01 / Explore Betterloop
Find out what your developers will receive.
An approved configuration has a path from its author to the developer device. Follow that path before you choose a rollout.
02 / Choose a starting problem
Where does your current setup lose consistency?
Developers receive different instructions
Trace the selected revision and the repository context each session receives.
Open this resource →Useful resources arrive without ownership
Decide who selects a resource, who reviews it, and where it can operate.
Open this resource →An installation leaves unanswered questions
Separate files on disk from native trust, active hooks, and supported controls.
Open this resource →A good local fix never reaches the next person
Prepare a reusable lesson with a clear task boundary and a meaningful check.
Open this resource →03 / The working path
Give agent setup an owner and a release process.
- 01
Choose the shared starting point
The organization owner records its requirements and selects reviewed resources.
- 02
Prepare the team revision
The team chooses an organization revision. It adds the context and permitted settings its work requires.
- 03
Enroll the developer device
Developers select their existing coding agents. Native authentication and trust remain part of setup.
- 04
Inspect a session in its repository
Compare the expected revision with the reported setup. Examine a supported decision in context.
- 05
Make a lesson available through review
Define where the lesson applies. Verify it. Obtain independent approval before expanding delivery.
04 / Ownership and review
Let teams adopt shared changes deliberately.
An organization and a team each own one rules configuration. Publishing creates a new version. A team reviews its baseline update before adoption.
05 / From local work to team knowledge
Make useful experience available beyond one session.
A lesson needs more than a completion message. Record its applicability and verification evidence. Review the proposed guidance before other developers receive it.
06 / Client choices
Connect the client. Examine its actual coverage.
Use native Claude Code or Codex, or evaluate the optional Mimir runtime. Each path has specific activation requirements and control boundaries.
Claude Code
Prepare its owned integration and inspect a fresh invocation of the supported hooks.
Open this resource →Codex
Complete native hook trust and diagnose installation, activation, and policy support separately.
Open this resource →Mimir — Alpha
Explore the optional standalone runtime and its separate evaluation path.
Open this resource →07 / Reviewed resource examples
Understand a resource before selecting it.
A skill, reference, and runtime control have different jobs. Read the intended workflow, dependencies, and limits before adding one to the shared configuration.
08 / Data and trust
Review the information each connection sends.
Runtime reports carry limited metadata. Public learning contributions carry reviewed summaries. Raw prompts, source code, command arguments, and credentials stay outside those records.
09 / Reading paths
Continue with the task in front of you.
Connect your first device
Start with the appropriate owner or developer procedure and its readiness checks.
Open this resource →Investigate a setup problem
Use focused guides for configuration differences, resource decisions, coverage, and shared lessons.
Open this resource →Understand the engineering choices
Study how content identity and verification records support reviewed changes.
Open this resource →