Project ideas from Hacker News discussions.

Claude Code now reads AGENTS.md if there is no Claude.md

📝 Discussion Summary (Click to expand)

1. Welcoming the move toward a shared standard
Many users see Anthropic’s recognition of AGENTS.md as a step toward interoperability and freedom from vendor lock‑in.
- “Finally doing something standards‑compliant instead of forcing users into a proprietary workflow.”askonomm
- “all i want is freedom to make a workflow where i don't feel tied to one provider…”sidrag22
- “we shouldn't have to do this on a per‑repo basis, Ant can choose to be a reasonable member of the ecosystem or not”verdverm

2. Technical quirks and missing pieces
Commenters point out practical hurdles: symlink handling (especially on Windows), lack of recursive detection for nested AGENTS.md, and still‑absent support for .agents/skills or MCP config.
- “Because Git can track symlinks and not hard links since they look like ordinary files.”wgd
- “I tried that before, it doesn’t work. Claude will not prioritize AGENTS instructions the same way it did for CLAUDE. ln was the only thing that worked for me.”gcampos
- Windows symlink discussion: “Windows supports it but git disables creating symlinks by default…” – lolpython
- “you'll need to wrap that in a find because of nested instruction files”verdverm
- “not as simple as that: nested AGENTS.md, skills, agent customization, MCP, … the list of misaligned harness features is still rather long”verdverm

3. Suspicions of reactive, market‑driven motives
A sizable contingent believes the change came only after user backlash and losing share to other harnesses, not from genuine community goodwill.
- “Remember they didn't do this because they wanted to help community, they did it because community was angry and they were losing users to other harnesses. Doesn't look like Anthropic care about dev community.”throwaw12
- “It's 'finally' because people have been asking for it for a long time… This was such a minimal thing to do… so finally seeing it happen, while nice, is too late for me.”verdverm
- “Anthropic in 2025: We can use our dominant market position to degrade the harness experiences of our competitors… Anthropic in 2026: We are losing our market position…”eleventen
- “They're probably losing market share. The only time you ever see tech companies make consumer-focused changes is when those consumers are jumping ship…”combobyte

4. Debate over model‑specific tuning vs a universal prompt file
Users split on whether a single AGENTS.md works across models or if model‑specific tuning remains necessary for optimal performance.
- “prompts should be tuned for models… blindly applying agents.md is probably an antipattern unless you want all models to basically converge to some common ill‑defined instruction following…”swyx
- “No thanks, I'm not tuning a bunch of files just for things to break when I switch models or a new one comes out.”arcanemachiner
- “By this logic you'd probably be wise to tier your claude.md by model (sonnet/opus) as well as effort level too, considering the varying failure modes.”TomGarden
- “In a 'one LLM only' environment, your instructions are by default tuned for said LLM. In a multi-LLM environment, roughly nobody will keep separate sets of instructions for each.”groby_b


🚀 Project Ideas

AgentLink

Summary

  • A cross‑platform CLI that creates, updates, and cleans platform‑appropriate links (symlinks/junctions) for AGENTS.md → CLAUDE.md (and .agents/skills → .claude/skills) recursively, removing the need for manual ln or git hooks.
  • Core value: zero‑config, works on Windows, macOS, and Linux, so developers never worry about platform‑specific link quirks again.

Details

Key Value
Target Audience Developers using Claude Code, Codex, or any LLM agent harness that relies on AGENTS.md/CLAUDE.md conventions
Core Feature Auto‑detects AGENTS.md (and nested .agents/skills) and creates/updates the appropriate links or junctions for the current OS, with a --watch mode to keep them in sync
Tech Stack Rust (for portable filesystem actions) + Clap CLI; optional Node.js wrapper for npm publish
Difficulty Medium
Monetization Hobby

Notes

  • HN commenters lamented the hassle: “we shouldn't have to do this on a per‑repo basis” (verdverm) and the need for post‑checkout hooks (nomel). AgentLink eliminates that boilerplate.
  • Enables seamless switching between agents and models without breaking existing workflows, a practical utility that would spark discussion on standardizing agent configuration.

AgentInject

Summary

  • VS Code extension that reads AGENTS.md (and optional model‑specific variants like CLAUDE.md, GEMINI.md) and automatically injects their contents into the LLM prompt context for any active agent session.
  • Core value: removes manual token waste and ensures the agent always sees the latest instructions without the user having to maintain duplicate files.

Details

Key Value
Target Audience VS Code users who interact with LLM coding agents (Claude Code, Codex, Copilot, etc.)
Core Feature On file save or agent activation, injects AGENTS.md content into the prompt; supports fallback hierarchy and model‑specific overrides via front‑matter
Tech Stack TypeScript + VS Code Extension API; uses tree‑sitter to parse markdown front‑matter
Difficulty Low
Monetization Hobby

Notes

  • Commenters complained about token waste: “Every turn means more tokens…” (jwolfe) and the desire for a harness that “can try its own file and fall back to generic” (pishpash). AgentInject does exactly that.
  • Provides a concrete way to experiment with model‑specific tuning while keeping a single source of truth, likely to generate lively debate on prompt engineering practices.

AgentHub

Summary

  • Hosted service (SaaS) where teams can store, version, and share AGENTS.md, skills, and MCP configurations; harnesses fetch the latest version via a simple API, optionally applying model‑specific transforms.
  • Core value: centralizes agent configuration, eliminating per‑repo drift and enabling organization‑wide standards without sacrificing flexibility.

Details

Key Value
Target Audience Teams and organizations that standardize on LLM agent workflows across multiple projects
Core Feature Git‑backed repository with web UI, CLI, and REST API; supports branch‑based versioning, pull‑request reviews, and webhook‑triggered updates to local harnesses
Tech Stack Go backend (Gin), PostgreSQL, React frontend; hosted on Fly.io or similar
Difficulty High
Monetization Revenue-ready: Tiered subscription (Free for individuals, $9/user/mo for teams)

Notes

  • Users expressed frustration with proprietary workflows: “Finally doing something standards‑compliant instead of forcing users into a proprietary workflow.” (askonomm) AgentHub offers a neutral, community‑driven alternative.
  • Would likely become a focal point for discussion on governance of agent configuration, similar to how language servers or package registries evolved.

AgentOptimizer

Summary

  • Static analysis tool that scans AGENTS.md (and related skill files) to estimate token cost, detect redundant or contradictory instructions, and suggest a minimal, high‑impact instruction set tailored to a target model.
  • Core value: reduces prompt bloat, improves agent reliability, and saves costly LLM tokens while preserving essential guidance.

Details

Key Value
Target Audience Power users and prompt engineers who fine‑tune agent behavior and care about token efficiency
Core Feature Parses markdown, builds an instruction dependency graph, runs a token estimator (using model‑specific tokenizer), and outputs a diff‑style report with recommendations
- Tech Stack Python (Click) + tiktoken for token counting; optional Rust core for speed
Difficulty Medium
Monetization Hobby

Notes

  • Many commenters worried about context clouding: “adding relevant information still eventually clouds context” (Madmallard) and the need to keep instructions brief. AgentOptimizer directly addresses that pain.
  • Provides measurable token savings, a point of interest for HN readers who enjoy performance‑focused tooling and would likely share benchmark results in the comments.

Read Later