feat(ENG-12325): add defender config option to StackOneToolSet by hiskudin · Pull Request #328 · StackOneHQ/stackone-ai-node

hiskudin · 2026-03-19T11:16:18Z

Summary

Adds a defender option to StackOneToolSet for explicit control over prompt injection detection behavior on every tool call.

Fixes a silent bug where defender_enabled was dropped by Zod validation before reaching the RPC call
Extends the RPC schema to forward block_high_risk, use_tier1_classification, use_tier2_classification per-request (backend changes tracked separately)
Field names match canonical DefenderSettings from @stackone/core

Behavior

Usage	Effect
`defender` omitted	SDK defaults: enabled, tier 1 + tier 2 on, never blocks
`defender: null`	Explicitly disabled — no scanning
`defender: { useProjectSettings: true }`	Defers to project dashboard settings
`defender: { enabled, blockHighRisk, ... }`	Explicit SDK-level config, overrides project settings

Default behavior (omit `defender`)

const toolset = new StackOneToolSet({ apiKey: '...' });
// → defender on, not blocking, tier 1 + tier 2 enabled

Disable entirely

const toolset = new StackOneToolSet({ apiKey: '...', defender: null });

Defer to project dashboard settings

const toolset = new StackOneToolSet({
  apiKey: '...',
  defender: { useProjectSettings: true },
});
// Combining with other fields throws ToolSetConfigError at construction

Explicit SDK-level config

const toolset = new StackOneToolSet({
  apiKey: '...',
  defender: {
    enabled: true,
    blockHighRisk: true,          // block HIGH/CRITICAL instead of just annotating
    useTier1Classification: true, // pattern-based detection
    useTier2Classification: true, // ML-based detection (requires onnxruntime-node)
  },
});

Parameters

Param	Type	Default	Description
`enabled`	`boolean`	`true`	Master switch. `false` disables all scanning.
`blockHighRisk`	`boolean`	`false`	Block tool calls with HIGH/CRITICAL risk. When `false`, risky content is annotated but still returned.
`useTier1Classification`	`boolean`	`true`	Fast pattern-based detection (regex rules for role markers, instruction overrides, etc.).
`useTier2Classification`	`boolean`	`true`	ML-based detection (ONNX model). Catches novel attacks patterns miss. Requires `onnxruntime-node` at runtime.

Changes

src/types.ts — DefenderConfig is now a discriminated union; adds DEFAULT_DEFENDER_CONFIG constant
src/schema.ts — adds block_high_risk, use_tier1_classification, use_tier2_classification to rpcActionRequestSchema
src/rpc-client.ts — forwards all new defender fields in request body
src/toolsets.ts — resolves defender config at construction, validates useProjectSettings exclusivity, applies three-way logic in createRpcBackedTool
src/index.ts — exports DEFAULT_DEFENDER_CONFIG

Notes

blockHighRisk, useTier1Classification, and useTier2Classification are forwarded in the RPC payload but require a backend change to ActionsRpcRequestDto to take effect. That work is tracked separately.

🤖 Generated with Claude Code

Adds a `defender` option to the `StackOneToolSet` constructor that allows controlling prompt injection detection behavior at the toolset level. - Add `DefenderConfig` interface with `enabled`, `blockHighRisk`, `useTier1Classification`, and `useTier2Classification` fields — matching canonical `DefenderSettings` names from `@stackone/core` - Add `defender_enabled` to `rpcActionRequestSchema` so it is no longer silently dropped by Zod validation - Forward `defender_enabled` through `RpcClient.rpcAction()` request body - Thread `defenderConfig.enabled` → `defender_enabled` in every RPC call made by `createRpcBackedTool` - Export `DefenderConfig` from the package index Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

pkg-pr-new · 2026-03-19T11:17:08Z

Open in StackBlitz

npm i https://pkg.pr.new/@stackone/ai@328

commit: d35d4a9

cubic-dev-ai

1 issue found across 5 files

Prompt for AI agents (unresolved issues)


Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="src/toolsets.ts">

<violation number="1" location="src/toolsets.ts:979">
P2: `dryRun` request serialization is now inconsistent with real RPC calls because `defender_enabled` is forwarded only in the live path.</violation>
</file>

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.}

src/toolsets.ts

Copilot

Pull request overview

Adds a toolset-level defender configuration to StackOneToolSet so consumers can control prompt-injection detection behavior (currently via per-request defender_enabled), and fixes a silent validation issue where defender_enabled was previously dropped before reaching the RPC call.

