A community brain.
Your data stays yours.
Opt-in collective intelligence, privacy-first by design.
Hive is an optional community cloud that lets every team benefit from anonymized patterns discovered across the Korva network — without ever sending identifiable code. It is off-limits by default-deny: nothing leaves your machine until it passes the privacy filter.
The problem Hive solves
Every team rediscovers the same patterns — the same webhook idempotency trick, the same migration pitfall. That knowledge stays locked inside individual vaults. Hive shares the shape of those lessons across the network, anonymized, so you inherit solutions instead of reinventing them.
What Hive does
Default-deny privacy filter
Every observation must pass a strict filter — PII, secrets and file paths removed — before a single byte is queued for upload.
Hybrid context
When enabled, vault_context queries your local SQLite and Hive in parallel with a 3-second hard timeout. Local results always take priority.
Instant kill-switch
Set KORVA_HIVE_DISABLE=1 and all outbound traffic stops immediately — no redeploy, no config edit, one environment variable.
Opt-in by design
Skills and private scrolls are never sent to Hive. Sync is content-addressed and limited to allow-listed observation types.
Hive in a context call
vault_context({ project: "payments" })
→ context: [...] // local SQLite
→ hive_context: [...] // community brain
→ hive_status: "ok" merged in 142ms
# If Hive is unreachable:
→ hive_status: "unavailable"
→ local context still returned ✓Hive at a glance
- Opt-in community cloud sync
- Default-deny privacy filter before any upload
- KORVA_HIVE_DISABLE=1 — instant kill-switch
- Hybrid local + community context
- Skills and private scrolls never leave your machine
korva hive statuskorva hive enablekorva hive push Tap the community brain
Hive ships with Korva, disabled-safe — enable it when your team is ready.
Install in 30 seconds