perf(coderd/x/chatd): persist workspace agent binding across chat turns (#23274)

## Summary

This change removes the steady-state "resolve the latest workspace
agent" query from chat execution.

Instead of asking the database for the latest build's agent on every
turn, a chat now persists the workspace/build/agent binding it actually
uses and reuses that binding across subsequent turns. The common path
becomes "load the bound agent by ID and dial it", with fallback paths to
repair the binding when it is missing, stale, or intentionally changed.

## What changes

- add `workspace_id`, `build_id`, and `agent_id` binding fields to
`chats`
- expose those fields through the chat API / SDK so the execution
context is explicit
- load the persisted binding first in chatd, instead of always resolving
the latest build's agent
- persist a refreshed binding when chatd has to re-resolve the workspace
agent
- keep child / subagent chats on the same bound workspace context by
inheriting the parent binding
- leave `build_id` / `agent_id` unset for flows like `create_workspace`,
then bind them lazily on the next agent-backed turn

## Runtime behavior

The binding is treated as an optimistic cache of the agent a chat should
use:

- if the bound agent still exists and dials successfully, we use it
without a latest-build lookup
- if the bound agent is missing or no longer reachable, chatd
re-resolves against the latest build and persists the new binding
- if a workspace mutation changes the chat's target workspace, the
binding is updated as part of that mutation

To avoid reintroducing a hot-path query, dialing uses lazy validation:

- start dialing the cached agent immediately
- only validate against the latest build if the dial is still pending
after a short delay
- if validation finds a different agent, cancel the stale dial, switch
to the current agent, and persist the repaired binding

## Result

The hot path stops issuing
`GetWorkspaceAgentsInLatestBuildByWorkspaceID` for every user message,
which is the source of the DB pressure this PR is addressing. At the
same time, chats still converge to the correct workspace agent when the
binding becomes stale due to rebuilds or explicit workspace changes.

coderd/database/queries.sql.go

@@ -3823,7 +3823,7 @@ WHERE
             $3::int
     )
 RETURNING
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 `
 
 type AcquireChatsParams struct {
@@ -3862,6 +3862,8 @@ func (q *sqlQuerier) AcquireChats(ctx context.Context, arg AcquireChatsParams) (
 			&i.Mode,
 			pq.Array(&i.MCPServerIDs),
 			&i.Labels,
+			&i.BuildID,
+			&i.AgentID,
 		); err != nil {
 			return nil, err
 		}
@@ -4095,7 +4097,7 @@ func (q *sqlQuerier) DeleteChatUsageLimitUserOverride(ctx context.Context, userI
 
 const getChatByID = `-- name: GetChatByID :one
 SELECT
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 FROM
     chats
 WHERE
@@ -4124,12 +4126,14 @@ func (q *sqlQuerier) GetChatByID(ctx context.Context, id uuid.UUID) (Chat, error
 		&i.Mode,
 		pq.Array(&i.MCPServerIDs),
 		&i.Labels,
+		&i.BuildID,
+		&i.AgentID,
 	)
 	return i, err
 }
 
 const getChatByIDForUpdate = `-- name: GetChatByIDForUpdate :one
-SELECT id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels FROM chats WHERE id = $1::uuid FOR UPDATE
+SELECT id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id FROM chats WHERE id = $1::uuid FOR UPDATE
 `
 
 func (q *sqlQuerier) GetChatByIDForUpdate(ctx context.Context, id uuid.UUID) (Chat, error) {
@@ -4154,6 +4158,8 @@ func (q *sqlQuerier) GetChatByIDForUpdate(ctx context.Context, id uuid.UUID) (Ch
 		&i.Mode,
 		pq.Array(&i.MCPServerIDs),
 		&i.Labels,
+		&i.BuildID,
+		&i.AgentID,
 	)
 	return i, err
 }
@@ -4998,7 +5004,7 @@ func (q *sqlQuerier) GetChatUsageLimitUserOverride(ctx context.Context, userID u
 
 const getChats = `-- name: GetChats :many
 SELECT
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 FROM
     chats
 WHERE
@@ -5089,6 +5095,8 @@ func (q *sqlQuerier) GetChats(ctx context.Context, arg GetChatsParams) ([]Chat,
 			&i.Mode,
 			pq.Array(&i.MCPServerIDs),
 			&i.Labels,
+			&i.BuildID,
+			&i.AgentID,
 		); err != nil {
 			return nil, err
 		}
@@ -5154,7 +5162,7 @@ func (q *sqlQuerier) GetLastChatMessageByRole(ctx context.Context, arg GetLastCh
 
