos                execute (MCP, daemon)   -->    ship plans
  Linear / Slack                                      -->    live MCP context

Capture  -->  Analyze  -->  Decide  -->  Update  -->  Deliver

• Capture — meetings, repos, conversations as structured signals
• Analyze — check new inputs against existing decisions
• Decide — version decisions with rationale and evidence
• Update — regenerate specs, artifacts, roadmap
• Deliver — push context to agents and teammates

• Decision/hypothesis graph (Postgres + pgvector)
• Artifacts: CLAUDE.md, .cursor/rules, AGENTS.md, skills
• MCP for Cursor, Claude Code, Codex
• Session capture from agent transcripts
• Linear / Jira mirroring; Granola, Otter, Google Meet imports
• Team workspaces

brew install notiky/tap/notiky
# or: curl -fsSL https://get.notiky.com | bash
notiky setup

Windows: irm https://get.notiky.com/install.ps1 | iex

See notiky-cli for details.

1. Sign up at notiky.com
2. Run notiky setup
3. Add Notiky MCP in Cursor or Claude Code
4. Capture a session or import a transcript

Read https://github.com/notiky/.github/blob/main/llms.txt

• AGENTS.md
• llms.txt

• Website
• CLI
• Contributing
• Security

Notiky is proprietary. Copyright 2026 Notiky.

The CLI is Apache 2.0.