Canakkale 5.3.0

📅 Release Date: February 16, 2026 - 📅 Next Release Date: To Be Announced

Release Summary

Autom Mate Canakkale 5.3.0 is a capability expansion and operational resilience–focused release that introduces the Inspector experience, strengthens AI Agent and Playground usability, improves Monitoring visibility and navigation, and resolves multiple stability issues across integrations, logging, exports, licensing, and variable handling.

This release delivers new productivity tooling through Inspector modules (Search, Find & Replace, Undo/History, and Overview design), adds new connector capabilities (Apollo.io), and expands AI-supported experiences such as BI Estimator auto-fill and UI indicators. In parallel, critical runtime and platform fixes address Elasticsearch log durability and performance, safe handling of missing data and inputs, improved export correctness for large XLSX outputs, and clearer owner/creator visibility.

Canakkale 5.3.0 also includes targeted security remediation work for backend dependencies and MateAgent components, and introduces clearer compatibility documentation between AutomCenter, MateAgent, and AI Agent versions.

---

Highlights of this Release

🅱️ Bug Fixes

• Fixed AI Agent Library download flow to activate the downloaded version correctly.

• Fixed Playground streaming behavior when an API key is missing.

• Prevented double-click Dev Mode enable from producing “UNKNOWN-ERROR”.

• Implemented automatic API key generation for new users, new tenants, and seeded administrators.

• Fixed XLSX export corruption for 1000+ rows caused by malformed worksheet XML.

• Fixed AI Agent model seed logic to run correctly when a tenant already exists.

• Fixed AI Agent tenant_id registration failing during first launch initialization.

• Fixed Stop action returning XML as stringified JSON instead of proper application/xml.

• Added graceful handling when server is accessed during Desktop Action execution.

• Fixed Monitoring → Open Action in Autom to navigate to the correct action position.

• Fixed incorrect Owner/Creator display across Vault, Monitoring, and Autom lists.

• Fixed AI Agent Composer license consumption incorrectly counting RAG flow as a second flow.

• Removed incorrect “Trial Expired” message when valid license authorization exists.

• Added validation for Delay action when non-numeric input is provided.

• Fixed MateDB Get returning default values instead of empty when key is missing.

• Fixed import-time issue where Error List does not detect missing custom action credentials.

---

✨ New Features

• Introduced Inspector Search module for global search across an Autom.

• Introduced Inspector Find & Replace for action fields.

• Added Undo/History for Inspector Find & Replace operations.

• Designed Inspector Overview screen (first-time experience) with key metrics.

• Developed Apollo.io connector/app support.

• Added support for External Trigger XML / SOAP XML ingestion (prevent empty object payloads).

• Added daily scheduled AI job to auto-fill BI Estimator values (runs 4 times/day).

• Added UI indicator for AI auto-filled BI Estimator values.

---

🛠️ Enhancements

• Improved Scheduled Autom Monitoring list clarity (fields and actions).

• Display AI Agent version in AI Agent list view.

• Display Mate Agent version on the interface.

• Autom List: display username when creator name is masked (--- ---).

• Log Summary: display Action Execution Count / Number of Transactions.

• Improved error message clarity by providing detailed missing-key path in variable errors.

• Improved Inspector usability by moving the Inspector toggle button to the right side.

• Added UI notification for missing-field errors in older automs without opening the action.

---

Detailed Version

🅱️ Bug Fix

• AI Agent Library Download Activation Issue

  Previously, downloading a library when no version existed could leave the downloaded version inactive. A dedicated activation step was added between download and refresh, using the library version change endpoint to activate the downloaded version automatically.

---

• Playground Streaming Blocked When API Key Is Missing

  Streaming requests were blocked entirely when no API key existed, because the stream URL could not be constructed. The Playground now detects the missing API key scenario earlier and handles it in a controlled way instead of failing at request time.

---

• Prevent Double Click on Dev Mode Enabler

  Double-clicking the Dev Mode toggle could trigger an “UNKNOWN-ERROR” notification. The toggle interaction is now protected against double-trigger behavior to prevent erroneous UI errors.

---

• Auto API Key Generation for New Users / Admin / Tenants

  API keys were not consistently present across scenarios such as newly created users, seeded administrators, or newly created tenants. The system now checks and generates API keys automatically for these scenarios to ensure consistency and prevent missing-key failures.

