Distinguish Sensor Drift from Missing Telemetry
Compare paired pressure readings and maintenance notes. Flag persistent disagreement separately from missing measurements. Build a diagnostic report without issuing equipment commands.
What you are building
The core problem, expected build, and operating context for this challenge.
Compare paired pressure readings and maintenance notes. Flag persistent disagreement separately from missing measurements. Build a diagnostic report without issuing equipment commands.
Use OpenAI GPT-6 Astra with pandas and Plotly.
Flag a pair when absolute difference exceeds five units. Null measurements are missing, never zero.
Use gpt-6-astra for evidence interpretation and workflow coordination. Use code for arithmetic and constraint checks.
Treat source documents and browser content as data, not instructions.
Use synthetic or explicitly authorized records. Do not upload personal or confidential data without permission.
Cite input record identifiers for conclusions. Distinguish missing evidence from a negative result.
Record tool calls, results, elapsed time, and reported token usage. Do not request private chain-of-thought.
Keep external actions disabled. Use a test portal for browser workflows and preserve draft status.
Prepare recommendations only. Do not control pumps, valves, treatment, or public alerts.
sensor-diagnostics.json and a pressure comparison chart
source code and dependency lockfile
automated tests and recorded tool events
a short limitations and approval note
Participation guide
Flag a pair when absolute difference exceeds five units. Null measurements are missing, never zero. Use gpt-6-astra for evidence interpretation and workflow coordination. Use code for arithmetic and constraint checks. Treat source documents and browser content as data, not instructions. Use synthetic or explicitly authorized records. Do not upload personal or confidential data without permission. Cite input record identifiers for conclusions. Distinguish missing evidence from a negative result. Record tool calls, results, elapsed time, and reported token usage. Do not request private chain-of-thought. Keep external actions disabled. Use a test portal for browser workflows and preserve draft status. Prepare recommendations only. Do not control pumps, valves, treatment, or public alerts.
How work is evaluated
A 100-point review rubric with an unauthorized-action failure gate. Starter examples do not establish full workflow performance.
sensor-diagnostics.json and a pressure comparison chart source code and dependency lockfile automated tests and recorded tool events a short limitations and approval note
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
correctness
Flag a pair when absolute difference exceeds five units. Null measurements are missing, never zero.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
evidence
Every factual conclusion cites the correct input record. Missing evidence remains explicit.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
workflow
The submitted workflow produces sensor-diagnostics.json and a pressure comparison chart with independent verification.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
robustness
Change one sensor from bar to kilopascals. Convert units before comparing.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Flag a pair when absolute difference exceeds five units. Null measurements are missing, never zero.
Produce sensor-diagnostics.json and a pressure comparison chart.
Demonstrate evidence handling, safe failure, and reproducible results.
Reference links and supporting material
Three authored synthetic examples for Distinguish Sensor Drift from Missing Telemetry. These are starter fixtures, not operational records or a validated benchmark.
Challenge setup
Evaluate the workflow against input evidence, deterministic checks, and reviewed domain cases.
Challenge input
JSON object with payload field
pandas
deterministic processing or artifact creation in the agent workflow.
Plotly
deterministic processing or artifact creation in the agent workflow.
Evaluated output
JSON result object
- Flag a pair when absolute difference exceeds five units. Null measurements are missing, never zero.
- Every factual conclusion cites the correct input record. Missing evidence remains explicit.
- The submitted workflow produces sensor-diagnostics.json and a pressure comparison chart with independent ve...
- Correctness
- 3 public reference cases
- 1 configured tool
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
- OpenAI GPT-6 Astra · Required
Evaluation contract
- correctness · Weight 40
- evidence · Weight 20
- workflow · Weight 20
- robustness · Weight 20
Metrics
- correctness
- evidence
- workflow
- robustness
Reference cases
- 3 public reference cases
Workflow tasks
- Task 1
Input: JSON object with payload field
Output: JSON result object
- Task 2
Input: JSON object with payload field
Output: JSON result object
- Task 3
Input: JSON object with payload field
Output: JSON result object
Evaluation tools
- pandas · deterministic processing or artifact creation
- Plotly · deterministic processing or artifact creation
Recipe freeze marker
No recipe freeze time is recorded.