# AI use cases for Old Coder (AmazingAng/old-coder) Repo: [AmazingAng/old-coder](https://github.com/AmazingAng/old-coder) | Use cases: 4 | Page: https://dedupedai.com/use-cases/amazingang-old-coder/ A markdown skill for coding agents that makes them write a test plan before coding and produce an evidence report after, so a human reviews those documents instead of the code. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Rerun the demo gauntlet to check the evidence yourself: https://dedupedai.com/use-cases/amazingang-old-coder/verify-rate-limiter-demo-gauntlet/ - Build a feature in Claude Code with a spec you approve first: https://dedupedai.com/use-cases/amazingang-old-coder/build-feature-with-spec-approval-claude-code/ - Add the Old Coder workflow to Codex or Cursor rules: https://dedupedai.com/use-cases/amazingang-old-coder/add-old-coder-to-agents-md-any-agent/ - Design and review an HTTP/JSON API against API gates: https://dedupedai.com/use-cases/amazingang-old-coder/design-review-http-json-api-old-coder-api/ ## How to rerun the demo gauntlet to check the evidence yourself with Old Coder Repo: [AmazingAng/old-coder](https://github.com/AmazingAng/old-coder) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Run the full gauntlet on the included rate limiter and confirm the reported numbers. **Who it is for:** A reviewer who wants to check the demo's claimed 41 tests, 100% coverage, and 22/22 planted bugs caught before trusting the approach. **The problem:** A report of green checks means little unless you can rerun the same checks yourself. **The outcome:** A fresh local run of the gauntlet that reproduces the report, with results you produced rather than the ones written in evidence.md. ### Steps 1. Clone the repo and open it: `git clone https://github.com/amazingang/old-coder` then `cd old-coder` (inferred, check the README) 2. Enter the demo folder: `cd demo-rate-limiter` 3. Create the environment and install: `python3 -m venv .venv && .venv/bin/pip install -r requirements-dev.txt -e .` 4. Run the whole report: `./tools/gauntlet.sh` 5. Compare the output with evidence.md and note any difference in test count, coverage, or mutation results. ### Prompt to paste into your agent ```text In the old-coder repo, go to demo-rate-limiter and run ./tools/gauntlet.sh after setting up .venv as the README describes. Report the test count, statement and branch coverage, and planted bugs caught, and state whether each matches evidence.md. ``` ### Caveat The demo is a rate limiter, and the README says the gauntlet cannot prove the spec is complete, so a green run shows only the constraints the demo encodes. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/amazingang-old-coder/verify-rate-limiter-demo-gauntlet/ ## How to build a feature in Claude Code with a spec you approve first with Old Coder Repo: [AmazingAng/old-coder](https://github.com/AmazingAng/old-coder) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Install Old Coder in Claude Code so the agent writes a test plan, waits for your yes, then ships with evidence. **Who it is for:** A developer using Claude Code who is about to have it change code in a high-stakes area such as money, logins, data, or concurrency and does not want to read every line. **The problem:** Reading all agent-written code does not scale, but approving only a short spec and a final report can keep the review manageable. **The outcome:** A feature built through the red, green, refactor cycle, with a SPEC you approved and an EVIDENCE report with checks marked pass, fail, or unverified. ### Steps 1. Install the skill for all projects: `npx skills add https://github.com/amazingang/old-coder --skill old-coder` 2. Or copy it into your Claude Code skills folder: `cp -r skills/old-coder ~/.claude/skills/` 3. Ask Claude Code for the change and invoke the skill with `/old-coder`. 4. Read the SPEC the agent writes with concrete must and must-not examples, and approve or reject it. 5. Read the EVIDENCE report, check any item marked unverified, and do not treat missing approval as high confidence. ### Prompt to paste into your agent ```text Using the old-coder skill in the repo, implement . First write the SPEC with concrete must and must-not examples and stop for my approval. After coding, produce an EVIDENCE report listing each check as pass, fail, or unverified, and confirm it came from a fresh run. ``` ### Caveat The skill is plain markdown that Claude Code loads from a skills folder, and the README states that it cannot authenticate its own checkers, so the SPEC approval is the main safeguard. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/amazingang-old-coder/build-feature-with-spec-approval-claude-code/ ## How to add the Old Coder workflow to Codex or Cursor rules with Old Coder Repo: [AmazingAng/old-coder](https://github.com/AmazingAng/old-coder) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Paste the Old Coder SKILL.md into AGENTS.md or a rules file so the agent follows the same gates. **Who it is for:** A team using Codex CLI, Cursor, or another instruction-following agent that does not load Claude Code skills. **The problem:** Without a shared workflow, each agent session may skip spec approval, weaken tests, or report checks that did not run. **The outcome:** A project whose agent instructions include the Old Coder workflow and its honesty rules, with the references directory kept beside it. ### Steps 1. Clone the repo: `git clone https://github.com/amazingang/old-coder` (inferred, check the README) 2. Copy the skill folder into your project: `cp -r old-coder/skills/old-coder /skills/` (inferred, check the README) 3. Add skills/old-coder/SKILL.md to your AGENTS.md, rules file, or system prompt. 4. Keep the references directory next to SKILL.md so the agent can load references/gauntlet.md. 5. Start a task and confirm the agent writes a SPEC and waits for approval before coding. ### Prompt to paste into your agent ```text Read skills/old-coder/SKILL.md in the repo and follow it for the task . Write the SPEC first and wait for my approval, then produce an EVIDENCE report that labels any unverified check as unverified. Verify that references/gauntlet.md was read before the gauntlet ran. ``` ### Caveat The README says the skill works with any agent that follows instructions, but it does not document how each agent loads AGENTS.md or rules files, so check that your agent actually reads the file. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/amazingang-old-coder/add-old-coder-to-agents-md-any-agent/ ## How to design and review an HTTP/JSON API against API gates with Old Coder Repo: [AmazingAng/old-coder](https://github.com/AmazingAng/old-coder) | Difficulty: Advanced | Time: 45 minutes | Basis: From the README Use the optional old-coder-api skill to check compatibility, auth, idempotency, pagination, and rate limits. **Who it is for:** A backend developer designing or changing an HTTP/JSON API who wants its contract checked before code is written. **The problem:** API changes can break clients or skip authorization, idempotency, or pagination without anyone noticing until production. **The outcome:** A reviewed API design whose gate decisions are recorded as SPEC constraints and gauntlet checks in the Old Coder workflow. ### Steps 1. Install the API skill: `npx skills add https://github.com/amazingang/old-coder --skill old-coder-api` 2. To use it with the workflow, install both skills: `npx skills add https://github.com/amazingang/old-coder --skill old-coder --skill old-coder-api` 3. Ask the agent to review your API change against the compatibility, authorization, idempotency, pagination, rate-limit, and operability gates. 4. Turn each gate decision into a SPEC constraint and a gauntlet check. ### Prompt to paste into your agent ```text Using the old-coder-api skill, review the API change described in for compatibility, authorization, idempotency, pagination, rate-limit, and operability. Output each gate as pass, fail, or unverified and list the resulting SPEC constraints and gauntlet checks. ``` ### Caveat The README describes the API skill's gates but not their detailed criteria, so the review is only as complete as the gate definitions the skill provides. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/amazingang-old-coder/design-review-http-json-api-old-coder-api/