Files
supabase/apps/studio/lib/ai/tools/studio-tools.ts
Matt Rossman d143571586 feat(assistant): trace-level scorers + server-side tool execution with needsApproval (#45654)
## Motivation

When Assistant runs a potentially destructive tool like `execute_sql`,
it stops the LLM request and prompts for client-side approval and
execution of the tool. After approval, a second request kicks off under
a separate trace. This has made scoring and
[Topics](https://www.braintrust.dev/blog/topics) classification
challenging, as the generated `output` is split across stateless
requests. The [span-level
scoring](https://www.braintrust.dev/docs/evaluate/custom-code#score-spans)
approach we've used thusfar (after the LLM call, we massage the result
into an `output` payload that's stuck onto the root span) has been
cumbersome and led to invalid scores / topics where only part of the
assistant response is considered. It's also inefficient, as we're
duplicating potentially large info (like the `search_docs` output) that
already exists within the trace.

An alternative to scoring spans is to [score
traces](https://www.braintrust.dev/docs/evaluate/custom-code#score-traces).
Braintrust [best
practices](https://www.braintrust.dev/docs/evaluate/score-online#best-practices)
advise:

> Use span scope for evaluating individual operations or outputs. Use
trace scope for evaluating multi-turn conversations, overall workflow
completion, or when your scorer needs access to the full execution
context.

We've also received [direct
guidance](https://supabase.slack.com/archives/C05QYJBLX89/p1777925770927149?thread_ts=1777905716.911979&cid=C05QYJBLX89)
from their team to use this approach.

## Changes

Migrates eval scorers from custom `AssistantEvalOutput` shape to
trace-level scoring via `trace.getThread()` / `trace.getSpans()`, with
thread parsing that scores the full latest Assistant turn and passes
prior conversation separately where relevant.

Moves `execute_sql` and `deploy_edge_function` from client-side
execution after approval to AI SDK `needsApproval` + server-side
`execute()`. SQL results returned to the model are gated by AI opt-in
level, so row data is only included with `schema_and_log_and_data`;
otherwise the tool returns the no-data-permissions sentinel.

Adds `metadata.isFinalStep` to disambiguate multiple LLM requests within
an "assistant" turn due to tool call requests/responses. For online
evals, this means we should configure automations to only score traces
with `metadata.isFinalStep = true` to ensure we're judging the complete
generated response.

Other minor kaizen changes:
- Renamed `promptProviderOptions` to `systemProviderOptions` to clarify
that this is associated with the "system" message and disambiguate from
the root `providerOptions`
- Adds `evals/trace-utils.ts` to handle Zod validation of the `unknown`
span shapes from Braintrust, to more easily access typed inputs/output
on tool spans.
- Bumps AI SDK floor version `^6.0.116` → `^6.0.174`
- Tweaked the "Conciseness" scorer to not unfairly dock points for the
new `[called tool_name]` labels in serialized assistant response

## Verification

In the studio staging build, I asked Assistant to create a todos table
with 3 sample todos. I manually approved the `execute_sql` call and saw
Assistant generate text before & after the call.

In Braintrust I verified two traces were produced (see [filtered
logs](https://www.braintrust.dev/app/supabase.io/p/Assistant/logs?v=Staging&tvt=trace&search={%22filter%22:[{%22text%22:%22metadata.environment%2520%253D%2520%27staging%27%22,%22label%22:%22metadata.environment%2520%253D%2520%27staging%27%22,%22originType%22:%22btql%22},{%22text%22:%22%2560Chat%2520ID%2560%2520%253D%2520%25221cb2ac45-e5e7-458c-9da4-3bf6863b8842%2522%22,%22label%22:%22Chat%2520ID%2520equals%25201cb2ac45-e5e7-458c-9da4-3bf6863b8842%22,%22originType%22:%22form%22}]})),
the first with `metadata.isFinalStep = false` and the second with
`metadata.isFinalStep = true`.

In the Braintrust staging scorers, I ran the preview Completeness scorer
on the second trace and verified it sees the complete Assistant response
including markers for tool calls ([link to
trace](https://www.braintrust.dev/app/supabase.io/p/Assistant%20(Staging%20Scorers)/trace?object_type=project_logs&object_id=b5214b62-ad1e-4929-9d5b-40b1daebe948&r=0ed0a4f8-8aff-4a34-bb1d-1df1d88a5070&s=ff9015f8-6bf7-4ab3-83a9-ca4e69e27e82))

<img width="1193" height="960" alt="CleanShot 2026-05-07 at 11 27 10@2x"
src="https://github.com/user-attachments/assets/509d4858-c3a1-4068-986d-3aa4d5617d1a"
/>

I also tested the `deploy_edge_function` workflow and verified it still
prompts for permission and warns on deployment of existing functions.

