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