whoami --hermes
A command-line agent that claims it can "run entire software engineering tasks on its own".
Actually good, or just more hype? I ran it for 72 hours straight. Here's the answer.
cat chapter_01.md
From brew install hermes to the first prompt: 4 min 22 s in total.
cold start → first-successful-task, average of three runs
hermes run "refactor auth module to use pkce"
# hermes v0.9.2 · session 42a1 [plan] "Analyse src/auth/*.ts → find oauth flow → extract to pkce" [read] src/auth/oauth.ts // 214 lines [read] src/auth/token.ts // 88 lines [think] "Found implicit flow; switch to code+pkce, needs a new state param" [edit] src/auth/oauth.ts +43 -17 [edit] src/auth/token.ts +12 -4 [test] pnpm vitest auth PASS 18/18 [commit] "feat(auth): migrate to oauth2 code+pkce" [push] origin feat/pkce-auth ok # total time 3m 14s · 14k tokens · $0.21
The same 48 GitHub issues, run through both agents
echo $VERDICT
Worth installing, not yet worth fully relying on.
Three commands, and in under 5 minutes you'll see it do its first job.
# 1. install $ brew install hermes-agent/tap/hermes # 2. auth (have your anthropic api key ready) $ hermes auth login # 3. first task $ cd ~/your-repo && hermes run "add a CHANGELOG.md from git log"
exit 0
Full traces, the PR list for all 48 tasks and the benchmark scripts are at github.com/lewis/hermes-review