Consistent code-comment style enforcer for LLM coding agents
Problem
Developers working with coding agents waste time manually cleaning up code comments before committing: every LLM comments differently, some horrendously, and there is no shared standard. Disabling comments entirely works but is hardly an alternative.
Opportunity
A linter/formatter-style tool (or agent skill/hook) that enforces a strict, configurable comment style guide - what to comment and what NOT to - rewriting all agent-generated comments into one consistent format before they reach production.
Market analysis
Real, daily friction for anyone using coding agents, and today's answers are crude: forbid-all-comments ESLint plugins or prompt instructions that agents ignore. The gap is a tool that enforces comment *policy*, not just comment syntax, but it shapes more like a popular open-source hook than a fundable product.
Market · Developers and teams using LLM coding agents daily; high-frequency pain, heavily discussed on HN and Reddit, but micro-scope as a product.
Pricing · Effectively zero willingness to pay as a standalone tool; lint/format tooling is free by convention. Any money is in team style-guide-as-config SaaS, a tough sell.
Pros
- + Universal, recurring annoyance with no purpose-built enforcer today.
- + Cheap to build: ESLint plugin plus agent hook covers the JS ecosystem fast.
- + Agent hooks (Claude Code, skills) provide a native distribution channel.
Cons
- − Prompt rules and existing plugins approximate it well enough for most teams.
- − Multilingual support (comment semantics per language) is a wide surface for a solo builder.
- − OSS-shaped tool: high adoption ceiling, near-zero revenue floor.
Existing / similar tools
Source
Hacker News (Ask HN)
The unexploited angle is that the valuable half of the problem is judgment, not formatting: deciding which comments deserve to exist requires understanding the code, which a deterministic linter cannot do. A hybrid that pairs cheap lint rules (length, placement, banned phrases like ‘now we’, restating the function name) with an optional LLM pass that deletes or rewrites low-signal comments would genuinely beat both prompt rules and forbid-all plugins. Shipped as an agent hook with a one-line install, it could become the default ‘comment hygiene’ layer; shipped as a paid SaaS, it almost certainly dies.