vault-research
от lraivisto
DevOps & CloudКратко
High-velocity research orchestration engine.
Описание
Что делает этот навык
ResearchVault manages persistent state for agentic research using a local SQLite database. It stores artifacts, findings, and links across sessions, and supports parallel research paths through branches and hypotheses. A built-in synthesis engine discovers connections between findings using local embeddings.
Persistent local state and embedded synthesis mean research context survives across agent runs without relying on external memory services.
Когда его использовать
- Tracking findings across a multi-session research project
- Running parallel hypotheses on the same research topic
- Monitoring URLs and queries continuously for new developments
- Coordinating multiple AI agents working on the same research vault
- Flagging low-confidence data for automated verification
Пример сценария
User asks: research rising underground metal bands in Finland
- 1
Initialize a new vault project with an objective using vault.py init
- 2
Ingest sources by running scuttle against relevant URLs
- 3
Run synthesize to discover and link related findings
- 4
Run verify plan to identify and queue low-confidence findings for follow-up
- 5
Start the MCP server to share vault state with other agents
A structured SQLite vault with linked findings, verified facts, and hypothesis branches ready for further agent consumption
Установка
clawhub install lraivisto/vault-researchВыполните команду CLI или вставьте промпт в Claude Code, Cursor, Codex — любой кодинг-агент, которым пользуетесь.