We welcome early adopters, testers, and contributors who are comfortable trying experimental developer tooling and sharing feedback.
What is Bitloops?
AI agents don’t understand your codebase.
They brute-force it.
They:
run grep, ls, cat
chain tool calls
pull too much context
burn tokens figuring out what matters
Bitloops replaces that.
It builds a local, queryable model of your repository so agents can fetch what they need directly—dependencies, files, tests, and prior context—in a single call.
Powered by DevQL, a GraphQL-based query layer for your codebase.
It also captures every agent interaction as checkpoints—giving you a trace of:
tool usage
decisions
reasoning
and how changes evolved
So you don’t just get results—you can inspect how they were produced.
Without vs With Bitloops
| | |
| ------------------------------------------------------------ | ------------------------------------------------------------- |
| Without Bitloops | With Bitloops |
| Each session starts from zero | Sessions build on shared repository memory |
| Agents send large amounts of repository context to the model | Only relevant code, architecture, and reasoning are retrieved |
| High token usage with noisy context | Fewer tokens with higher-quality context |
| Reviewers see diffs with no reasoning trail | Commits link to the developer–AI conversation |
| AI reasoning disappears between sessions | Agent reasoning remains searchable |
| Limited governance over AI-generated changes | Teams can review and audit AI reasoning through commits |
| Multiple iterations to reach a correct result | Higher chance of getting it right the first time |
Installation
Native Install (Recommended)
macOS, Linux, WSL:
curl -fsSL https://bitloops.com/install.sh | bash
Windows PowerShell:
irm https://bitloops.com/install.ps1 | iex
Windows CMD:
curl -fsSL https://bitloops.com/install.cmd -o install.cmd && install.cmd && del install.cmd
Getting Started
From the repository you want Bitloops to capture, run:
bitloops init --install-default-daemon
This bootstraps the default daemon service if needed, creates or updates .bitloops.local.toml, and installs Bitloops hooks for the current project.
To pause or resume capture later:
bitloops enable
bitloops disable
bitloops disable removes Bitloops-managed agent prompt surfaces for the agents recorded in the repo policy.
bitloops enable reinstalls those same managed surfaces and resumes capture.
Work as usual and commit normally. Bitloops will capture the relevant
developer-agent context around those changes.
Dashboard
After setup, open localhost:5667 to view checkpoints and DevQL data.
Uninstall
Remove Bitloops-managed agent prompt surfaces from the current Bitloops project:
bitloops disable
Remove Bitloops-managed artefacts from your machine as well:
bitloops uninstall --full
Supported Agents
[x] Claude Code
[x] Codex
[x] Cursor
[x] Gemini
[x] Copilot
[x] OpenCode
What is DevQL?
DevQL is Bitloops’ query layer for asking structured questions about your repository, checkpoints, dependencies, tests, and captured development context.
No. None of your code is sent to our servers. By default Bitloops keeps its
config, cache, state, and local data in platform app directories on your
machine, with store backends controlled by the daemon config.
Is this totally free for real?
You bet!
What kind of databases do I need?
None for normal use. Bitloops ships with local storage defaults, so setup is plug-and-play.
Advanced users can configure external storage backends, but that is optional and not required to get started.
Why do you use telemetry and why should I opt-in?
Telemetry data help us understand which features users are using the most and
help us guide our development. The telemetry data are not connected to specific
users and are analysed and considered in aggregate. Bitloops asks for consent
when the default daemon config is first created, and later interactive init
or enable runs only ask again if consent becomes unresolved for an existing
daemon config.
Community & Support
Contributing
We welcome contributions from the community! Your input helps make Bitloops better
for everyone. See CONTRIBUTING.md to get started.
We welcome:
bug reports
documentation improvements
performance improvements
new integrations with AI coding tools
architectural ideas and discussions
If you are unsure where to start, open an issue and we will help guide you.
Code of Conduct
We're committed to fostering an inclusive and respectful community. Read our
CODE_OF_CONDUCT.md for guidelines.
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
