Jack Timonen @Jack_Timonen

@arshad_mir We have 3 of them :)

@arshad_mir I'm not a technical co-founder myself but I posted a link to our Slack.

"waiting for a tool that never happened". What kind of tool were you looking for? While agents are great at navigating your local codebase, as they can grep, search, and read files to understand how your application works, the problem is that modern software doesn't stop at the repository boundary. A large part of the system lives in frameworks, libraries, SDKs, and other open-source dependencies. Agents can usually see where your code calls into those dependencies, but they often can't navigate and inspect them in the same way. And even when an agent reads the docs, they only tell it what to call, not how it actually behaves. For that, you need the source. GitHits gives agents access to: 1/ Code examples based on real implementations from repositories, issues, discussions, and pull requests, linked back to the implementation code 2/ Code navigation across packages and repositories: search, grep, file listing, and exact line reads without cloning 3/ Package inspection for dependencies, vulnerabilities, changelogs, and upgrade changes 4/Documentation access across hosted docs and repository-backed docs GitHits does this by building a version-aware index of open-source code on demand, usually in 10-20 seconds for an average repository. And there's a forever free tier available.

@rs545837 Regarding Rust, I'd love to know what you think about this: youtube.com/watch?v=K1hxE7…

every "agent orchestrator" this month is an electron app pretending to be your terminal i run 6 agents at once. i just want to know which one's done, which one's stuck, and see the semantic diff before i merge. so i rebuilt it in rust. it lives inside tmux like a sidebar that doesn’t get in your way and syncs across sessions: - live status for every agent across all worktrees (amp, claude code, codex, opencode, pi, droid) - worktrees collapse into one group with running agents + diff stats - hit l to review the diff in a popup, never leave your terminal - 97% less memory per pane. no electron. no new app. it's called opensessions. would you run your agents in this?

To stop the AI hallucinating code that looks right but doesn't, you could also test giving your agent access to 1) Code examples, based on real implementations from repos, issues, discussions, and pull requests, linked back to the implementation code 2) Code Navigation across packages and repos: search, grep, file listing, and exact line reads without cloning 3) Package Inspection for dependencies, vulnerabilities, changelogs, and upgrade changes @GitHits_com CLI enables all that and there's a forever free tier available.

When context is missing, and a model keeps doing retries harder, you could also test giving your agent access to 1) Code examples, based on real implementations from repos, issues, discussions, and pull requests, linked back to the implementation code 2) Code Navigation across packages and repos: search, grep, file listing, and exact line reads without cloning 3) Package Inspection for dependencies, vulnerabilities, changelogs, and upgrade changes @GitHits_com CLI enables all that and there's a forever free tier available.

Would you be interested in giving your agent access to 1) Code examples, based on real implementations from repos, issues, discussions, and pull requests, linked back to the implementation code 2) Code Navigation across packages and repos: search, grep, file listing, and exact line reads without cloning 3) Package Inspection for dependencies, vulnerabilities, changelogs, and upgrade changes? Asking, because @GitHits_com CLI enables all that and it doesn't cost anything to try.

Context7 is great for docs. However, documentation is just an abstraction layer, whereas the truth is in code. But what if your agent got access to – Code examples, based on real implementations from GitHub repos, issues, discussions, and pull requests, linked back to the implementation code – Code Navigation across packages and repos: search, grep, file listing, and exact line reads without cloning – Package Inspection for dependencies, vulnerabilities, changelogs, and upgrade changes Asking, because @GitHits_com CLI enables all that. Forever free tier available. :)

@walidbelr @thsottiaux What did you do to get it?

@thsottiaux @jlitola

@theo 😁

Amir, a long-time Todoist user here. 👋 Since you're getting slop, would it make sense to test what happens if your agent got access to – Code examples, based on real implementations from repos, issues, discussions, and pull requests, linked back to the implementation code – Code Navigation across packages and repos: search, grep, file listing, and exact line reads without cloning – Package Inspection for dependencies, vulnerabilities, changelogs, and upgrade changes Asking, because @GitHits_com CLI enables all that and it doesn't cost anything to try.

@mohbii Is this limited to men?

@Mat_Sherman @harris @boardyai is good for intros

See how GitHits helps Codex to resolve DuckDB C++ migration. Vanilla Codex failed at the task, while with GitHits Codex used fewer tokens, converged faster, and produced a correct solution. youtube.com/watch?v=5JXSiG…

@ycombinator: build a product users 🩷 Our power users realize the impact of giving agents access to real implementations from open source, because the problem with AI coding agents is that they work with partial information: model training data, partial docs, and the limited context available in your current repo. That’s where agents start looping and hallucinating: - Chasing hallucinated APIs that don't exist - Guessing integration patterns from incomplete docs - Wrestling with missing behavior that changed between versions - Treating package internals like a black box To make the right choices, your agent needs to see inside the open source stack: the packages, SDKs, issues, PRs, and docs scattered across GitHub, package registries, and websites. GitHits brings that context to your agent: 1/ Real implementation examples, so the agent can see how an API, library, or framework is used in working projects, not just in toy snippets or outdated blog posts. Code examples are distilled from repositories, issues, discussions, and pull requests, with links back to the source code. 2/ Dependency source navigation, so the agent can inspect what a package does under the hood, without cloning the repo (find symbols, search files, follow implementation paths, and read exact line ranges). 3/ Package inspection, so the agent can make safer dependency changes by checking metadata, vulnerabilities, changelogs, and upgrade context before editing code. 4/ Documentation access, so the agent can combine official up-to-date docs with real code usage, whether the docs are hosted separately or live inside the repo. npx githits@latest init Thanks for the shoutout, @robinebers !

Robin Ebers · AI for Non-Coders @robinebers

5 days ago

don't know why i keep promoting them for free but this is why i love @GitHits_com so much had a problem in my openusage v2 app asked how to solve it boom, perfect fix without guesses

2 2 10 694 5

@Avantgarrdde @brettcalhounn Here's a recent comment from our user. It makes a difference when your agent has access to real implementations, because whatever you're building, another engineer has likely already solved the same issue.

@hhawk @OpenAI @thsottiaux Ok, thanks for sharing. Anyway, if your agent doesn't know how to implement something, GitHits could be able to help. We just received this feedback.