---

• XLSX Export Corruption for 1000+ Rows

  XLSX exports could become unreadable in Excel clients due to an extra </worksheet> being injected every 1000 rows in sheet XML. The export logic was corrected to preserve a valid worksheet XML structure for large datasets.

---

• AI Agent Seed Data Bug for Models

  Model seeding was skipped when the tenant already existed, preventing required model/config setup from completing. Seeding logic was updated to run model-related seeding under the correct tenant-exists conditions.

---

• AI Agent Tenant ID Registration Fails on First Launch

  AI Agent initialization must register tenant_id once per environment, but the registration could fail during first launch. Initialization registration was corrected to ensure reliable one-time registration behavior.

---

• Stop Action Returns XML as Stringified JSON

  When Stop action response type was set to application/xml, the response body was still returned as JSON containing XML string content. Stop action now returns proper XML body with correct application/xml content-type behavior.

---

• Graceful Error Handling When Server Is Accessed During Desktop Action Execution

  Manually accessing the server during a running desktop action could cause unexpected failure without clear handling. The execution path now handles this scenario more gracefully and returns clearer failures instead of abrupt unexpected behavior.

---

• Monitoring “Open Action in Autom” Does Not Navigate to Action Position

  Monitoring could open the Autom page but fail to navigate to the referenced action location on the canvas. Navigation now correctly locates and focuses the referenced action.

---

• Incorrect Owner Displayed for Vault / Monitoring / Automs

  Owner/Creator fields were incorrectly shown as “administrator” even when created by another user, creating audit inconsistencies. Owner attribution now reflects the correct creator identity across affected screens.

---

• AI Agent Composer Consumes 2 Flows

  AI Agent Composer licensing incorrectly counted both the agent flow and its RAG flow, consuming 2 flows instead of 1. License consumption logic was corrected to avoid counting the RAG flow as an additional flow in this scenario.

---

• Remove “Trial Expired” Message When License Authorization Exists

  Systems could continue displaying “Trial Expired” even after a valid license with the AI Agent add-on was entered. UI license messaging was corrected to reflect the authorized state properly.

---

• Delay Action Ignores Non-Numeric Input Without Validation

  Delay action could complete without waiting when a non-numeric value was provided to Delay (ms), without validation or error. Input validation was added so invalid delay inputs are handled correctly and do not silently execute incorrectly.

---

• MateDB Get Returns Default Value Instead of Empty When Key Not Found

  When a key did not exist, the target variable could retain its default value instead of becoming empty, causing incorrect downstream logic. MateDB Get now sets the variable to empty when the key is missing.

---

• Error List Not Finding Custom Actions on Import

  After import, missing credentials for custom actions were not detected reliably in Error List, leaving fields unset without surfacing errors. Import validation and error detection were corrected to surface missing custom action credential dependencies properly.

✨ New Features

🛠️ Enhancements

---

User Notices & Guidance

This section highlights important behavioral changes, usage recommendations, and practical tips introduced in Canakkale 5.3.0. Reviewing these notes will help you adapt existing workflows smoothly and ensure optimal system usage. workflows smoothly and make the most of the new versioning and storage capabilities.

Warnings (Critical Behavior Changes)

• Authorization header behavior in RWS Actions now respects explicit user configuration. Systems relying on implicit JWT overrides should review integration settings.

• Expired licenses now return explicit expired states instead of generic responses. External monitoring systems should handle this state accordingly.

• Improved execution chaining may expose previously hidden logical flow misconfigurations. It is recommended to review complex error-handling chains after upgrading.

---

Hints (Productivity Tips)

• Use Monitoring hierarchy view to fully leverage improved parent-child log visibility.

• After updating credentials in Vault, no manual re-selection is required—flows will automatically use updated values.

• Review nested variable mappings in complex payload scenarios to ensure proper data propagation.

• Test chained Go To logic in Developer versions before promoting to Active.

---

Best Practices (Recommended Usage)

• Perform development and testing activities in Developer versions before releasing changes.

• Regularly monitor complex flow chains that rely on error handling and Go To transitions.

• Keep credential management centralized in Vault to benefit from automatic synchronization.

• Leverage Monitoring hierarchy improvements as part of routine operational validation.