Repository navigation
✨ Add agent disclosure and screenshot comment layers - #398
Merged
Merged
Conversation
Disclose known agent assistance on review writes without transmitting harness identifiers. Support precise screenshot feedback and share credential selection, scope checks, and attribution with raw API commands.
This comment has been minimized.
This comment has been minimized.
Keep disclosure on for recognized agent harnesses. Explicit settings can name the agent, and invalid settings stop the review write.
This comment has been minimized.
This comment has been minimized.
Add --layer current|baseline|diff and --point x,y to vizzly comment. Both require --screenshot and are rejected with --reply-to, since replies inherit their thread's layer and location. Remove --type; the server still defaults it to general.
Robdel12
changed the title
Vizzly - Visual Test ResultsCLI Reporter - Waiting for buildNo builds received yet for this pull request. CLI TUI - 1 change needs review
Meaningful diffs needing review (1)vizzly-help · 1202×1430 · 2.8% diff
|
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
Comments and review decisions should disclose agent assistance while keeping the logged-in user as the author. An agent reviewing a screenshot can also leave feedback on the exact spot it inspected, on the image it was actually looking at.
Approach
The CLI recognizes Codex, Claude Code, Pi, Gemini CLI, and OpenCode at the shared HTTP write boundary. It sends only the agent label and keeps session IDs, thread IDs, and prompts local. Use
--via-agentorVIZZLY_VIA_AGENTto name an agent explicitly. Ambiguous markers and invalid overrides stop the write.vizzly commentsupports screenshot IDs, threaded replies, and a location on a specific layer:--layer current|baseline|diffpicks the image. It defaults tocurrenton the server.--region x,y,width,heightis measured in that layer's bitmap pixels. Current and diff share the current image's pixels, while baseline uses the baseline image, which can be a different height.--point x,ypins a spot as integer percentages (0–100). It can't be combined with--region.--screenshotand are rejected with--reply-to, since replies inherit their thread's layer and location.layerfor screenshot comments.--typeis removed. The product no longer shows comment types, and Approve/Reject/Question read as review decisions when they weren't. The server still defaultstypetogeneraland accepts it from older clients. Scripts that pass-t/--typewill now fail with an unknown option, so this needs a release note.Raw API comments and decisions use the same disclosure path. Review commands use the shared credential and scope checks, so explicit tokens take precedence and the CLI reports which credential it selected.
Evidence
The CLI review tests cover all five harnesses, explicit agent selection, rejected opt-outs, credential precedence, read-only scopes, and the comment request bodies for layered regions, points, and replies, along with every location validation. The built CLI completed comment and decision workflows against a real authenticated server and database before the layer change. Layers and points have so far been checked only against the test server.
Release after the server supports agent labels, screenshot comment layers, and points.