**References**
- https://www.braintrust.dev/docs/evaluate/custom-code#score-traces
-
https://ai-sdk.dev/docs/ai-sdk-core/tools-and-tool-calling#tool-execution-approval

Supercedes https://github.com/supabase/supabase/pull/45556 and
https://github.com/supabase/supabase/pull/45339

Closes AI-473

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **New Features**
* Tool actions (SQL execution, edge-function deploy) now require
explicit user Approve/Deny before proceeding.

* **Improvements**
* Assistant pauses for approval responses before sending follow-ups,
giving clearer control over risky actions.
  * Deploy/replace flows show confirmation and clearer replace warnings.
* Evaluation/scoring updated to use richer trace data for more accurate
assistant performance signals.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-05-12 15:24:21 -04:00

119 lines
4.1 KiB
TypeScript

import { tool } from 'ai'
import { z } from 'zod'
import { deployEdgeFunction } from '@/data/edge-functions/edge-functions-deploy-mutation'
import { executeSql } from '@/data/sql/execute-sql-query'
import type { AiOptInLevel } from '@/hooks/misc/useOrgOptedIntoAi'
import {
EDGE_FUNCTION_PROMPT,
PG_BEST_PRACTICES,
REALTIME_PROMPT,
RLS_PROMPT,
} from '@/lib/ai/prompts'
import { NO_DATA_PERMISSIONS } from '@/lib/ai/tools/tool-sanitizer'
import { fixSqlBackslashEscapes } from '@/lib/ai/util'
const KNOWLEDGE = {
pg_best_practices: PG_BEST_PRACTICES,
rls: RLS_PROMPT,
edge_functions: EDGE_FUNCTION_PROMPT,
realtime: REALTIME_PROMPT,
} as const
type KnowledgeName = keyof typeof KNOWLEDGE
export const executeSqlInputSchema = z.object({
// Transform at parse time so the corrected SQL is what gets stored in
// toolCall.input — ensuring evals and logs reflect what actually runs.
sql: z.string().describe('The SQL statement to execute.').transform(fixSqlBackslashEscapes),
label: z.string().describe('A short 2-4 word label for the SQL statement.'),
chartConfig: z
.object({
view: z.enum(['table', 'chart']).describe('How to render the results after execution'),
xAxis: z.string().optional().describe('The column to use for the x-axis of the chart.'),
yAxis: z.string().optional().describe('The column to use for the y-axis of the chart.'),
})
.describe('Chart configuration for rendering the results'),
isWriteQuery: z
.boolean()
.default(false)
.describe(
'Whether the SQL statement performs a write operation of any kind instead of a read operation'
),
})
export const loadKnowledgeInputSchema = z.object({
name: z
.enum(Object.keys(KNOWLEDGE) as [KnowledgeName, ...KnowledgeName[]])
.describe('The knowledge to load'),
})
export type StudioToolsContext = {
projectRef?: string
connectionString?: string
authorization?: string
aiOptInLevel?: AiOptInLevel
}
export const getStudioTools = (ctx: StudioToolsContext = {}) => {
const { projectRef, connectionString, authorization, aiOptInLevel = 'schema' } = ctx
const authHeaders = authorization
? { 'Content-Type': 'application/json', Authorization: authorization }
: undefined
return {
execute_sql: tool({
description:
'Asks the user to execute a SQL statement and return the results. Requires user approval before executing.',
inputSchema: executeSqlInputSchema,
needsApproval: true,
execute: async ({ sql }) => {
const { result } = await executeSql(
{ projectRef, connectionString, sql },
undefined,
authHeaders
)
return aiOptInLevel === 'schema_and_log_and_data' ? result : NO_DATA_PERMISSIONS
},
}),
deploy_edge_function: tool({
description:
'Asks the user to deploy a Supabase Edge Function from provided code. Requires user approval before deploying.',
inputSchema: z.object({
name: z.string().describe('The URL-friendly name/slug of the Edge Function.'),
code: z.string().describe('The TypeScript code for the Edge Function.'),
}),
needsApproval: true,
execute: async ({ name, code }) => {
await deployEdgeFunction({
projectRef: projectRef ?? '',
slug: name,
metadata: {
entrypoint_path: 'index.ts',
name,
verify_jwt: true,
},
files: [{ name: 'index.ts', content: code }],
authorization,
})
return { success: true }
},
}),
rename_chat: tool({
description: `Rename the current chat session when the current chat name doesn't describe the conversation topic.`,
inputSchema: z.object({
newName: z.string().describe('The new name for the chat session. Five words or less.'),
}),
execute: async () => {
return { status: 'Chat request sent to client' }
},
}),
load_knowledge: tool({
description:
'Load detailed knowledge about a Supabase topic before answering questions about it.',
inputSchema: loadKnowledgeInputSchema,
execute: ({ name }) => KNOWLEDGE[name],
}),
}
}