Skip to content

n8n: engines, idempotency, retention, incognito and preview options for the products release — hold until production - #3

Merged
uday390 merged 1 commit into
mainfrom
feat/products-release
Sep 18, 2026
Merged

uday390 merged 1 commit into
mainfrom
feat/products-release

Conversation

@buildwithuday221

Copy link
Copy Markdown
Contributor

Hold until production

Do not merge or tag until the products release (engines extract / deep-extract, credits, plan gates) is live in production. A tag push publishes to npm via publish.yml, and the new options would 400/402 against today's API.

What changed

Source of truth: deep-read-service/docs/api/reference.md on staging (Choosing how to process, Value Grounding, File types, Rate Limits, Webhooks, Error Responses).

Node (nodes/DeepRead/DeepRead.node.ts) — OCR Extract and Structured Extraction only:

  • Engine selector: Plan Default (field omitted; Free and Standard run extract, Enterprise runs deep-extract), Extract (one OCR pass), Deep Extract (two passes, an LLM judge, a second read that checks each extracted field value). Sent as the pipeline form field. Descriptions note that every job reports its product and that fast/standard still work as aliases.
  • Options collection: Idempotency Key (same key on a retry returns the same job; a different request with the same key is 409), Incognito (Enterprise; no preview, no field locations), Preview (default on; off means no page images, no preview link, no field locations), Retention (Days) (1–365, Enterprise).
  • Booleans travel as the strings "true"/"false" in the multipart form; each option is only sent when the user added it, so a plan without the add-on is never gated by a default.
  • Searchable PDF re-described as a Deep Extract add-on on Enterprise plans (the "Standard accuracy tier only" / "default pipeline" wording is gone). searchablePdf and timeout kept as they were.
  • Response shape unchanged.

README — engines and options section, accepted file types by plan, plans (Free 2,000 pages a month resetting on the signup day; Standard prepaid credits from $10 per 1,000 pages: Parse $10, Extract $20, Deep Extract $40; Enterprise custom — "Pro $99 for 50,000" dropped), and a short note that webhooks (Standard and up) are signed with X-DeepRead-Signature (HMAC-SHA256; secret at GET /dashboard/v1/webhooks/secret). The intro no longer names OCR vendors; it describes the two engines as the reference does.

Version 0.2.2 → 0.3.0 (package.json, package-lock.json) plus a CHANGELOG.md entry.

Verification

  • npm run lint and a direct eslint nodes credentials package.json: clean.
  • npm run build: TypeScript build successful; the compiled node carries pipeline, idempotency_key, incognito, preview and retention_days.
  • The repo has no test script (package.json has only lint, build, dev, release), so there were no tests to run.
  • Not published to npm.

🤖 Generated with Claude Code

https://claude.ai/code/session_01FUBR4w2x5qDxjVhrby1uqE

…or the products release

Engine selector (Plan Default / Extract / Deep Extract) sent as the
pipeline form field; an Options collection on OCR Extract and Structured
Extraction with Idempotency Key, Incognito, Preview and Retention (Days),
booleans sent as the strings "true"/"false" the API's form fields expect.
Searchable PDF is described as a Deep Extract add-on on Enterprise plans
and every job reports its product (parse, extract, deep-extract).

README: plans are Free (2,000 pages a month, resetting on the signup
day), Standard (prepaid credits from $10 per 1,000 pages: Parse $10,
Extract $20, Deep Extract $40) and Enterprise (custom); accepted file
types by plan; webhooks (Standard and up) signed with
X-DeepRead-Signature. Version 0.2.2 -> 0.3.0 with a CHANGELOG entry.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FUBR4w2x5qDxjVhrby1uqE
@uday390
uday390 merged commit b7afb24 into main Sep 18, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants