Add Base Prompt Rule functionality to Guardrail Policies #3598

krngrover94 · 2025-11-21T04:21:45Z

Introduced Base Prompt Rule in GuardrailPoliciesAction to auto-populate base prompts based on selected agent collections.
Updated GuardrailPolicies DTO to include BasePromptRule with properties for enabling, auto-detection, and confidence score.
Enhanced CreateGuardrailModal to manage Base Prompt Rule settings, including UI elements for configuration.
Modified GuardrailPolicies component to handle Base Prompt Rule data in the UI.
Added logic to fetch detected base prompts from agent collections.

- Introduced Base Prompt Rule in GuardrailPoliciesAction to auto-populate base prompts based on selected agent collections. - Updated GuardrailPolicies DTO to include BasePromptRule with properties for enabling, auto-detection, and confidence score. - Enhanced CreateGuardrailModal to manage Base Prompt Rule settings, including UI elements for configuration. - Modified GuardrailPolicies component to handle Base Prompt Rule data in the UI. - Added logic to fetch detected base prompts from agent collections.

devsecopsbot

🤖 AI Security analysis: "Automated analysis found no security issues in the five changed files. No immediate fixes required, but residual risks such as runtime logic flaws and dependency vulnerabilities may remain and warrant verification."

Risk Level	AI Score
🟢 NO RISK	5.0/100

Top 0 security issues / 0 total (Critical: 0, High: 0, Medium: 0, Low: 0)

Title	Location	Recommendation
—	—	No issues to display

🔗 View Detailed Report

…ic_base_prompt_intent_guardrail

devsecopsbot

🤖 AI Security analysis: "Automated scan found no issues across nine changed files, but changes could still introduce undetected vulnerabilities or logic errors. Ensure manual review and full CI/security testing for sensitive areas."

Risk Level	AI Score
🟢 NO RISK	5.0/100

Top 0 security issues / 0 total (Critical: 0, High: 0, Medium: 0, Low: 0)

Title	Location	Recommendation
—	—	No issues to display

🔗 View Detailed Report

…with names

devsecopsbot

🤖 AI Security analysis: "Automated scan found no issues across nine changed files, but changes could still introduce undetected vulnerabilities or logic errors. Ensure manual review and full CI/security testing for sensitive areas."

Risk Level	AI Score
🟢 NO RISK	5.0/100

Top 0 security issues / 0 total (Critical: 0, High: 0, Medium: 0, Low: 0)

Title	Location	Recommendation
—	—	No issues to display

🔗 View Detailed Report

rakshak-akto · 2025-12-02T06:15:23Z

apps/dashboard/src/main/java/com/akto/action/GuardrailPoliciesAction.java

+     * 
+     * Fetches detectedBasePrompt from the first selected agent collection
+     */
+    private void populateBasePromptIfNeeded(GuardrailPolicies policy) {


Duplicate logic in cyborg as well, can think of keeping it in some common place like libs/dao/src/main/java/com/akto/util

Also, we are not showing the auto detected base prompt in UI. Why do we need this then?

.../web/polaris_web/web/src/apps/dashboard/pages/guardrails/components/steps/BasePromptStep.jsx

rakshak-akto · 2025-12-02T06:28:21Z

.../web/polaris_web/web/src/apps/dashboard/pages/guardrails/components/CreateGuardrailModal.jsx

                                                <HorizontalStack gap="2" blockAlign="center">
-                                                    <Text
-                                                        variant="bodyMd"
+                            <Text 


Fix / restore the formatting in entire renderAllSteps block

.../web/polaris_web/web/src/apps/dashboard/pages/guardrails/components/steps/BasePromptStep.jsx

…ve placeholder detection logic. Update CreateGuardrailModal for better layout consistency and adjust BasePromptStep for confidence threshold labeling.

devsecopsbot

🤖 AI Security analysis: "Automated scan found no issues, but absence of findings does not guarantee safety. Multiple modified files increase regression and undetected-vulnerability risk; validate with additional targeted testing and reviews."

Risk Level	AI Score
🟢 NO RISK	10.0/100

Top 0 security issues / 0 total (Critical: 0, High: 0, Medium: 0, Low: 0)

Title	Location	Recommendation
—	—	No issues to display

🔗 View Detailed Report

rakshak-akto

As discussed,

lets show base prompt for all agents
show some messaging if base prompt could not be detected or confidence score is not enough or no single template found.
in future iteration, we can allow option to edit as well
In policy page, simply show a checkbox (no additional text prompts needed). This rule can be called something on the lines of Base prompt template enforcement (agent only)

rakshak-akto · 2025-12-04T06:04:34Z

...ashboard/web/polaris_web/web/src/apps/dashboard/pages/observe/api_collections/ApiDetails.jsx

                tabs={[
                    ValuesTab,
                    SchemaTab,
+                    ...(detectedBasePrompt ? [BasePromptTab] : []),


should show if isAgent = true (which is controlled by gen-ai tag
As discussed, show some messaging from backend ("Couldn't detect base prompt" or "No single / Multiple base prompt detected")

rakshak-akto · 2025-12-04T06:09:35Z

apps/dashboard/web/polaris_web/web/src/util/func.js

+   * @param {string} text - The text to search for placeholders
+   * @returns {Array} Array of placeholder objects with {start, end, phrase}
+   */
+  findPlaceholders: function(text) {