 const getStaleChats = `-- name: GetStaleChats :many
 SELECT
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 FROM
     chats
 WHERE
@@ -5192,6 +5200,8 @@ func (q *sqlQuerier) GetStaleChats(ctx context.Context, staleThreshold time.Time
 			&i.Mode,
 			pq.Array(&i.MCPServerIDs),
 			&i.Labels,
+			&i.BuildID,
+			&i.AgentID,
 		); err != nil {
 			return nil, err
 		}
@@ -5250,6 +5260,8 @@ const insertChat = `-- name: InsertChat :one
 INSERT INTO chats (
     owner_id,
     workspace_id,
+    build_id,
+    agent_id,
     parent_chat_id,
     root_chat_id,
     last_model_config_id,
@@ -5263,18 +5275,22 @@ INSERT INTO chats (
     $3::uuid,
     $4::uuid,
     $5::uuid,
-    $6::text,
-    $7::chat_mode,
-    COALESCE($8::uuid[], '{}'::uuid[]),
-    COALESCE($9::jsonb, '{}'::jsonb)
+    $6::uuid,
+    $7::uuid,
+    $8::text,
+    $9::chat_mode,
+    COALESCE($10::uuid[], '{}'::uuid[]),
+    COALESCE($11::jsonb, '{}'::jsonb)
 )
 RETURNING
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 `
 
 type InsertChatParams struct {
 	OwnerID           uuid.UUID             `db:"owner_id" json:"owner_id"`
 	WorkspaceID       uuid.NullUUID         `db:"workspace_id" json:"workspace_id"`
+	BuildID           uuid.NullUUID         `db:"build_id" json:"build_id"`
+	AgentID           uuid.NullUUID         `db:"agent_id" json:"agent_id"`
 	ParentChatID      uuid.NullUUID         `db:"parent_chat_id" json:"parent_chat_id"`
 	RootChatID        uuid.NullUUID         `db:"root_chat_id" json:"root_chat_id"`
 	LastModelConfigID uuid.UUID             `db:"last_model_config_id" json:"last_model_config_id"`
@@ -5288,6 +5304,8 @@ func (q *sqlQuerier) InsertChat(ctx context.Context, arg InsertChatParams) (Chat
 	row := q.db.QueryRowContext(ctx, insertChat,
 		arg.OwnerID,
 		arg.WorkspaceID,
+		arg.BuildID,
+		arg.AgentID,
 		arg.ParentChatID,
 		arg.RootChatID,
 		arg.LastModelConfigID,
@@ -5316,6 +5334,8 @@ func (q *sqlQuerier) InsertChat(ctx context.Context, arg InsertChatParams) (Chat
 		&i.Mode,
 		pq.Array(&i.MCPServerIDs),
 		&i.Labels,
+		&i.BuildID,
+		&i.AgentID,
 	)
 	return i, err
 }
@@ -5702,6 +5722,50 @@ func (q *sqlQuerier) UnarchiveChatByID(ctx context.Context, id uuid.UUID) error
 	return err
 }
 
+const updateChatBuildAgentBinding = `-- name: UpdateChatBuildAgentBinding :one
+UPDATE chats SET
+    build_id = $1::uuid,
+    agent_id = $2::uuid,
+    updated_at = NOW()
+WHERE
+    id = $3::uuid
+RETURNING id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
+`
+
+type UpdateChatBuildAgentBindingParams struct {
+	BuildID uuid.NullUUID `db:"build_id" json:"build_id"`
+	AgentID uuid.NullUUID `db:"agent_id" json:"agent_id"`
+	ID      uuid.UUID     `db:"id" json:"id"`
+}
+
+func (q *sqlQuerier) UpdateChatBuildAgentBinding(ctx context.Context, arg UpdateChatBuildAgentBindingParams) (Chat, error) {
+	row := q.db.QueryRowContext(ctx, updateChatBuildAgentBinding, arg.BuildID, arg.AgentID, arg.ID)
+	var i Chat
+	err := row.Scan(
+		&i.ID,
+		&i.OwnerID,
+		&i.WorkspaceID,
+		&i.Title,
+		&i.Status,
+		&i.WorkerID,
+		&i.StartedAt,
+		&i.HeartbeatAt,
+		&i.CreatedAt,
+		&i.UpdatedAt,
+		&i.ParentChatID,
+		&i.RootChatID,
+		&i.LastModelConfigID,
+		&i.Archived,
+		&i.LastError,
+		&i.Mode,
+		pq.Array(&i.MCPServerIDs),
+		&i.Labels,
+		&i.BuildID,
+		&i.AgentID,
+	)
+	return i, err
+}
+
 const updateChatByID = `-- name: UpdateChatByID :one
 UPDATE
     chats
@@ -5711,7 +5775,7 @@ SET
 WHERE
     id = $2::uuid
 RETURNING
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 `
 
