Skip to content

docs: add eSignet and OpenFn integration pages - #967

Merged
jeremi merged 3 commits into
mainfrom
docs/esignet-openfn-integrations
Sep 9, 2026
Merged

jeremi merged 3 commits into
mainfrom
docs/esignet-openfn-integrations

Conversation

@jeremi

@jeremi jeremi commented Sep 9, 2026 •

Copy link
Copy Markdown
Member

Summary

Two explanation pages, seated under the "Use from applications" groups and cross-linked from the integration patterns page, the glossary, and the terms specification.

  • Authenticate citizens with eSignet over Base Registry Engine (BReg) (explanation/esignet-authentication-over-breg, status: current). Describes registrystack/esignet-relay-authenticator v0.3.0: who owns what between eSignet, BReg, and Mint; the challenge-first login flow with a fresh-consent second lookup; the governed :lookup request shape; what the registry must grant (lookup-only access profile, a required scope only the provider holds, selector profile, separate admin client); claim mapping and the pairwise pseudonymous subject; the fixed error codes; a configuration example; the boundaries (BReg only, Relay deferred, no OTP delivery, enrollment, wallet login, or identity assurance); and the Solmara Lab journey. Seated in the BReg section.
  • Automate registry workflows with OpenFn (explanation/openfn-adaptors, status: draft, adaptors at 0.1.0 and unreleased). Describes registrystack/openfn-language-registry-stack: the three adaptors and the @registrystack/client@0.27.0 namespaces underneath; credential shapes (explicit vs profile mode, static vs private_key_jwt); what each adaptor does; the result branches and the no-retry, explicit-idempotency, strong-ETag rules; the synthetic agriculture pilot (three workflows, split principals, breg-webhook-signature-v1 event bridge, deduplicating destination); what the pilot does not establish; and loading the adaptors into another OpenFn. Seated in the Evidence Gateway, Relay, and BReg sections.

Other changes:

  • astro.config.mjs: four sidebar entries.
  • scripts/information-architecture.test.mjs: the exact slug lists for the Evidence Gateway consumer group and the BReg section now include the new pages.
  • explanation/integration-patterns.mdx: one sentence in "Workflow engines" pointing at the OpenFn page.
  • reference/glossary.mdx and spec/rs-terms.mdx: the eSignet entries now say which way the integration runs (a separately maintained provider is an ordinary governed client of a registry) instead of saying no service integrates with it, and link the new page.

Evidence

Registry Stack-side claims carry {/* Evidence: ... */} anchors (BReg lookup route and $select/accessProfile, authorize_profile_claims and the access.profile.no_required_scope finding, Mint private_key_jwt and the minted scope claim, the access-token verifier profile both runtimes build, the webhook signature domain, the agricultural-holdings starter, the Node client facade, the Evidence client profile loader, the Relay V2 Evidence-family exclusion and V1 retirement list). Claims about the two external repositories carry TODO[evidence] comments per the spec/rs-arc-g.mdx precedent. The OpenFn page names the upstream commit it was checked against; the eSignet page pins to the v0.3.0 release tag.

Review notes

  • Security-sensitive content. Both pages document authentication and authorization integrations (Mint private_key_jwt clients, access-profile grants, pairwise subjects, HMAC event verification). The pages describe behaviour and recommend narrow grants; they change no code. Worth a read by someone who knows the eSignet provider and the OpenFn pilot to confirm the described behaviour matches.
  • All five review findings were verified against source and fixed in 087bff4c3, with the reasoning in each thread. Two mattered for security guidance: a dedicated access profile does not isolate the provider without a requiredScopes entry, and neither BReg nor Evidence Gateway accepts a citizen's eSignet ID Token as a registry credential.
  • Upstream, not fixed here. The OpenFn Relay adaptor still exposes listEvidenceOfferings against /metadata/evidence-offerings, a route the Relay V2 cutover retired; the page warns against building on it, but the operation belongs in the OpenFn repository's own backlog. Solmara Lab's docs/esignet.md still describes the v0.2.0 JAR, the nia-esignet Mint client, and the Relay esignet-userinfo lookup, so it disagrees with this page until it is rewritten for 0.3.0.

Checks

From docs/site/: npm test (565 pass) and npm run check:production, which is what CI runs: the full source check set (content, evidence anchors, draft links, docset, release manifests, archive lock, cutover, notary surface, markdown, Vale with 0 errors and 0 warnings on the touched pages, OpenAPI, config vocabulary, tutorial dry runs, SVG), the production-shaped build, and the built-output checks (accessibility, llms.txt, SEO, 93,386 internal links).

@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Sep 9, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-09-09T15:39:37.400684Z 087bff4 New commits
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@jeremi
jeremi enabled auto-merge (rebase) September 9, 2026 15:04

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7ac053f867

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Adopters asked how Registry Stack fits beside an identity front door and a
workflow engine. Two explanation pages answer that from the Registry Stack
side: the eSignet provider that authenticates citizens through one governed
Base Registry Engine lookup, and the OpenFn adaptors that wrap the Evidence
Gateway, Base Registry Engine, and Registry Relay clients for Lightning
workflows. Both are seated under "Use from applications" and cross-linked
from the integration patterns page and the glossary.

Signed-off-by: Jeremi Joslin <jeremi@joslin.fr>
The eSignet page told operators a dedicated access profile isolates the
provider, but `accessProfile` is caller-selected and an empty
`requiredScopes` set admits any caller that satisfies the principal claim,
so the profile now needs a scope registered on the provider's Mint client
alone. It also claimed BReg and Evidence Gateway accept eSignet ID Tokens
and UserInfo JWTs: both runtimes verify access tokens only, so that
paragraph now says a citizen's ID Token is not a registry credential.

The OpenFn page promised the adaptors keep raw records out of workflow
state while later saying record data stays in `data`, described change
application as always a separate lifecycle step when `automatic` and
`planner` modes apply inline, and listed evidence-offering discovery as a
current Relay capability though Relay V2 claims no Evidence family and the
V1 route is retired.

eSignet 0.3.0 is released, so the page pins its links to the tag and is no
longer a draft, and the terms entry says which way the integration runs.

Signed-off-by: Jeremi Joslin <jeremi@joslin.fr>
@jeremi
jeremi force-pushed the docs/esignet-openfn-integrations branch from 7ac053f to 087bff4 Compare September 9, 2026 15:35
The anchor named the generated products/registry-relay page, which a clean
checkout does not hold, so the Evidence anchors job failed on a tree the
local check passed against a built site.

Signed-off-by: Jeremi Joslin <jeremi@joslin.fr>
@jeremi
jeremi merged commit a8273a4 into main Sep 9, 2026
37 checks passed
@jeremi
jeremi deleted the docs/esignet-openfn-integrations branch September 9, 2026 15:50
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.

1 participant