▲ 164 points
back
64 comments
In my research grep is fine if you don’t care about tokens and you have less than 100k files. The direct corpus interaction paper [1] shows a breakdown past this level. In my personal experience you get a bit better relevance than a BM25 search engine with grep plus an agent. But it requires you to eat tokens.
If you think grep is great, it’s because you’ve been social engineered to organize your content to be findable. We document why something is useful to an agent. We put it in a logical place.
Just organizing content is at least half of building search, agentic or not. It’s one reason Google is successful, we’re all trying to make our content findable by the search engine. It’s not all technology :)
> If you think grep is great, it’s because you’ve been social engineered to organize your content to be findable. ...
This is such a strange train of thought. How do did you get there?
The social engineering thing runs deep. For example, if you grep for “Key” method, chances are the type/class name would stand on the same line. This is the case in Go and, I think, in many other programming languages (ironically, not C).
Lines are a fundamental building block of text and it’s not unreasonable to optimize them.
)
You can minimize the token waste using rtk as a proxy, and Claude will happily use rtk.
Don’t presume this study has anything to do with programming. They measured an agent’s ability to search long conversations, not code.
> We evaluate on a 116-question representative subset of the LongMemEval benchmark (Wu et al., 2025), which tests an agent’s ability to answer questions over long conversations spanning multiple sessions.
I get a sense that I was click-baited by article's title with the classic trope of "X is all you need". This research is a solid contribution, but is far from all we need to understand grep vs semantic search in agent retrieval.
Combining regex filtering with semantic ranking using multi-vector embeddings has yielded good results for me. I use ColGREP from the LightOn team asa daily driver - https://github.com/lightonai/next-plaid/blob/main/colgrep/RE...
I have always used traditional grep to search codebases. It serves me better than an IDE when there’re lots of scattered and frequent queries.
grep’s design is surprisingly winning, exceeding expectations to this day.
you might be interested in https://github.com/boyter/cs
pretty fast and neat project to search code interactively with a lot of optimizations on finding the right thing
Tangential, I have a hook that rewriters grep to rg but lately I wonder if this is actually wasteful as the model is so biased to grep, is there a way to shim/alias perhaps?
My CLI does something close to this:
https://github.com/gitsense/gsc-cli
`gsc grep` is just an alias for `gsc rg`, mostly because agents are much more likely to reach for “grep” than “rg”.
It works pretty well, but it is not a perfect drop-in replacement. `grep` and `ripgrep` differ in a few details, especially around glob/wildcard behaviour and flags. What I found works is to not use `grep` in search examples, and have the CLI spit out an error message for the AI saying this is `ripgrep`, so it needs to use `ripgrep` syntax.
I've been on a look out for any harness that properly secures a protocol to the LLM, but they're all just "here's some tools, hopefully you don't use bash for everything".
Many harnesses are doing this already, "Grep" is the tool name, ripgrep is the implementation
It depends on if it is using Grep the harness tool or Grep from the bash tool
If performance is the concern, ugrep will get you most of the way there relative to gnu grep, and should be fully grep compatible in terms of syntax:
https://github.com/Genivia/ugrep#aliases
Claude Code may ship with ugrep already.
It seems ridiculous that, for example, Copilot running in Visual Studio working on a C# codebase finds stuff in code by grepping around instead of using the Roslyn-driven code symbol and semantic database built into Visual Studio. I'm guessing it's because the people they get to work on AI stuff are AI People who probably only write in Python
It is sort of funny when Copilot hasn’t been integrated with Microsoft’s stuff. But it does make some sense from a business point of view. Make it work with grep, it works everywhere.
There's a lot more examples of grep usage than Visual code search in the training set.
Codex does this in VSCode as well.
Compilations break all the time and those symbols either become useless or it’s just quicker to use grep.
This is a surprising result. With structured inputs like source code, I’d expect grep to outperform semantic search, but natural language’s errors and inconsistencies seem to leave so many cracks for information to fall through.
This paper is based on quality so I don't think it should be that surprising if you take loops into consideration. What the agent finds in the first pass, can help if formulate the next grep if needed.
If you are truly bitter-lesson pilled - give the agent all the tools and let it decide which to use.
