I Spy Code
A good explanation leaves evidence.
I Spy Code is an AI coding assistant workspace preview for investigating unfamiliar code. Explore unfiltered and uncensored model choice with your selected provider, and check each explanation against repository evidence.
What are you investigating?
FIELD GUIDE / START WITH A QUESTION| File | Question | Method |
|---|---|---|
| 01 | Can you make it happen again? | Reduce the report to a reproducible input and a comparison case. |
| 02 | Where did the value change? | Follow transformations across module boundaries. |
| 03 | Does the patch prove its claim? | Compare changed behavior with the evidence from relevant checks. |
| 05 | What would disprove the theory? | Choose an observation that separates plausible explanations. |
ILLUSTRATIVE CASE / A MISSING ZERO
The row existed. The filter removed it.
A quantity of zero disappears from an import. The observation is the missing row; a truthiness check is one possible explanation. Inspect the filter before blaming storage.
What evidence would settle this?
Trace the zero through parsing and filtering, then compare it with a missing quantity. A meaningful regression check should retain zero while rejecting the truly absent value. This is a worked example, not a claim about the reference software.
A question for your assistant
Read only. Explain the observed behavior of [function] using its implementation and callers. Cite repository evidence. Separate facts, hypotheses, and open questions. Propose one check that could disprove the leading explanation.
PROJECT + ENDPOINT + FIRST QUESTIONPrepare your first investigation