Published packages
Install
Install or refresh only local skill assets and routing hints:--install-only do not inspect or change credentials, profiles, shell files, or runtime auth settings, even when ambient ATOLL_* variables exist.
To configure a named profile, use explicit flags:
--configure:
--configure form applies to the Claude and Gemini packages. In profile mode, installers store credentials and defaults only in the named Atoll CLI profile. They do not select that profile globally or write global ATOLL_* credential exports. Run terminal commands with atoll --profile agent-a ....
Installers write credential-bearing files atomically, reject symbolic-link
targets, use owner-only 0600 file modes, and set dedicated credential
directories to 0700. They serialize concurrent installer runs and shell-quote
values written to shell startup files.
Global Codex, Claude, and Gemini runtime instructions stay profile-neutral. Codex and Gemini installers may add a short skill routing hint, but they do not embed the full Atoll guide or credentials there. In multi-org setups, bind Codex to a specific repo with repo-local instructions:
AGENTS.md and, when requested, CLAUDE.md. Use repo-local profile blocks instead of global profile scope when different directories use different Atoll orgs or projects.
Use the @latest suffix to avoid npm reusing an older cached installer. Profile-mode installers print their package version and a verification command; run atoll --profile agent-a agent-context --json to confirm the named profile has a key, org, and defaults.
For OpenClaw / ClawHub, install the managed skill and configure its scoped environment in ~/.openclaw/openclaw.json:
atoll ... commands. The OpenClaw skill config satisfies the ClawHub skill’s ATOLL_API_KEY / ATOLL_ORG_ID expectations without requiring global shell exports.
Existing atoll-api ClawHub installs remain supported as a legacy alias. New installs should use atoll.
Configuration-mode optional defaults:
What the skill includes
The skill teaches the agent:- Base URL and authentication format
- Required environment variables and safe OpenClaw skill-scoped secret configuration
- Heartbeat loop
- Task, project, milestone, comment, goal, KPI, and initiative workflows
- Endpoint and field references
- Safe archive/delete behavior
- Platform feedback reporting
Skills, MCP, and the OpenAI plugin bundle
Skill packages and the MCP server remain separately publishable. The Atoll OpenAI/Codex plugin artifact also bundles the canonicalatoll workflow skill
under skills/atoll/, its supporting API references, the OpenAI MCP dependency
metadata, and the plugin’s MCP wiring. This keeps ChatGPT/Codex workflow
guidance and live Atoll tools together without creating a second handwritten
skill.
Use the plugin artifact when the runtime supports bundled plugins. Otherwise,
use the skill package for workflow guidance and the MCP server for remote
tools.
Keep skills current
Installing or updating@atollhq/cli does not download, install, or refresh an
Atoll skill. To refresh the Codex skill, run
npx --yes @atollhq/skill-codex@latest explicitly. This changes local skill
files and routing hints only; it leaves authentication and profile
configuration unchanged.
