dothinker AI nav
中文 Search

Developer

Develop with Claude Code and Codex

Turn requirements into AI-agent tasks, then close the loop with tests, diff review, and human acceptance.

Best for

Software teams and independent developers that want faster delivery without losing review control.

Final output

Task breakdown, implementation diff, test results, and code review notes.

Workflow snapshot

Last checked: 2026-05-19

Complexity: 5-stage workflow
Sources: 6 Review Notes: Treat AI output as a draft and verify facts, rights, platform rules, and business claims before publishing.
Inputs

Clear goal, constraints, target file scope, and acceptance criteria. Existing issue, design, error log, or failing test output.

Tool chain

Claude Code -> Codex -> Cursor -> Clacky -> Lovable

Final output

Task breakdown, implementation diff, test results, and code review notes.

Human review

The change scope matches the task and avoids unrelated refactors. The diff is readable and test commands plus results are recorded.

  1. 01 Write a precise task spec
  2. 02 Make the agent read first
  3. 03 Change in small slices
  4. 04 Run tests and type checks
  5. 05 Review the diff before merge

Input Materials

  • Clear goal, constraints, target file scope, and acceptance criteria.
  • Existing issue, design, error log, or failing test output.
  • Allowed commands and directories that should not be modified.

Review Checklist

  • The change scope matches the task and avoids unrelated refactors.
  • The diff is readable and test commands plus results are recorded.
  • Data, auth, or deploy changes have a rollback path.

Common Failure Modes

  • The agent implements before reading enough context.
  • Tests are weakened to make them pass.
  • Cross-file edits become too large to review.

Output Template

Delivery note: goal / changed files / behavior change / test results / residual risk / follow-up tasks.

Recommended Tool Stack

Tools are organized by workflow role. Unlisted tools can be added to the library later.

1

Claude Code

Code implementation

Read the repository in the terminal, edit files, run commands, and complete development tasks.

2

Codex

Implementation and review

Apply planned code changes, run checks, and report risks.

3

Cursor

Interactive editing

Support local edits, codebase questions, and human handoff.

4

Clacky

App prototype

Turn natural-language requirements into a runnable full-stack app draft, then have engineers review architecture, code, and deploy settings.

5

Lovable

Fast validation

Turn a product idea into a clickable web app prototype for requirement clarity and early user feedback.

Complete Workflow

Use AI outputs as drafts; facts, copyright, platform rules, and business claims need human review.

  1. Stage 01

    Write a precise task spec

    Describe goal, scope, acceptance criteria, blocked areas, and checks to run.

    Reusable prompt 
    Turn this requirement into task cards suitable for an AI coding agent: {requirement}
  2. Stage 02

    Make the agent read first

    Ask for understanding, relevant files, and an implementation plan before editing.

  3. Stage 03

    Change in small slices

    Assign one clear subtask at a time to avoid broad refactors.

  4. Stage 04

    Run tests and type checks

    Use failures as diagnostic input while keeping human judgment over the fix direction.

  5. Stage 05

    Review the diff before merge

    Check unrelated changes, hidden side effects, mistaken deletions, and missing tests.

FAQ

Can this workflow publish automatically?

Not recommended. AI is useful for drafts, variants, and checklists, but facts, asset rights, and platform rules need human confirmation.

What if my tool stack is different?

Keep the workflow roles: ideation, generation, editing, review, and learning. Substitute specific tools with existing team accounts.

Sources

Last checked: 2026-05-19

  • Claude Code overview Anthropic Docs · Source used to verify the referenced tool capability and workflow boundary.
  • Using Codex with your ChatGPT plan OpenAI Help Center · Source used to verify the referenced tool capability and workflow boundary.
  • Cursor Codebase Indexing Cursor Docs · Source used to verify the referenced tool capability and workflow boundary.
  • ClackyAI Clacky · Source used to verify the referenced tool capability and workflow boundary.
  • Lovable Documentation Lovable · Source used to verify the referenced tool capability and workflow boundary.
  • WayToAGI 编码分类 WayToAGI · Source used to verify the referenced tool capability and workflow boundary.

Review Notes

  • Treat AI output as a draft and verify facts, rights, platform rules, and business claims before publishing.
  • Tool pricing, quotas, and capabilities may change; check official sources before purchase or automation.