Claude Code plugin

The gather plugin wires Hunta memory into Claude Code so it happens by construction, not by hoping the model remembers to use a tool.

Install

/plugin marketplace add hunta-ai/gather
/plugin install gather@hunta

Then point it at your tenant (mint an agent key in the console):

export GATHER_URL=https://mcp.hunta.ai   # or your own deployment on Estate
export GATHER_TOKEN=<your-agent-key>

Recall and capture are live on your next session.

What it does

• Auto-recall on every prompt. A UserPromptSubmit hook queries your memory and injects the top facts into the turn, so the agent starts with what it already knows.
• Re-inject after compaction. A SessionStart(compact) hook re-runs recall so injected context survives context resets.
• Loss-proof capture. PreCompact and SessionEnd hooks post a content-free breadcrumb to your staging inbox. The transcript is never uploaded.
• Curation commands. /gather:flush reviews pending candidates; /gather:status shows local recall hit-rate and latency.

Fail-open by construction

A slow or unreachable memory server never blocks a turn: hooks time out at 8 seconds, and every error path exits cleanly with no output. Automatic capture only ever adds candidates to staging, so canon is never touched on any error path. Worst case is a turn with no injected context, never a stalled turn.

Privacy

Everything goes only to the GATHER_URL you configure. Recall sends the prompt text as a search query; capture sends a breadcrumb (event, cwd, session id, timestamp) and never the transcript. Full data-flow table: docs/plugin.md.