Conversations & History
The Intellia CLI exposes two read/search surfaces over your org’s activity:
conversations(aliasconv) — search, browse, and dispatch handoffs across the conversations the calling user can access.history— search across everything the org has done: conversations, feed outputs, and workflows.
Both groups emit machine-readable JSON. The --json flag is the default for most commands and is accepted as a no-op everywhere for compatibility. Add --pretty to any command for pretty-printed output. Results are scoped to what the calling user’s role and org membership grant — you only ever see conversations you can access and history within your own org.
Conversations
intellia conversations <command> [options]
intellia conv <command> [options] # aliaslist
List conversations the calling user has access to, most recent first. Returns a paginated envelope — read paginationHint in the response for a copy-pasteable next-page command, or use nextCursor directly.
intellia conv list
intellia conv list --agent-id <agentId> --limit 50
intellia conv list -q "lease renewal" --cursor 20| Option | Description |
|---|---|
--agent-id <id> |
Filter by agent UUID |
-q, --query <q> |
Filter by title / agent-name substring |
-l, --limit <n> |
Max results (default 20, max 100) |
-c, --cursor <n> |
Numeric offset for pagination |
--pretty |
Pretty-print JSON |
show
Print all messages from a single conversation in chronological order. Aliased as get.
intellia conv show <conversationId>
intellia conv get <conversationId> --pretty| Option | Description |
|---|---|
--pretty |
Pretty-print JSON |
search
Search past message bodies. Defaults to keyword search (tsvector); pass --semantic for embedding-based cosine search.
Keyword mode returns a paginated envelope with a paginationHint. Semantic mode returns the top-K matches by embedding similarity and ignores --cursor — raise --limit to widen the result set instead.
intellia conv search "bond lodgement"
intellia conv search "tenant wants to break lease" --semantic --limit 10
intellia conv search "invoice" --agent-id <agentId> --from-date 2026-01-01 --to-date 2026-03-31| Option | Description |
|---|---|
--semantic |
Use semantic (embedding) search instead of keyword |
--agent-id <id> |
Restrict to a single agent |
--conversation-id <id> |
Restrict to a single conversation |
--from-date <iso> |
Earliest message createdAt (ISO 8601) |
--to-date <iso> |
Latest message createdAt (ISO 8601) |
-l, --limit <n> |
Max results (default 20, max 100) |
-c, --cursor <n> |
Numeric offset for pagination (keyword mode only) |
--pretty |
Pretty-print JSON |
The response includes a mode field set to keyword or semantic.
send-to-agent
Fire-and-forget: dispatch a message to another agent. The reply lands as a system message in the calling conversation when the sub-agent finishes. End your turn after dispatch — do not poll for the result.
The callback conversation defaults to the INTELLIA_CONVERSATION_ID environment variable (set automatically inside an agent code-interpreter session). Outside such a session you must pass --callback-conversation-id explicitly or the command errors.
intellia conv send-to-agent <agentId> "Summarise the latest inspection report"
intellia conv send-to-agent <agentId> "Draft a reply" --callback-conversation-id <conversationId>| Option | Description |
|---|---|
-c, --callback-conversation-id <id> |
Conversation that receives the system-message reply. Defaults to INTELLIA_CONVERSATION_ID. |
--pretty |
Pretty-print JSON |
The response returns the jobId, status, and dispatchedAt for the dispatched handoff.
jobs
Inspect pending and completed agent-handoff jobs initiated by this user. Has two subcommands.
jobs status
Show status and a result preview for a single handoff job.
intellia conv jobs status <jobId>| Option | Description |
|---|---|
--pretty |
Pretty-print JSON |
jobs list
List recent handoff jobs initiated by this user, most recent first.
intellia conv jobs list
intellia conv jobs list --status pending
intellia conv jobs list --status completed --limit 50| Option | Description |
|---|---|
--status <s> |
Filter by status: pending, completed, failed, or orphaned |
-l, --limit <n> |
Max results (default 20, max 100) |
-c, --cursor <s> |
Pagination cursor from the previous response |
--pretty |
Pretty-print JSON |
History
intellia history <command> [options]search
Find prior work by keyword across conversations, feed outputs, and workflows — including archived items. Use it when a user refers to something they did before but can’t pinpoint it, then link the best match. Results are scoped to the calling user’s org.
intellia history search "end of financial year report"
intellia history search "rent increase" --types conversation,output
intellia history search "vacate notice" --limit 25 --no-include-deleted| Option | Description |
|---|---|
--types <list> |
Comma-separated subset of conversation, output, workflow (default all) |
-l, --limit <n> |
Max results per type (default 8, max 25) |
--no-include-deleted |
Exclude archived / soft-deleted items (included by default) |
--pretty |
Pretty-print JSON |
The response returns results plus a count of the total matches across all types.