Changes:

Introduces DefenderConfig and exports it from the public entrypoint.
Extends the RPC request Zod schema to accept defender_enabled and forwards it in the RPC client request payload.
Adds defender to StackOneToolSet config and threads defender.enabled → defender_enabled into RPC calls.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
`src/types.ts`	Adds `DefenderConfig` interface describing toolset-level defender settings.
`src/toolsets.ts`	Accepts/stores `defender` config on the toolset and injects `defender_enabled` into RPC calls.
`src/schema.ts`	Updates `rpcActionRequestSchema` to validate `defender_enabled` so it isn’t stripped.
`src/rpc-client.ts`	Includes validated `defender_enabled` in the JSON payload sent to `/actions/rpc`.
`src/index.ts`	Re-exports `DefenderConfig` for SDK consumers.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.

src/toolsets.ts

src/schema.ts

- Include defender_enabled in dryRun payload to match live RPC path - Clarify JSDoc that only enabled is currently applied per-request - Add tests for defender_enabled forwarding in rpc-client and toolsets Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

…nd SDK defaults - Replace `DefenderConfig` interface with a discriminated union: - `{ useProjectSettings: true }` → defer to project dashboard settings - `{ enabled?, blockHighRisk?, useTier1Classification?, useTier2Classification? }` → SDK-level config - Add `DEFAULT_DEFENDER_CONFIG`: enabled=true, blockHighRisk=false, tier1+tier2 on - Omitting `defender` now applies SDK defaults (enabled, not blocking) rather than deferring to project settings — this makes SDK behavior explicit and predictable - `defender: null` explicitly disables defender for all tool calls - Constructor throws `ToolSetConfigError` if `useProjectSettings: true` is combined with other fields - Extend `rpcActionRequestSchema` and `RpcClient` to forward `block_high_risk`, `use_tier1_classification`, `use_tier2_classification` per-request (backend support for these is tracked separately) - Export `DEFAULT_DEFENDER_CONFIG` from package index Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

cubic-dev-ai

3 issues found across 5 files (changes from recent commits).

Prompt for AI agents (unresolved issues)


Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="src/toolsets.ts">

<violation number="1" location="src/toolsets.ts:995">
P2: `dryRun` no longer matches the real RPC payload for defender settings.</violation>
</file>

<file name="src/rpc-client.ts">

<violation number="1" location="src/rpc-client.ts:56">
P1: Don't forward the unsupported defender tuning fields yet. Only `defender_enabled` is wired through the RPC API today, so sending the other three keys can break RPC calls until the backend DTO is updated.</violation>
</file>

<file name="src/types.ts">

<violation number="1" location="src/types.ts:271">
P1: Defaulting omitted `defender` to an explicit config overrides existing project settings for every RPC call.</violation>
</file>

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.}

src/rpc-client.ts

cubic-dev-ai · 2026-03-19T16:00:36Z

src/types.ts

+ * Defender is enabled but outputs are never blocked — scans run and results are annotated only.
+ */
+export const DEFAULT_DEFENDER_CONFIG = {
+	enabled: true,


P1: Defaulting omitted defender to an explicit config overrides existing project settings for every RPC call.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At src/types.ts, line 271: <comment>Defaulting omitted `defender` to an explicit config overrides existing project settings for every RPC call.</comment> <file context> @@ -239,28 +239,37 @@ export interface ClaudeAgentSdkOptions { + * Defender is enabled but outputs are never blocked — scans run and results are annotated only. + */ +export const DEFAULT_DEFENDER_CONFIG = { + enabled: true, + blockHighRisk: false, + useTier1Classification: true, </file context>

src/toolsets.ts

…ults - Move defenderFields computation before dryRun block so both paths share the same logic - Fix TypeScript error: defenderConfig.enabled was accessed on the useProjectSettings union variant - Update tests to reflect new behavior: omitting defender now applies SDK defaults instead of undefined Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

The ternary expression used mixed tabs+spaces alignment which oxfmt rejects. Replaced with an if-else block that uses consistent tab indentation. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

