promptall

PromptAll for n8n: ask several AI models at once, or check an answer's claims, with one PromptAll key.

Package Information

Released: this month
Downloads: 0 weekly / 0 monthly
Latest Version: 0.1.0-beta.1
Author: iAi Systems LLC

Available Nodes

Documentation generating…

PromptAll icon
PromptAll
Ask several AI models at once, or check an answer's claims, with one PromptAll key

Documentation

@promptall/n8n-nodes-promptall

The PromptAll node for n8n. It can ask several AI models at once, or check an answer's claims, using one PromptAll key (pa_sk_…, created in the app under 🔑 Keys).

It is a thin client of https://api.promptall.app/v1. Model lists come from /v1/models, and limits, prices and routing are the server's. It holds no provider keys and sends no telemetry (see ../README.md).

Operations

Operation In Out
Ask prompt, one or more models (from /v1/models), optional system prompt and max answer tokens one item per model: model_requested, model (what actually answered), route, lane, downshift, answer, cost_usd, usage
Check text, up to 12,000 characters { id, summary, claims: [{ text, verdict, reason }] }, or one item per claim
  • cost_usd is null when the route doesn't report a cost. It is never estimated.
  • downshift is daily or monthly when a ⭐ Pro or ⭐ Pro Max allowance was used up and a promptall/… model answered on free models instead.
  • Errors (401, 402, 403, 413, 429 and others) show the server's own message, plus a line on what to do. With Continue On Fail, a refused model becomes an error item and the other models still answer.

Develop

No build step and no dependencies: plain CommonJS, which n8n loads as is.

node --test test/lib.test.js test/node.test.js

To try it in a local n8n, point N8N_CUSTOM_EXTENSIONS at this folder and start n8n.

Files

  • lib/promptall.js: the API client. It has no n8n imports and is where all behaviour lives.
  • nodes/PromptAll/PromptAll.node.js: the n8n glue (parameters, execute, model picker).
  • credentials/PromptAllApi.credentials.js: the API key and base URL. The connection test calls /v1/me.
  • docs/help-automation-tools.md: the help-centre article "Use PromptAll from n8n, Zapier or Make".

Install

In n8n: Settings → Community nodes → Install, and enter @promptall/n8n-nodes-promptall.

Publishing (owner)

MIT, published as @promptall/n8n-nodes-promptall from the promptall npm org, by trusted publishing from GitHub Actions (.github/workflows/publish-n8n-node.yml): no npm token is stored in GitHub.

Once, the first time (npm can only attach a trusted publisher to a package that exists):

npm login                                   # the promptall org owner, with 2FA
npm publish dist/n8n/promptall-n8n-nodes-promptall-0.1.0-beta.1.tgz --tag beta --access public

Then on npmjs.com: the package → Settings → Trusted Publisher → GitHub Actions: organization iaisystemsllc, repository promptall-app, workflow publish-n8n-node.yml, environment npm. Then Settings → Publishing access → Require two-factor authentication and disallow tokens.

Every release after that: bump version in package.json (and the CHANGELOG), merge, then either push a tag n8n-node-v<version> or run Actions → Publish n8n node with a dist-tag (beta for prereleases, latest for releases). The workflow runs the tests, refuses a version that's already on npm or a prerelease aimed at latest, and publishes.

Installing in n8n: Settings → Community Nodes → Install → @promptall/n8n-nodes-promptall@beta (or @0.1.0-beta.1).

Discussion