Skip to content

[E13–S9 10/15] Add web document processing and rendition views - #234

Open
salmonumbrella wants to merge 1 commit into
kenn-io:mainfrom
salmonumbrella:contrib/s4-web-processing
Open

[E13–S9 10/15] Add web document processing and rendition views#234
salmonumbrella wants to merge 1 commit into
kenn-io:mainfrom
salmonumbrella:contrib/s4-web-processing

Conversation

@salmonumbrella

Copy link
Copy Markdown
Contributor

What changed

  • Adds a web processing drawer for one exact selected document version, with executable profile selection, reviewed provider flow, consent state, retained-artifact consequences, and durable job progress.
  • Shows rendition and embedding coverage independently, including missing, rebuilding, unavailable, partial, and complete states, without turning an optional provider failure into a false all-or-nothing result.
  • Adds a verified sanitized Markdown view with immutable build, source, and profile identities; completeness and warnings; body-relative navigation; and inert rendering.
  • Adds source-fenced retrieval proof for the selected version, including the actual search mode, degradations, result evidence, and bounded trace.
  • Keeps browser authority narrow: processing survives a closed response after durable enqueue, rendition reads are not cached and are integrity-checked, and the daemon-lifetime session gains only the exact processing and read routes.

Why

Docbank already has the daemon API and CLI, but the primary human interface could not review what leaves the vault, grant or inspect consent, follow a durable processing job, read the verified rendition, or see why retrieval degraded. This puts those boundaries in one version-pinned workflow without giving the browser provider credentials or general daemon authority.

Usage

Select a file in the web app and choose Process and retrieve. Pick an executable profile, review the exact flow and retained classes, then choose Consent and run when consent is required. After processing, use Read sanitized Markdown to inspect the verified rendition or search the selected version directly from the drawer.

Reviewed processing plan

Reviewed processing plan

Partial provider result

Partial provider result

Verified sanitized Markdown

Verified sanitized Markdown

All captures are from the real daemon and web app against a temporary synthetic vault.

Refs #176

Stacks on #233.

@roborev-ci

roborev-ci Bot commented Aug 28, 2026

Copy link
Copy Markdown

roborev: Combined Review (312f8a3)

Verdict: Two medium-severity security issues require attention.

Medium

  • document/openaiembed/client.go:232-235: The configured egress policy is validated and fingerprinted but not enforced. New preserves the caller’s transport, allowing document/query content and bearer credentials to bypass CIDR, proxy, or transport restrictions. Install a sealed providerhttp transport or reject unsealed clients.

  • document/bridge/client.go:107-109: The rendition bridge only disables cookies and redirects while retaining the caller’s ambient transport, permitting HTTP operator-network origins. Source documents and bearer credentials could be exposed through a malicious proxy or network observer. Require a bounded egress policy and sealed transport, or reject unbounded clients and non-loopback HTTP origins.


Reviewers: 2 done | Synthesis: codex, 6s | Total: 32m40s

@mariusvniekerk mariusvniekerk changed the title Add web document processing and rendition views [E13–S9 10/15] Add web document processing and rendition views Aug 30, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Development

Successfully merging this pull request may close these issues.

1 participant