knotless documentation

Guides

How to do things with knotless, a task at a time.

Getting started: install, log in, push, clone, commit without a checkout, show who you are.
Collaborating: write access, protected branches, forks, pulls, reviews, merging, issues, moderation.
Private repositories: create, share and browse them.
Agents: connect an MCP client, locally or over HTTP.
Review agents: a bot account that reviews your pulls, with Claude Code or another model.
Running knotless-serve: the cache, appview, gateway and web pages.
Workflows: run CI with knotless-run.

Actions: everything you can do, and its command, MCP tool, XRPC method and web page.
Commands: knotless, knotless-serve, knotless-run, knotless-triage and knotless-agent, their flags, and the environment.
MCP tools.
HTTP and XRPC API: what knotless-serve answers, route by route.
Workflow files: the YAML, what a step sees, and the secrets file.
knotless-run internals: the state file, caches on disk, containers and volumes.
Performance: round trips and benchmarks.
Lexicons: the record and method schemas.
Deploying: how knotless.dev runs, and monitoring.
Observability: what serve and the runner log and count, and why.

How knotless works, and why. Start with the overview.

Overview: goals, non-goals, the pieces, the PDS’s limits.
Storage: records, keys, push, fetch, compaction.
Naming and identity: renames, redirects, transfers, commit authors.
Access: OAuth, scopes, private repositories.
Gateway: collaborators’ writes, protected branches.
Pulls, issues and moderation: the conversation, merging, forks.
knotless-serve: the cache, appview, moderation service and web pages.
knotless-run: deciding what to run, scheduling, caches, secrets, checks, the merge gate.
The merge queue: batches and bisection.
Integrity and trust: who can do what.
Prior art.