 type UpdateChatByIDParams struct {
@@ -5741,6 +5805,8 @@ func (q *sqlQuerier) UpdateChatByID(ctx context.Context, arg UpdateChatByIDParam
 		&i.Mode,
 		pq.Array(&i.MCPServerIDs),
 		&i.Labels,
+		&i.BuildID,
+		&i.AgentID,
 	)
 	return i, err
 }
@@ -5780,7 +5846,7 @@ SET
 WHERE
     id = $2::uuid
 RETURNING
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 `
 
 type UpdateChatLabelsByIDParams struct {
@@ -5810,6 +5876,8 @@ func (q *sqlQuerier) UpdateChatLabelsByID(ctx context.Context, arg UpdateChatLab
 		&i.Mode,
 		pq.Array(&i.MCPServerIDs),
 		&i.Labels,
+		&i.BuildID,
+		&i.AgentID,
 	)
 	return i, err
 }
@@ -5823,7 +5891,7 @@ SET
 WHERE
     id = $2::uuid
 RETURNING
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 `
 
 type UpdateChatMCPServerIDsParams struct {
@@ -5853,6 +5921,8 @@ func (q *sqlQuerier) UpdateChatMCPServerIDs(ctx context.Context, arg UpdateChatM
 		&i.Mode,
 		pq.Array(&i.MCPServerIDs),
 		&i.Labels,
+		&i.BuildID,
+		&i.AgentID,
 	)
 	return i, err
 }
@@ -5917,7 +5987,7 @@ SET
 WHERE
     id = $6::uuid
 RETURNING
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 `
 
 type UpdateChatStatusParams struct {
@@ -5958,29 +6028,36 @@ func (q *sqlQuerier) UpdateChatStatus(ctx context.Context, arg UpdateChatStatusP
 		&i.Mode,
 		pq.Array(&i.MCPServerIDs),
 		&i.Labels,
+		&i.BuildID,
+		&i.AgentID,
 	)
 	return i, err
 }
 
-const updateChatWorkspace = `-- name: UpdateChatWorkspace :one
-UPDATE
-    chats
-SET
+const updateChatWorkspaceBinding = `-- name: UpdateChatWorkspaceBinding :one
+UPDATE chats SET
     workspace_id = $1::uuid,
+    build_id = $2::uuid,
+    agent_id = $3::uuid,
     updated_at = NOW()
-WHERE
-    id = $2::uuid
-RETURNING
-    id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels
+WHERE id = $4::uuid
+RETURNING id, owner_id, workspace_id, title, status, worker_id, started_at, heartbeat_at, created_at, updated_at, parent_chat_id, root_chat_id, last_model_config_id, archived, last_error, mode, mcp_server_ids, labels, build_id, agent_id
 `
 
-type UpdateChatWorkspaceParams struct {
+type UpdateChatWorkspaceBindingParams struct {
 	WorkspaceID uuid.NullUUID `db:"workspace_id" json:"workspace_id"`
+	BuildID     uuid.NullUUID `db:"build_id" json:"build_id"`
+	AgentID     uuid.NullUUID `db:"agent_id" json:"agent_id"`
 	ID          uuid.UUID     `db:"id" json:"id"`
 }
 
-func (q *sqlQuerier) UpdateChatWorkspace(ctx context.Context, arg UpdateChatWorkspaceParams) (Chat, error) {
-	row := q.db.QueryRowContext(ctx, updateChatWorkspace, arg.WorkspaceID, arg.ID)
+func (q *sqlQuerier) UpdateChatWorkspaceBinding(ctx context.Context, arg UpdateChatWorkspaceBindingParams) (Chat, error) {
+	row := q.db.QueryRowContext(ctx, updateChatWorkspaceBinding,
+		arg.WorkspaceID,
+		arg.BuildID,
+		arg.AgentID,
+		arg.ID,
+	)
 	var i Chat
 	err := row.Scan(
 		&i.ID,
@@ -6001,6 +6078,8 @@ func (q *sqlQuerier) UpdateChatWorkspace(ctx context.Context, arg UpdateChatWork
 		&i.Mode,
 		pq.Array(&i.MCPServerIDs),
 		&i.Labels,
+		&i.BuildID,
+		&i.AgentID,
 	)
 	return i, err
 }