What did Hlido score Bullet?

Bullet scored 73/100 (STEADY) in Hlido's independent, hands-on review.

Does any vendor pay Hlido for placement?

No. Hlido takes no money from the agents it rates — scoring weights stay private and the evidence behind every verdict is public.

Bullet

Name: Bullet review
Item: Bullet
Rating: 73
Author: Hlido Editor

Coding · tested 2026-08-21 · re-test due 2026-11-21 · by the Hlido desk, not the vendor

In short: A free, speed-obsessed coding agent (GUI + CLI) that routes/searches/executes with a tighter loop than the norm — YC-backed, with a headline 95.8% SWE-bench Verified claim you'll want to confirm.

4 PASS · 0 FAIL of 4 public-surface claims

Visit Bullet → scorecard.json

Quick answer

Bullet scores 73/100 (STEADY) on Hlido’s independent, hands-on test (reviewed 2026-08-21). STEADY (73) for a coherent, well-articulated coding agent with sensible efficiency protocols (model routing, targeted search, parallel execution with loop interception), cross-platform GUI+CLI, a free no-key start, and Y

Hlido Editor · 2026-08-21

Bullet is a coding agent built around one thesis: keep up with the developer. Its 'system' is three sensible protocols — Route/Escalate (send straightforward work to fast models, escalate to reasoning models only when the task demands it), Search/Acquire (targeted search and file reads instead of embedding the whole repo), and Parallel Execution (independent tool calls run together; duplicate calls and stuck loops are intercepted before they waste time). That's a coherent, credible design that maps to real pain — teams 'burning hours waiting on agent runs' — and the founders' note ('we were building with Claude Code every day, watching capable models move through needlessly slow machinery') is honest about where it came from. It ships cross-platform (macOS/Linux/Windows) as a GUI plus a free CLI (npm i -g @trybullet/cli), no key required to start, backed by Y Combinator. The thing to flag loudly: the site leads with '95.8% on SWE-bench Verified', a very strong claim that a homepage can't substantiate and that should be checked against the published results before you weight it. Speed and loop-efficiency claims are likewise the whole pitch and can only be judged by using it. For developers frustrated with slow agent runs, it's a genuinely interesting, free option — just verify the benchmark and try it on your own repo.

Why STEADY

STEADY (73) for a coherent, well-articulated coding agent with sensible efficiency protocols (model routing, targeted search, parallel execution with loop interception), cross-platform GUI+CLI, a free no-key start, and YC backing. Not higher because its headline differentiators — the 95.8% SWE-bench Verified claim and the speed/loop-efficiency promises — are exactly what a homepage cannot substantiate and must be independently verified. Not FADING because the design is credible, the product is real and shipping, and the free CLI makes it directly testable.

Public-surface checklist

PASS Homepage loads (required)
PASS Primary value prop (required) — A free, speed-obsessed coding agent (GUI + CLI) that routes/searches/executes wi
PASS Cta present
PASS Evidence or demo — 1 screenshot(s) captured

What we saw

1 screenshot captured by the Hlido engine during the reviewed run (run-dc777aea094124d8-codewithbullet-com). Our own captures — not vendor marketing material.

Bullet — run screenshot 1 (home.png) — home.png

What it does well

Coherent efficiency design: Route/Escalate (fast model by default, escalate only when needed), targeted Search/Acquire instead of whole-repo embedding, Parallel Execution with duplicate/stuck-loop interception
Cross-platform GUI (macOS/Linux/Windows) plus a free CLI (npm i -g @trybullet/cli), no key required to start
Honest, specific founders' note grounding it in real Claude-Code-workflow frustration
Free to use with no subscription, and Y-Combinator-backed

What it fails at

Leads with a very strong '95.8% on SWE-bench Verified' claim that the homepage cannot substantiate — needs independent confirmation
Speed and loop-efficiency are the core pitch but can only be judged by hands-on use, not from the surface
Crowded category (Cursor, Claude Code, aider, opencode) — differentiation rests on unverified performance
Maturity/adoption beyond YC backing and star-free landing copy isn't evidenced on the surface

Red flags

The '95.8% on SWE-bench Verified' headline is a strong, benchmark-topping claim presented on a marketing page — verify it against the published results before weighting it in a decision.
Speed/loop-efficiency, the entire differentiator, is unverifiable from the surface; benchmark it on your own repository.

Best for

Developers frustrated with slow agent runs who want a speed-focused coding agent to try for free
People who want both a GUI and a CLI from the same agent, cross-platform
Anyone wanting to A/B a routing/parallel-execution approach against their current coding assistant

Not recommended for

Buyers who need the SWE-bench claim independently verified before adopting
Teams standardised on a mature coding agent with proven track record and ecosystem
Users who need enterprise features/support (this is a free, early product)

Compared to

Aider speed-focused-new-gui-cli-agent-vs-mature-open-source-cli-agent
aider is a mature, widely-used open-source CLI coding agent with a strong track record; Bullet is a newer, free GUI+CLI agent competing on speed via model routing and parallel execution, with a bold (unverified) SWE-bench claim. Choose aider for a proven, transparent CLI; try Bullet if slow agent runs are your main pain and you'll verify its speed/benchmark yourself.

Agent relevance

CLI Behavioral-testable

Agentic-Commerce Readiness 47/100 · SURFACE-ONLY

Independent readiness for agent delegation & transaction. How it’s scored · check live

A coding agent available as a cross-platform GUI and a free CLI (npm i -g @trybullet/cli, no key to start), so its behaviour is directly testable on a real repository. It is an end-user coding agent rather than an agent-callable service, but the free CLI makes hands-on evaluation straightforward.

Agent-friendly score: 7/10

Evidence

A fast coding agent that routes, searches and executes to keep up with you; GUI for macOS/Linux/Windows + free CLI — source (2026-08-21) verified
Three protocols: Route/Escalate, Search/Acquire (targeted search, no whole-repo embed), Parallel Execution with loop interception — source (2026-08-21) verified
CLI: npm install -g @trybullet/cli; macOS & Linux, Node 18+, free, no key required to start — source (2026-08-21) verified
Headline claim: 95.8% on SWE-bench Verified (link: 'read the results') — source (2026-08-21) verified
Free to use, no subscriptions; backed by Y Combinator — source (2026-08-21) verified

scorecard.json · transparency passport · registry · methodology

More: compare agents · best of · developer tools · incident registry

Verdict by Hlido Editor, our automated editorial system · Method: public-surface-tier-1+editorial-narrative-v2 · Methodology version 2026.05 · Next review due 2026-11-21

How this page was produced. The scores, claim verdicts and evidence come from automated hands-on testing of the product’s public surface. The written analysis is drafted by an AI system, and pages publish without a person reviewing each one. Hlido publishes this record and answers for it — tell us if anything here is wrong and we will correct it.

Embed this trust badge

Live, always-current independent score — free to embed on your site or README. No vendor pays for placement.

Markdown

[![Hlido trust score](https://hlido.eu/badge/codewithbullet.svg)](https://hlido.eu/check/?agent=codewithbullet)

HTML

<a href="https://hlido.eu/check/?agent=codewithbullet"><img src="https://hlido.eu/badge/codewithbullet.svg" alt="Hlido trust score"></a>