Agentic Task System (ATS)

What is Agentic Task System (ATS)?

ATS is an MCP server and CLI that gives your AI agent memory from the task system you already use — TickTick, Taskmaster, Beads, Obsidian, Notion, GitHub, Airtable, Google, or all at once via the composite adapter with Reciprocal Rank Fusion (RRF) retrieval. It is an adapter, not a migration — it makes your existing task app agent-native without re-homing any data.

How to use Agentic Task System (ATS)?

Install the CLI and an adapter globally (npm install -g @reneza/ats-cli @reneza/ats-adapter-ticktick), configure with ats config use ticktick, authenticate with ats auth login, then query with ats find "deployment runbook". For MCP clients, install @reneza/ats-mcp and set the ATS_ADAPTER environment variable; the binary (ats-mcp) works over stdio with Claude Code, Claude Desktop, Cursor, Windsurf, and OpenCode.

Key features of Agentic Task System (ATS)

• Two-way bus for agent reads and writes.
• First-fetch relevance via hybrid RRF-fused retrieval.
• Durable typed links (decision, depends-on, output, supersedes).
• Execution context: intent, lifecycle, security, ledger, promote, hierarchy.
• Bounded NDJSON events with recovery and dedup.
• Task graph linking structured nodes with proof and edges.
• Session-index handoff for durable task-linked summaries.
• Curated at write time using “trunk” themes.

Use cases of Agentic Task System (ATS)

• Give a coding agent persistent memory between sessions using your task app.
• Fuse retrieval across Notion, GitHub, and TickTick with one ats find query.
• Hand off context (intent, links, lifecycle) from one agent to another in a fresh session.
• Manage a task backlog with dependency links and auto-computed next actions.
• Audit agent actions via the action ledger while keeping writes add-only.

FAQ from Agentic Task System (ATS)

What makes ATS different from other agent memory solutions?

Most projects build a new store that drifts when not fed, while ATS uses your existing task app as the persistent memory. It combines dense, sparse, and keyword retrieval with RRF ranking, adds typed links, lifecycle validity, and an action ledger — whereas alternatives like memory files or vector-DB require manual upkeep or a separate database.

Does ATS work with the task app I already use?

Yes. It provides adapters for TickTick, Obsidian, Notion, GitHub, Airtable, Google (Sheets/Docs/Slides), Taskmaster, Beads, and OKF. A composite adapter fuses multiple sources into one deduplicated corpus. You keep using your current app; ATS adds the retrieval and context layer.

Do I need to host anything to use ATS?

You can deploy the backend on Render (one-click button) and the operator deck on Cloudflare Pages. The MCP server runs on your machine or VPS via Docker or the ats-mcp binary. After deploying, set the ATS_MCP_TOKEN and point your MCP client at the server’s URL.

What MCP clients are supported?

ATS works with any client that can launch a stdio MCP server, including Claude Code, Claude Desktop, Cursor, Windsurf, and OpenCode. The binary and ATS_ADAPTER environment variable are the same; only the config file location and wrapper key differ per client.

Where does my data live? Does ATS create a new database?

Your data stays in your existing task apps (TickTick, Notion, local markdown, etc.). ATS also uses a private Qdrant vector store and Ollama embeddings when deployed, but writes are add-only — it never drops or modifies data you added. Metadata is stored in flat YAML frontmatter on the task body and in ## Related / ## References sections.