Show HN: OldHand A Claude/Codex plugin to verify the development flow end-to-end
OldHand is a Claude and Codex plugin shown on Hacker News that aims to verify the development flow end to end. The author built it as a Claude Code or Codex…
By Dillip Chowdary • Aug 05, 2026 • Source: HN Claude/Codex/Fable
OldHand is a Claude and Codex plugin shown on Hacker News that aims to verify the development flow end to end. The author built it as a Claude Code or Codex skill after heavy day-to-day use of AI agents for writing code. The convenience of those agents is clear; the gap appears when work must map to a Jira or Asana ticket in a corporate environment. Tasks get finished, but the output is not always up to the mark.
The core product idea is end-to-end verification of the development flow rather than only generating code. The author points to two failure modes that show up under ticket-driven work: AI hallucination, and unnecessary code when a reusable function already exists. OldHand is framed as a skill layered on Claude Code or Codex so that agent output can be checked against the real workflow constraints of ticket-backed delivery.
Advertisement
Tech Pulse Daily
Get tomorrow's pulse first
Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.
For engineers and builders, that matters because agent-assisted coding is already convenient enough to use on real tickets, yet quality still depends on catching invented work and missed reuse. In environments where Jira or Asana define scope and acceptance, incomplete or redundant changes create review load and drift from the ticket. A verifier aimed at the full flow targets that gap between "the agent wrote something" and "this matches how the team actually ships work."
The competitive context is the growing use of Claude and Codex style coding agents inside corporate ticket systems. Agents make writing code fast; ticket trackers still define what "done" means. OldHand sits in the middle as a Show HN project from the Claude, Codex, and Fable orbit: not another generator, but a plugin meant to keep agent output honest against an existing process.
The practical takeaway is to treat agent output on Jira or Asana work as incomplete until reuse and hallucination risks are checked. Watch whether OldHand can enforce that verification end to end in real ticket flows, especially catching duplicate logic when a function already exists, rather than only speeding up first drafts.
Advertisement
🔎 More interesting news
- Show HN: Clayrune – Run Claude Code agents in parallel without losing context
- Agent skills that bring team coding standards to Claude Code and Codex
- AI coding agents are blowing through budgets — Replit, Kilo Code, and Symbotic explain…
- Introducing Database Operations Agents: The future of autonomous database management
- Today's full Tech Pulse briefing →