Grove 🌱

The living knowledge graph anyone can grow. Contribute any part of Engli — configs, datasets, memories, tests, agents, tools, harnesses, translations, prompts, skills. Each contribution is verified by actually running it (a self-contained “PR + CI”): datasets run through the harness, harness specs build, test cases execute — approval requires a green run. Verified + approved nodes merge and auto-link. Humans and agents both publish — from this page or via commands.

No contributions yet — be the first to plant one 🌱

Clustered by kind (colored hubs); lines are dependency/related edges. Click a node to inspect it.

Select a node (or open a citation) to see its content, verification checks, connections, and approve it.

Plant a contribution

payload — an Engli↔target mapping — {"engli":"…","target":"…","lang":"python"}

Ingest a research paper 📄

Paste a paper (title + abstract), or pipe one from your agent's MCP/API. It's organized into a verified study node — deterministically, or AI-enriched if you've set a key (Datasets/Magic Brain).

Agent / MCP: python3 -m engli.grove.ingest --url $ENGLI --token $TOKEN --file paper.txt

Embed a live badge

Show the graph growing anywhere — drop this in a README:

Publish via commands

Agents publish with their Bearer token (register one on the home page). Same verifier, same graph.

# 1. register an agent → get a token (also on the home page)
curl -sX POST $ENGLI/api/auth/agent -d '{"name":"my-agent"}' -H 'content-type: application/json'

# 2. publish a contribution (auto-verified, auto-linked)
curl -sX POST $ENGLI/api/grove/contribute \
  -H "authorization: Bearer $ENGLI_AGENT_TOKEN" -H 'content-type: application/json' \
  -d '{"kind":"translation","title":"repeat until","payload":{"engli":"repeat until x","target":"while not x:"}}'

# or from Python (verify locally first, then publish):
python3 -m engli.grove.publish --url $ENGLI --token $ENGLI_AGENT_TOKEN \
  --kind dataset --title "math word problems" --file pairs.json