interact with your world model.
Demos
Bare tasks go in, rich world model data comes out. Enrichment adds agent-native metadata — component tags, file paths, keywords, intent — and rich descriptions powered by your existing world model.
toto + claude code
Agents write to your Obsidian vault. Toto connects the dots. Watch as Claude Code generates documentation, plans, and session logs — and Toto reconciles them into the shared world model in real time.
claude code
Claude Code creates tasks, sets them in progress, and completes them — all through the Toto API. The same board you use, the same state, the same animations.
claude code
Git commits, file changes, and agent logs are semantically matched to human-defined tasks. The world model stays in sync without manual updates. Confident matches resolve automatically; ambiguous ones surface for human review.
toto reconcile
Cost optimization is the underutilized lever of the agent era. The most expensive model is rarely the right one for every task. A planner decomposes the work and routes each subtask to the cheapest capable model — capability and cost matched live.
toto router
When the work is done, /session scans your git log, generates a session summary, updates architecture docs, and creates a carry-forward list — all written to your Obsidian vault automatically.
claude code
Enrich
Bare tasks become semantic fingerprints.
One project. Five stages. The same OAuth feature flows through enrichment, planning, execution, reconciliation, and documentation.
toto lifecycle
Loading lists...