CommunityNews
AGENTS.md outperforms skills in our agent evals
A compressed 8KB docs index in AGENTS.md achieved 100% on Next.js 16 API evals. Skills maxed at 79%. Here’s what we learned and how to set it up.
Read in full here:
First Post!
mira_tokarev
That 8KB index result matches what I’ve seen: a small, curated map often beats a larger bag of snippets because the agent can actually keep it in working context. The trade-off is staleness—I’d treat AGENTS.md as a stable contract (entry points, invariants, test commands) and fetch volatile details on demand.
For evals, I’d also separate retrieval cost from task success and test after deliberate context truncation. Did the benchmark measure whether the index stayed correct as APIs changed, or only a fixed snapshot?
Popular Ai topics
Other popular topics
Sponsor Spotlight
Enabling companies to succeed by building software people love.
Categories:
Sub Categories:
Popular Portals
- /elixir
- /rust
- /wasm
- /ruby
- /erlang
- /phoenix
- /keyboards
- /python
- /js
- /rails
- /security
- /go
- /swift
- /vim
- /java
- /clojure
- /typescript
- /emacs
- /haskell
- /svelte
- /onivim
- /kotlin
- /react
- /c-plus-plus
- /crystal
- /tailwind
- /gleam
- /ocaml
- /elm
- /vscode
- /flutter
- /html
- /ash
- /deepseek
- /zig
- /opensuse
- /centos
- /php
- /react-native
- /scala
- /lisp
- /sublime-text
- /textmate
- /nixos
- /debian
- /agda
- /deno
- /django
- /kubuntu
- /nodejs
- /arch-linux
- /ubuntu
- /spring
- /revery
- /manjaro
- /julia
- /quarkus
- /lua
- /diversity
- /laravel
Devtalk Sponsors
Practical resources that improve the lives of professional developers.
We build trusted fault-tolerant systems that scale to billions of users.
Real-time error tracking, performance insights, and observability for devs.
Supporting innovation across the BEAM ecosystem.
We build reliable cloud platforms for business-critical systems.
Develop your skills with books, videos, and courses.
Catch errors, track performance, monitor hosts and more.
Enabling companies to succeed by building software people love.
Courses that move you from confusion to "Aha, now I get it!"
Publishing Linux Magazine, ADMIN, and MakerSpace for 25+ years.