SDK defaults will override project-level defender settings when no explicit defender config is passed. A console.warn nudges users to either pass an explicit config or opt into useProjectSettings: true. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

cubic-dev-ai

No issues found across 8 files

Replace flat defender_enabled/block_high_risk/use_tier*_classification fields with a nested defender_config object. Keep defender_enabled for backward compat. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

knip flagged defenderConfigRequestSchema as an unused export since it is only used internally within schema.ts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

glebedel · 2026-03-25T23:53:54Z

@hiskudin you now have merge conflicts

hiskudin requested a review from a team as a code owner March 19, 2026 11:16

Copilot AI review requested due to automatic review settings March 19, 2026 11:16

hiskudin changed the title ~~feat: add defender config option to StackOneToolSet~~ feat (ENG-12325): add defender config option to StackOneToolSet Mar 19, 2026

hiskudin changed the title ~~feat (ENG-12325): add defender config option to StackOneToolSet~~ feat(ENG-12325): add defender config option to StackOneToolSet Mar 19, 2026

cubic-dev-ai bot reviewed Mar 19, 2026

View reviewed changes

src/toolsets.ts Outdated Show resolved Hide resolved

Copilot AI reviewed Mar 19, 2026

View reviewed changes

src/toolsets.ts Outdated Show resolved Hide resolved

src/toolsets.ts Outdated Show resolved Hide resolved

src/schema.ts Show resolved Hide resolved

hiskudin and others added 2 commits March 19, 2026 11:49

cubic-dev-ai bot reviewed Mar 19, 2026

View reviewed changes

hiskudin and others added 2 commits March 19, 2026 16:48

fix: rewrite defenderFields as if-else to fix oxfmt formatting

f3f21e9

The ternary expression used mixed tabs+spaces alignment which oxfmt rejects. Replaced with an if-else block that uses consistent tab indentation. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

hiskudin marked this pull request as draft March 19, 2026 17:22

hiskudin and others added 3 commits March 20, 2026 10:47

refactor: ran format:oxfmt

6dc5766

Merge branch 'main' into feat/defender-config

37687f1

hiskudin marked this pull request as ready for review March 20, 2026 11:12

cubic-dev-ai bot reviewed Mar 20, 2026

View reviewed changes

hiskudin and others added 3 commits March 23, 2026 12:03

feat: nest defender fields under defender_config object

02bf4a2

Replace flat defender_enabled/block_high_risk/use_tier*_classification fields with a nested defender_config object. Keep defender_enabled for backward compat. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

fix: add defender_config to mock handler body type

bea8b05

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

fix: remove unused export from defenderConfigRequestSchema

d35d4a9

knip flagged defenderConfigRequestSchema as an unused export since it is only used internally within schema.ts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(ENG-12325): add defender config option to StackOneToolSet#328

feat(ENG-12325): add defender config option to StackOneToolSet#328
hiskudin wants to merge 11 commits intomainfrom
feat/defender-config

hiskudin commented Mar 19, 2026 •

edited

Loading

Uh oh!

pkg-pr-new bot commented Mar 19, 2026 •

edited

Loading

Uh oh!

cubic-dev-ai bot left a comment

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

cubic-dev-ai bot left a comment

Uh oh!

Uh oh!

cubic-dev-ai bot Mar 19, 2026 •

edited

Loading

Uh oh!

Uh oh!

cubic-dev-ai bot left a comment

Uh oh!

glebedel commented Mar 25, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

hiskudin commented Mar 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Behavior

Default behavior (omit defender)

Disable entirely

Defer to project dashboard settings

Explicit SDK-level config

Parameters

Changes

Notes

Uh oh!

pkg-pr-new bot commented Mar 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cubic-dev-ai bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

cubic-dev-ai bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

cubic-dev-ai bot Mar 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

cubic-dev-ai bot left a comment

Choose a reason for hiding this comment

Uh oh!

glebedel commented Mar 25, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

hiskudin commented Mar 19, 2026 •

edited

Loading

Default behavior (omit `defender`)

pkg-pr-new bot commented Mar 19, 2026 •

edited

Loading

cubic-dev-ai bot Mar 19, 2026 •

edited

Loading