---
title: "S-Corp taxes with two agents"
canonical: "https://agenticfeats.com/c/s-corp-taxes-with-two-agents-claude-code/"
agents: ["Claude Code", "Codex"]
evidence_grade: "C"
outcome: "completed"
job_type: "Finance and Admin"
roles: ["Finance", "Founder or owner"]
source_url: "https://common-sense.com/insights/doing-taxes-with-ai-coding-agents"
source_date: "2026-02-15"
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

# S-Corp taxes with two agents

Claude Code wrote Python scripts that compute the author's S-Corp tax forms from QuickBooks and bank data, while Codex reviewed the code.

Result: Codex caught real errors, including a wrong Schedule L field mapping, and the author found their own 2025 mistakes.

## How it was set up

QuickBooks data and bank statements in a private Git repo. Claude Code with Opus 4.6 writes the scripts. Codex with a high-reasoning model reviews them in multiple rounds.

## Worth knowing

The source does not say whether a CPA reviewed the final filings. The author has used a CPA since 1996.

## Prompt

```text
Review the tax-form scripts in [repo path]. Trace the data from [source JSON] through to the generated [CSV or form]. List findings ordered by severity, flagging any wrong field mappings.
```

## Evidence note

Author's first-person account with a concrete error example; no repo

## Sources

- Original: https://common-sense.com/insights/doing-taxes-with-ai-coding-agents

## The receipt

- Agent: Claude Code, Codex
- Evidence: Claim (grade C, claim only)
- Outcome: Completed
- Job type: Finance and Admin
- Roles: Finance, Founder or owner
- Source: Blog
- Published: 2026-02-15
- Added: 2026-10-13