Gonfire
Show HN: Sneakily steer candidates toward naive brute-force solutions
Last verified:
What is Gonfire?
Gonfire is an assessment platform built for engineering teams to evaluate how candidates actually work with AI. The core assumption is that candidates will use AI during coding tasks, and the way they steer the AI is the signal worth measuring. Instead of judging only the final pull request, Gonfire captures every interaction—prompts, edits, reverts, test runs—throughout the candidate's session, providing visibility into their planning, decision-making, and recovery processes .
The platform replaces traditional take-home assignments that no longer work in the AI era. Today, most candidates ship clean PRs generated mostly by AI assistants, making the output look similar regardless of whether the candidate carefully designed the solution or one-shot prompted the AI. Gonfire solves this by showing the full process: how candidates plan before prompting, where they push back when AI does the wrong thing, and when they revert and try different approaches .
Gonfire is designed for engineering leaders and hiring teams who are tired of grading take-homes and guessing what candidates actually did. The platform works with any GitHub repo or zip file as starter source, supporting TypeScript, Python, Rust, and Go. Setup is low-effort—just paste a repo URL and write a brief—and candidates have minimal setup required compared to local environment hell .
Key features include live capture of all AI interactions end-to-end, behavioral attribution tracking which prompt led to which decision, decision point identification where candidates paused/reverted/accepted AI output, and a full interaction trail per session. Review time drops from 2+ hours staring at a diff to approximately 30 minutes reviewing the assessment .
Gonfire pricing
Pricing model: Freemium
No pricing yet. Gonfire is in early access and calibrating the product with a small set of design partners before publishing pricing. If you want to use it on a real role, you need to get in touch and they will set up an account and walk you through it .
Gonfire pros
- Captures every AI interaction end-to-end during the assessment
- Shows how candidates plan before prompting
- Reveals where candidates push back when AI does wrong
- Tracks when candidates revert and try different angles
- Reduces reviewer time from 2+ hours to ~30 minutes
- Low setup cost—just paste repo URL and write brief
- Minimal candidate setup required, no dependency hell
- Works with GitHub repos or zip files
- Supports TypeScript, Python, Rust, and Go
- Behavioral attribution shows which prompt led to which decision
- Identifies decision points where candidates paused or reverted
- Full interaction trail per session, not just accuracy number
- Replaces guessing with direct process observation
- AI interactions are explicitly recorded with candidate consent
- Async work with configurable timing
Gonfire cons
- No pricing published yet—early access only
- Custom rubrics not yet available (on roadmap)
- Limited to small set of design partners currently
- Only supports TypeScript, Python, Rust, Go (not other languages)
- Admin-provisioning required for accounts (no self-signup)
- Does not position against HackerRank or CodeSignal
- No AI-written vs human-written line labeling
- Does not replace culture/team-fit interviews
- Early access stage with limited features
- Requires candidates to use AI (may not fit all roles)
- Only works with repos/zips you specify
- No mention of mobile/interview mode options
- No integration API mentioned for ATS systems
- No public success stories or case studies yet
- indicates empty testimonials
Frequently asked questions about Gonfire
What source code can candidates work with?
The starter source can be any GitHub repo or zip file. Gonfire has been tested with TypeScript, Python, Rust, and Go .
Is AI interaction capture covert or disclosed to candidates?
No, it is not covert. The candidate setup page explicitly states that their AI interactions will be recorded and evaluated as part of the assessment. There is no covert capture .
Can I use custom evaluation rubrics?
Not yet. Custom rubrics are on the roadmap .
How do I get an account?
Reach out and Gonfire will walk you through one. They are admin-provisioning accounts for design partners .
What is the reviewer time commitment with Gonfire?
Reviewer time is approximately 45 minutes total—15-30 minutes reviewing the assessment plus 30 minutes for an optional debrief anchored to specific moments, compared to 2+ hours with traditional take-homes .
What languages does Gonfire support?
Gonfire has been tested with TypeScript, Python, Rust, and Go .
Does Gonfire try to label lines as human-written vs AI-written?
No, Gonfire does not try to label individual lines as human-written vs AI-written. The candidate uses AI for the whole thing—that is the point. The label that matters is behavioral: which prompt led to which decision, observed in order .
What is the setup process for hiring teams?
Setup is low cost—just paste a repo URL and write a brief. This is compared to high setup cost for traditional take-homes which requires environment, prompt, rubric, and evaluation criteria setup .
Does Gonfire replace the entire interview loop?
No, Gonfire is not trying to replace the entire interview loop. Culture and team-fit interviews still belong on the calendar. Gonfire replaces the part where you stare at a take-home PR and try to guess what the candidate actually did .