Rebuild Requesty node on Responses API (@n8n/node-cli, verification-ready) - #2
Merged
Conversation
- Response Format option (Text / JSON Object / JSON Schema) for real API-enforced structured output via response_format + strict mode - Use Responses API toggle (useResponsesApi) - Enable Web Search with context size and country, wired as a web_search provider tool
Structured output on the Responses API is expressed via text.format
(json_schema / json_object), not the Chat Completions response_format.
Accepts either a bare JSON schema or a full {name, strict, schema} wrapper.
The Requesty node is now Responses-API-first. The 'Use Responses API' option is removed and useResponsesApi is always true.
Lets self-hosted customers point the node at their own gateway. The override applies to both the model-list request and chat requests, and falls back to the public gateway when empty.
Exposes low/medium/high reasoning effort for reasoning-capable models, wired to the Responses API reasoning.effort parameter.
Drops the country/location option so web search behaves uniformly across the different providers unified behind the Requesty gateway.
Updates the options table and feature list to cover structured output, web search, reasoning effort and the self-hosted Base URL override.
Branch 3 now exercises the Requesty node's Response Format = JSON Schema (real API-enforced structured output) instead of the prompt-based Structured Output Parser.
Previously reasoning effort defaulted to medium and was sent on every request, which non-reasoning models can reject. Adds a 'Default' (unset) option and omits reasoning unless the user picks a level.
The node is always on the Responses API now, so the qualifier is obsolete.
Self-hosted customers can now set their gateway URL on the credential so the connection test validates against their own deployment. The node's Base URL option overrides it per-node, falling back to credential then the public gateway.
Adds jest + ts-jest with 13 tests covering supplyData: always-on Responses API, base URL override, structured output text.format (json object, wrapped + bare schema), web search tool, reasoning effort, and invalid-schema error handling. Tests live in __tests__/ to stay outside the n8n-node lint scope; only dist/ is published.
Enables 'npm ci --legacy-peer-deps' in CI and publish workflows for fast, deterministic installs.
Uses 'npm ci --legacy-peer-deps' against the committed lockfile and adds the test step.
…ckage - Replace prepublishOnly (which blocked npm publish) with a prepack build - Drop the interactive release script in favor of CI publishing - Disable incremental builds so a clean build always emits JS and the 250kB tsbuildinfo cache never ships in the tarball
Triggers on release published (and manual dispatch), installs with the lockfile, lints, builds, tests, then runs 'npm publish --provenance'. Replaces the tag-triggered 'npm run release' which could self-trigger.
Drops the old node (nodes/Requesty), gulpfile, tslint, index.js and stale eslint/prettier/npmignore configs left over from the original package, leaving only the @n8n/node-cli based rebuild.
Selecting Response Format = JSON Schema with an empty/missing schema field crashed with "Cannot use 'in' operator to search for 'schema' in undefined". Now throws a clear error telling the user to provide a schema, and validates the parsed value is an object. Adds tests for the empty and undefined cases.
These were inside the Options collection, where n8n fails to render the editor (no input area) and the conditional show/hide is unreliable — users could select JSON Schema but had nowhere to enter the schema. Moving them to top-level fields fixes both the editor and the displayOptions that reveals the schema only for JSON Schema format.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Clean rebuild of the Requesty n8n community node on the
@n8n/node-clitooling, replacing the original pre-rewrite package.What this is
A Requesty Chat Model sub-node usable anywhere n8n accepts a chat model (AI Agent, Basic LLM Chain, etc.), backed by the OpenAI Responses API via
@n8n/ai-node-sdk.Highlights
text.format(strictjson_schema/json_object) — API-enforced, accepts bare schema or{name, strict, schema}wrapper/modelsQuality / tooling
supplyDatawiringn8n-node lintpasses in strict mode (verification rules)npm publish --provenanceon GitHub Release (OIDC),package-lock.jsoncommittednodes/Requesty, gulpfile, tslint, etc.)Known external dependency
Requires the Requesty gateway to emit
annotations: []onoutput_textcontent parts per the OpenResponses spec (OutputTextContent.required); otherwise the Responses parser throwsCannot read properties of undefined (reading 'map').🤖 Generated with Claude Code