Automated Zero-Day Triage Pipeline
In response to increasing zero-day attacks on edge devices, this challenge tasks you with building a resilient cybersecurity analytics engine. You will leverage Claude 4.5 via the Haystack framework to analyze vulnerability disclosures and generate structured threat intelligence. To ensure the reliability of long-running security scans and remediation steps, you will use Temporal for workflow orchestration. The system must ingest raw security bulletins, identify affected firmware versions using LLM-based parsing, and trigger automated testing environments to validate potential mitigations without human intervention.
What you are building
The core problem, expected build, and operating context for this challenge.
Develop an automated security response workflow using Claude 4.5 for LLM-driven threat analysis and Temporal for reliable distributed execution.
How work is evaluated
The evaluation focuses on the accuracy of the LLM parsing, the reliability of the Temporal workflow, and the structural integrity of the generated STIX intelligence.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Master the implementation of Haystack Pipelines for multi-stage security data processing
Design Temporal Workflows to manage the lifecycle of a security incident from discovery to patch
Build a custom Haystack Component that interfaces with Claude 4.5 to generate exploit mitigation strategies
Orchestrate Docker-based sandbox environments for automated vulnerability verification
Optimize prompt engineering for Claude 4.5 to reduce false positives in threat classification
Integrate real-time alerting systems using Temporal Signals to update security analysts on workflow progress
Implement persistent state management to track vulnerability status across large-scale enterprise networks
Reference links and supporting material
Sample data for 1 tasks
Challenge setup
The evaluation focuses on the accuracy of the LLM parsing, the reliability of the Temporal workflow, and the structural integrity of the generated STIX intelligence.
Challenge input
Raw text blob (security advisory)
Agent execution
No agent tool is listed for this challenge.
Evaluated output
JSON object with keys: cve_id, affected_versions, severity_score, mitigations
- Temporal workflow must reach a 'Completed' state without manual intervention.
- Extraction Precision target: 0.95
- Python execution harness
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
No tool records are attached.
Evaluation contract
- Temporal workflow must reach a 'Completed' state without manual intervention.
Metrics
- Extraction Precision ยท Target 0.95
Workflow tasks
- Task 1
Input: Raw text blob (security advisory)
Output: JSON object with keys: cve_id, affected_versions, severity_score, mitigations
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.