菜单
Deploy a temporary n8n test instance (or generate a local docker run command) via the internal "Nathan" bot, from the repo instead of Slack. Use after opening a PR to offer the user a live test instance, or whenever someone asks to spin up / deploy a test instance for a branch.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Builds and maintains configuration-based evaluations on a workflow with the eval-config tool. Use when the user asks to set up, add, view, change, or remove an evaluation, score, grade, or judge a workflow's output, or measure answer quality against a test dataset. This is the only eval form Instance AI handles — it does not touch on-canvas evaluation nodes.
Adds or updates n8n Public API v1 endpoints, OpenAPI specs, and handler wiring. Use when creating public API handlers, registering paths in openapi.yml, or adding OpenAPI tags. Prefer @PublicApiController for new endpoints (API-37+).
Checks if a community pull request is ready for human review. Verifies CLA signature, PR title format, description completeness, test coverage, and cubic-dev-ai issues, then triages to the right Linear team or recommends a close. Use when given a PR number or branch name to review, or when the user says /community-pr-readiness-check, or asks to check if a PR is ready for review.
Classifies automation requests using two decisions: anchor (which primitive owns the top-level control flow — workflow-anchored, agent-anchored, needs-clarification, or out-of-scope) and embeds_other (whether the other primitive appears embedded inside — an agent step inside a workflow, or a workflow invoked as an agent tool). Must be used before deciding the intent of any automation request, including compound requests with multiple independent automations, mid-build extensions to an existing workflow or agent, one-off questions or reports that need external systems you cannot query directly, and requests that need clarification before an anchor can be chosen, before choosing workflow-builder, planning, or an agent-oriented design.
Default path for all single-workflow work: new one-off workflows, existing- workflow edits, verification repairs, and workflow-local data tables. Write or edit a workspace source file, then call build-workflow with filePath. Do not load planning or create-tasks first. Load planning only when multiple coordinated workflows or shared cross-task data tables require a dependency-aware task graph.
Authors a new Instance AI workflow eval case — written locally as JSON, calibrated against a real build, then pushed to the LangTracer suite CI runs — build cases, behaviour/process cases, credential cases, and seeded (mid-conversation) cases — with intent-driven expectations. Use when adding or changing an Instance AI workflow eval, or debugging why one is flaky.
| name | n8n:nathan |
| description | Deploy a temporary n8n test instance (or generate a local docker run command) via the internal "Nathan" bot, from the repo instead of Slack. Use after opening a PR to offer the user a live test instance, or whenever someone asks to spin up / deploy a test instance for a branch. |
| allowed-tools | Bash(pnpm nathan:*), Bash(node scripts/nathan.mjs:*), Bash(git diff:*), Read |
Nathan is n8n's internal bot that deploys throwaway test instances from a branch
or Docker image. It's normally driven from Slack (/nathan ...); this skill runs
the same commands from the repo via pnpm nathan.
No tunnel setup needed. Nathan replies asynchronously through a short-lived
public tunnel that the script opens for you (npx localtunnel). A deploy to a
new test-<name> also polls that instance URL directly, so it still reports
success even if the tunnel drops. (Redeploying a name that's already up skips
the poll — it can't tell the old instance from the new — so it falls back to the
tunnel; prefer a fresh name when you need the reliable report.)
A token in ~/.n8n/dev/nathan-token. If a command reports no token, ask the
user for one — point them at the form
(https://internal.users.n8n.cloud/form/d6d34a2f-4899-4ee8-afc8-f8c41a8a243d),
where they log in with their n8n account and copy the token from the response —
then save it for them (the script's interactive paste prompt needs a real
terminal, so as an agent use the subcommand):
pnpm nathan set-token '<PASTED_TOKEN>'
Use judgement — don't offer for every branch. Offer when the change is something a person would actually want to click through in a running instance:
Skip the offer (a test instance adds nothing) for:
When skipping, don't nag — at most mention once that a test instance is available on request.
After you open a PR for a branch that warrants it, offer a live test instance.
Also re-offer whenever you git push new commits to such a branch — the
deployed instance reflects the pushed code, so a new push means a new build worth
redeploying. Don't just ask a bare yes/no — look at the diff and propose a
sensible profile, then let them confirm or adjust. For example:
"Want a test instance for this? Based on the diff I'd deploy it with instance AI enabled (
--ai) since it touches the AI assistant. Sound good, or a different license?"
Inspect what the PR changes (git diff --stat origin/master...HEAD and the file
paths / feature area), then choose:
| PR touches… | Suggest | Why |
|---|---|---|
AI features — @n8n/nodes-langchain, @n8n/instance-ai, the AI assistant/builder, N8N_AI_*, "askAi"/agent code | --ai | Enables instance AI (and defaults the license to pro2) so the AI features actually run |
License-gated / enterprise features — .ee.ts files or /ee/ dirs, license checks (@n8n_io/license-sdk, hasFeature), SSO/SAML/OIDC/LDAP, RBAC/roles/scopes, projects, variables, external secrets, source control/environments, log streaming, insights, folders | --enterprise | The feature is gated behind a license and won't be testable on community |
| A specific gated feature/quota you want on/off | --license pro2 --featureOverride <featureKey>:<value> | Bakes the override into a generated license (community/enterprise can't be overridden) |
| Anything else — core nodes, generic UI, non-gated bug fixes | (nothing — community default) | No license needed |
If both AI and enterprise apply, combine them: --ai --enterprise. When unsure,
state your best guess and ask. Run pnpm nathan help for the full flag reference.
pnpm nathan deploy <branch-name> test-<short-name> [flags]
Nathan builds the branch image (a few minutes) and the command prints the instance
URL (https://test-<short-name>.stage-app.n8n.cloud, login test@n8n.io /
helloWorld7). Relay that URL to the user.
pnpm nathan help # full option reference
pnpm nathan deploy my-branch test-my-feature # community license
pnpm nathan deploy my-branch test-sso --enterprise # enterprise license
pnpm nathan deploy my-branch test-ai --ai # instance AI (license -> pro2)
pnpm nathan deploy nightly test-nightly # deploy the n8nio/n8n:nightly image
Key flags (after the deploy args): --license community|enterprise|starter|pro1|pro2|trial,
--enterprise, --ai, -e KEY=value (repeatable), --featureOverride key:value
(needs a generated license).
local caveatpnpm nathan local ... generates a runnable docker run bundle, but Nathan
delivers it as Slack file attachments (run-n8n.sh + .env), not to the
terminal — they land in #updates-pnpm-nathan
(https://n8nio.slack.com/archives/C0BGVHZ0SCW). deploy and help return fully
in the terminal.
Do not invent a NATHAN_SLACK_CHANNEL. Leave it unset (it defaults to
#updates-pnpm-nathan); only set it if the user explicitly gives you a channel id.
Queue mode, multi-main, and non-SQLite databases are not supported by Nathan.