What works today
getmyorg is in invite-only beta. This page says what's ready and what isn't yet, so you know what to expect. It's updated with each release.
Ready
- Install in Claude Code and Codex, on macOS, Linux and Windows through
WSL2. The plugin keeps the getmyorg program at the same version as itself,
and
/myorg:updateupdates both to the latest release. - Sign in with an invite, through
/myorg:login. Your sign-in renews itself. - The project map.
/myorg:initmaps your repo and starts the semantic map. Git hooks keep the code map current as you commit, switch branches and pull, and/myorg:refreshupdates it by hand. The semantic map is written by your agent, so it updates when you run/myorg:semantic refresh, which also charts more of the code when there's nothing new to process, until the map is healthy (--chartgets there in one go);/myorg:prioritiesand the HQ Guide tell you when it's due./myorg:describe-mapwrites a one-line description of each part of the map, and/myorg:auditchecks your rules, docs and map for gaps. - Reading the map:
/myorg:briefin the terminal and/myorg:hqin the browser, with your features, the code map, your org and the command guide./myorg:whyexplains where a rule, claim or file's standing comes from. - One map for the whole team. The semantic map is committed with your code, and a teammate's clone, a pull or a branch switch loads it.
- Feature groups you curate. Your agent proposes them;
/myorg:semantic curate declarekeeps your own, and/myorg:codifysets their importance and release stage. - A status bar in Claude Code, through
/myorg:config statusline on: how much of the code is charted (▦ 99%), folder, model and effort (Opus 5.5 (high)), a circle that fills as the context does (its number appears from 75%), and the next thing that needs you. HQ's Guide explains each symbol. - Rules that last between sessions:
/myorg:codify,/myorg:answerand/myorg:override, for one repo or a folder of repos./myorg:anchorkeeps the project's goal./myorg:retrospectivescores a session against your rules, and/myorg:learnturns the corrections you made into lasting rules. - Planning:
/myorg:dispatchroutes a request through your org,/myorg:suggestproposes what to build next, and/myorg:triagesorts your open issues./myorg:chattalks a problem through over several turns before anything is built, then offers to file what you decided as an issue. - Procedures your agent runs end to end:
/myorg:pr-refineworks through a pull request's review comments,/myorg:debugfinds a bug's root cause before fixing it,/myorg:estimatesizes an issue or PR by the agent work it takes, and/myorg:split-workbreaks a big issue into PR-sized changes. - Written standards those procedures follow.
/myorg:pr-refine,/myorg:debug,/myorg:estimate,/myorg:split-work,/myorg:triageand/myorg:suggestload getmyorg's standards for their kind of work, such as org-wide, engineering and product, and end by naming the ones they followed. - Ask another agent:
/myorg:second-opinionasks another coding agent you have installed for a bounded answer or a critical review, and folds its answer into yours. In Claude Code,/myorg:codexasks Codex by name. Each needs that agent's CLI, signed in. - Health checks:
/myorg:doctorfinds problems with the install and repairs what it can. - Guardrails inside your sessions: getmyorg checks your agent's commands,
file reads and edits. Risky commands are stopped or bounced back to the
agent, and everyday work isn't interrupted. The band-aid detector reads what
each edit adds and tells the agent when it hides a problem instead of fixing
it: swallowed errors, file-wide lint disables, a bare
@ts-ignore, skipped tests. At strict it stops the worst of them; off skips it. The level is strict, guarded, relaxed (the default) or off. In Codex, trust the plugin's hooks in/hooksfirst. - Your settings, in one place:
/myorg:configshows your guard level, your autonomy profile, the team's org preset and the status line, each with where it came from, and changes them. The guard level and profile are yours, in every repo, or for one repo with--repo; a paid plan starts at theautonomousprofile and the free plan atassisted. The preset follows your plan unless the team sets one for the repo. - Gate mode, for when Claude Code's prompts are off: with
Bash(*)allowed or--dangerously-skip-permissions, nothing asks about a command the guards don't recognise./myorg:config gate onhas the guard ask you, in Claude Code's permission prompt, about every command that isn't routine; the agent can't clear those prompts itself./myorg:configsuggests it when your settings look that way. Codex's hooks can't ask, so it's Claude Code only.
Not yet
- Specialists as
@mentions. getmyorg doesn't add its specialists to your host's list of agents. Reach them through/myorg:dispatch. - Unattended runs. When an agent needs a decision, it asks you in the session and waits. Questions aren't yet sent to Slack or kept for later.
- Gemini CLI and OpenCode. These docs cover Claude Code and Codex only.
/myorg:geminiand/myorg:opencodeask those CLIs from Claude Code, but only how they're called is tested so far, not a run end to end. - Windows without WSL.
Known rough edges
- None we know of right now.
Something here wrong, or missing? Tell us: https://github.com/praxiom-systems/getmyorg-community/issues/new/choose