Simple regex match won't work?
Search for regex groups like these?
{...} {{...}}

…ase prompt auto-detection logic. Update BasePromptStep to enhance intent verification messaging and simplify state management. Adjust GuardrailPolicies DTO to reflect changes in base prompt handling.

devsecopsbot

🤖 AI Security analysis: "Automated scan found no issues, but absence of findings does not guarantee safety. Multiple modified files increase regression and undetected-vulnerability risk; validate with additional targeted testing and reviews."

Risk Level	AI Score
🟢 NO RISK	10.0/100

Top 0 security issues / 0 total (Critical: 0, High: 0, Medium: 0, Low: 0)

Title	Location	Recommendation
—	—	No issues to display

🔗 View Detailed Report

…d on the presence of a 'gen-ai' tag in the API collection. Remove detectedBasePrompt from ApiCollection DTO to streamline data handling.

devsecopsbot

🤖 AI Security analysis: "Review found no security issues in the change. Residual risk remains from undetected vulnerabilities or dependency/configuration regressions; continue automated scanning and verify security-focused tests before merging."

Risk Level	AI Score
🟢 NO RISK	10.0/100

Top 0 security issues / 0 total (Critical: 0, High: 0, Medium: 0, Low: 0)

Title	Location	Recommendation
—	—	No issues to display

🔗 View Detailed Report

krngrover94 requested a review from Ark2307 November 21, 2025 04:21

devsecopsbot bot reviewed Nov 21, 2025

View reviewed changes

krngrover94 requested a review from rakshak-akto November 21, 2025 07:46

Merge branch 'master' of github.com:akto-api-security/akto into agent…

021aa70

…ic_base_prompt_intent_guardrail

devsecopsbot bot reviewed Nov 30, 2025

View reviewed changes

Refactor CreateGuardrailModal to support new server selection format …

91a0fcc

…with names

devsecopsbot bot reviewed Nov 30, 2025

View reviewed changes

rakshak-akto reviewed Dec 2, 2025

View reviewed changes

Enhance ApiDetails to display detected base prompt template and impro…

2c430ee

…ve placeholder detection logic. Update CreateGuardrailModal for better layout consistency and adjust BasePromptStep for confidence threshold labeling.

devsecopsbot bot reviewed Dec 3, 2025

View reviewed changes

rakshak-akto reviewed Dec 4, 2025

View reviewed changes

rakshak-akto previously approved these changes Dec 4, 2025

View reviewed changes

Refactor GuardrailPoliciesAction and CreateGuardrailModal to remove b…

8f7fb1e

…ase prompt auto-detection logic. Update BasePromptStep to enhance intent verification messaging and simplify state management. Adjust GuardrailPolicies DTO to reflect changes in base prompt handling.

krngrover94 dismissed rakshak-akto’s stale review via 8f7fb1e December 4, 2025 09:31

devsecopsbot bot reviewed Dec 4, 2025

View reviewed changes

Refactor ApiDetails to conditionally display the base prompt tab base…

6a4d873

…d on the presence of a 'gen-ai' tag in the API collection. Remove detectedBasePrompt from ApiCollection DTO to streamline data handling.

devsecopsbot bot reviewed Dec 4, 2025

View reviewed changes

krngrover94 requested a review from rakshak-akto December 4, 2025 10:28

Ark2307 approved these changes Dec 4, 2025

View reviewed changes

krngrover94 merged commit 16861d0 into master Dec 4, 2025
6 checks passed

Add Base Prompt Rule functionality to Guardrail Policies #3598

Add Base Prompt Rule functionality to Guardrail Policies #3598

Uh oh!

Conversation

krngrover94 commented Nov 21, 2025

Uh oh!

devsecopsbot bot left a comment

Choose a reason for hiding this comment

Uh oh!

devsecopsbot bot left a comment

Choose a reason for hiding this comment

Uh oh!

devsecopsbot bot left a comment

Choose a reason for hiding this comment

Uh oh!

rakshak-akto Dec 2, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

rakshak-akto Dec 2, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

devsecopsbot bot left a comment

Choose a reason for hiding this comment

Uh oh!

rakshak-akto left a comment

Choose a reason for hiding this comment

Uh oh!

rakshak-akto Dec 4, 2025

Choose a reason for hiding this comment

Uh oh!

rakshak-akto Dec 4, 2025

Choose a reason for hiding this comment

Uh oh!

devsecopsbot bot left a comment

Choose a reason for hiding this comment

Uh oh!

devsecopsbot bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants