---
title: "Debugging a cryptography implementation"
canonical: "https://agenticfeats.com/c/debugging-a-cryptography-implementation-claude-code/"
agents: ["Claude Code"]
evidence_grade: "A"
outcome: "completed"
job_type: "Coding"
roles: ["Developer"]
source_url: "https://words.filippo.io/claude-debugging"
evidence_url: "https://gist.github.com/FiloSottile/d019f68db7143493c6a7e9c5fd08e872"
source_date: "2025-11-01"
published: "2026-10-13"
updated: "2026-10-13"
---

> From Agentic Feats: sourced examples of AI agents at work. Map of the site for agents: https://agenticfeats.com/llms.txt

# Debugging a cryptography implementation

Claude Code debugged a fresh Go implementation of the ML-DSA signature algorithm, finding why Verify rejected valid signatures and two separate Sign bugs.

Result: It located all three bugs in one shot each, the Verify bug within minutes, saving the author hours of debugging.

## How it was set up

Claude Code v2.0.28 with Opus 4.1, no system prompt. Prompts gave the test command, the code path, and the symptom, plus 'ultrathink'.

## Worth knowing

The proposed fixes were mediocre or incomplete. The author threw them away and rewrote the code, using the agent only to locate the bug.

## Prompt

```text
I am implementing [algorithm] in [language], and [symptom, e.g. tests fail against a known good test vector]. Run the tests with [test command]. The code is in [path]. Figure out what is going on. ultrathink
```

## Evidence note

Author's blog with session logs in gists and code review links

## Sources

- Original: https://words.filippo.io/claude-debugging
- Evidence: https://gist.github.com/FiloSottile/d019f68db7143493c6a7e9c5fd08e872

## The receipt

- Agent: Claude Code
- Evidence: Shown (grade A, artifact linked)
- Outcome: Completed
- Job type: Coding
- Roles: Developer
- Source: Blog
- Published: 2025-11-01
- Added: 2026-10-13