- regex (grep) - hybrid search (bm25+vector)
this X vs Y is uninteresting when the answer can be both.
That assumes that the agent knows which one is better. And to bake in which one is better via post-training would require a study like this to establish where each one works well
Exactly this, and this tool called qmd is what I use for the hybrid search portion. It also uses local LLMs to provide summaries on your own markdown data too. My agents use both depending on what type of search they are doing, and both provide good results.
Both is usually the right answer, since you can use LLMs to do query expansion and effectively increase the recall performance of your retrieval algo
it will only use tools it was trained on? what's the benfit of givig it all the tools.
I'm still disappointed that ai can't use ctags, its used for finding strings and patterns, its right there.
I recently watched the new Palantir + Kirkland & Ellis fund formation platform demo, and I was surprised to see how effective the union of structured data was in an agent harness. We're used to dealing with flat files and comparing here basic ways of searching, essentially, long strings, but using Palantir's "Ontology" graph framework, I think Kirkland is going to be able to achieve some exception and differentiating outcomes in legal tech. The whole idea assumes that they've got great structured data already, and perhaps that's the real valuable unknown, but giving an agent those tools is super powerful.
I wrote about it[1] and came away with a different view on both Palantir and the future of agentic workflows personally.
[1] sorry, LinkedIn: https://www.linkedin.com/pulse/fund-managements-killer-app-d...
That was great, thanks for the write-up. It’s rare to get a peek into Palantir’s ontology-forward approach. I’ve certainly been curious.
> But it would make no sense to have an LLM regurgitate an existing form document token-by-token rather than call a piece of 1994 software like Hotdocs to populate some placeholders.
This is a real “oof”, isn’t it. Very difficult to understand what they were going for here. Perhaps they just assumed no one in the intended audience would pick it up. But it certainly is enough of a red flag that it made me go back to the top of your write-up for a re-read, thinking about their whole pipeline in much more sceptical terms.
Table 2 and 3 tell you basically all you need to know. When you use a harness that is tuned towards programing (Codex and Claude Code), grep wins. When you use a neutral harness, vector search wins.
So far every Grep vs RAG discussion I've seen conflates overlapping factors. The most common is simply that a company rebuilt their pipeline from scratch and fixed a bunch of problems. The worst is when they go from one-shot RAG to multi-step Grep and completely miss the fact that multi-step RAG would likely get them similar results.
At the end of the day, the most important thing is knowing the _product features_ your users care about and making sure that's represented in the pipeline.
As far as i know Claude Code also uses LSP and tree-sitter to find things in your source code.
I'm getting a little tired of the "X is all you need" formulation
This paper oversells on the title. Like, what is chronos, which embedding model was used, which reranker, how was the reranking done, why is chronos much better than claude code
> grep generally yields higher accuracy
And a lot more tokens, and slower speed. Yes you can get more accuracy if you suck tons more data into context.
But compare this to more advanced code agent methods like Tree Sitter, PageRank, LSP, that build semantic maps to provide more relevant context. Grep alone can't do that
This has been posted before, but a dead-simple pattern that helps enormously with steering the model to the right code area is a DESIGN.md that it creates, updates, and references periodically.
Is <blank> the only ML paper title?
Feels important, but I wish they also had compared against something like MeiliSearch or Algolia.
i recently switched to https://github.com/dmtrKovalenko/fff. but i haven't notice any big diffrence yet..
From the article:
> LongMemEval rewards recovering literal witnesses: exact dates, counts, preferences, and spans that often remain stable under tokenization.
Is this saying they chose a benchmark that is biased towards doing well against literal string matching, thus works well with grep, and then (gasp) showed that grep did well, finally declaring "grep is all you need"?
The examples in the benchmark's demo image(1) are all examples you could see grep doing well on. A conversation about bikes, then a query about bike(s) where "bike" is a common token hit. But not stuff like a conversation about a Beethoven sonata, then a question about classical music, where embedding based approach would shine.
(1) https://github.com/xiaowu0162/LongMemEval/blob/main/assets/l...
I'm curious to see what patterns it's grepping.
If grep were enough, SQLite wouldn't exist.
Surely 'strings' would be even better?
it’s interesting to do a follow up where giving an agent both to choose from and comparing accuracy.