reoclo init
reoclo init sets up the current project to use Reoclo: it links the directory to an organization and downloads the reoclo agent skills your tools need. Run it once per project after reoclo login.
reoclo login # if you haven't alreadyreoclo initThis will:
- Link the project to an organization by writing a
.reoclofile (see Working across organizations). Everyreoclocommand run anywhere in this directory tree then targets that org — concurrently with your other projects, with no global “active org” to switch. - Install the reoclo skills into
.claude/skills/so skill-aware agents (Claude Code and others) can drive Reoclo from your terminal.
Choosing the organization
Section titled “Choosing the organization”In an interactive terminal with more than one organization, init shows a picker (defaulting to your active org). To set it non-interactively, pass the org explicitly:
reoclo init --org acmeThe org must be one you authorized during reoclo login.
Initializing under a specific profile
Section titled “Initializing under a specific profile”If you run init under a non-active profile — reoclo --profile staging init, or with $REOCLO_PROFILE set — the org is resolved against that profile’s backend, and init pins the profile in .reoclo too:
// .reoclo written by `reoclo --profile staging init`{ "profile": "staging", "org": "platform" }This matters because an org slug is only meaningful relative to its backend — the same slug (e.g. platform) can exist in both staging and production. Pinning the profile makes the binding reproduce the exact org you selected; without it, the slug would re-resolve against your active profile (a different backend) later. A plain init on your active profile writes just { "org": "<slug>" }, so the project floats with whatever profile is active.
Options
Section titled “Options”| Flag | Effect |
|---|---|
--org <slug> | Organization to link (must be in your OAuth grant). Skips the picker. |
--skills <list> | Comma-separated subset of skills to install (default: all). |
--no-skills | Skip installing skills. |
--mcp | Also register the reoclo MCP server in .mcp.json (merges; never clobbers existing servers). |
--force | Overwrite an existing .reoclo without asking. |
-y, --yes | Assume “yes” for prompts (for non-interactive use). |
--org is the global org flag — it works the same here as on any command.
What it writes
Section titled “What it writes”| Path | Contents |
|---|---|
.reoclo | { "org": "<slug>" } — the project’s org binding (plus "profile": "<name>" when initialized under a non-active profile; see above). Commit it so teammates inherit the binding. |
.claude/skills/<skill>/ | Each downloaded skill (one directory with a SKILL.md plus supporting files). |
.mcp.json | (with --mcp only) a reoclo entry under mcpServers running reoclo mcp. |
Skills are downloaded from the public reoclo/skills repository and require tar on your PATH (present by default on macOS, Linux, and Windows 10+). If tar is unavailable, init prints a manual git clone fallback and still completes the org link.
CI / scripted use
Section titled “CI / scripted use”reoclo init --org acme --no-skills --yesinit requires authentication — in CI, set REOCLO_AUTOMATION_KEY or run after reoclo login.
See also
Section titled “See also”- Working across organizations — the
.reoclobinding and org precedence. reoclo mcp— the MCP server--mcpregisters.