Cline

The most relevant user comments come from HackerNews, where developers with hands-on AI coding experience express measured skepticism. One user (+178) praises Cline specifically: 'I like cline... it found a good mix between vibesus take the wheel and hey, I'm still here and I enjoy doing this,' noting it works surprisingly well with local models (tested with devstral). Another user (+178) whose company is building a similar tool concluded that 'chunking and vector similarity search are fundamentally not great approaches for code RAG,' effectively validating Cline's divergence from traditional RAG. However, several users push back: one (+178) argues codebase indexing remains 'a killer feature in Cursor' with tens of thousands of reference files, insisting Cline may miss critical context without it. Broader AI coding sentiment is cautious — users compare LLM coding assistants to 'a senior dev who comes to work drunk,' useful only when tightly supervised (+198). Multiple commenters report hallucination loops, incorrect API usage, and the fundamental problem that if you must explain every step, 'it makes the whole idea of AI coding moot' (+198). One team enacted a rule: if you can't explain an LLM-generated PR without another LLM, you can't submit it. Juniors are seen as 'the least likely to benefit' (+198) because they lack the expertise to catch subtle errors.

No aggregate ratings were found for this product during the last harvest.