Add dynamic browser runtime capture proof annotations - #114
Conversation
- Replace mermaid diagrams with ASCII flows that render on both npm and GitHub - Restore Choose Your Starting Workflow decision table for first-time installers - Restore --tools install examples so CLI scoping stays discoverable - Add GSD scale datapoint (81 commands / 78 workflows / 33 agents) to back the "narrower" claim - Tighten Tessl row and add comparison-as-of-date footnote - Fix Handoff label in lifecycle diagram (no gsdd-handoff workflow exists)
- Replace abstract tagline with what the tool actually does - Cut synthetic sections (What This Is / How It Works as separate H2s) - Restore Choose Your Starting Workflow decision table - Rewrite comparison table rows in plain language - Remove ceremony from runtime support paragraph - Drop Configuration and Troubleshooting sections to docs
docs: clarify Workspine front door
Phase 64 of v1.9 (Closure UX And Regression Sweep): finalize the closure UX work so closeout-report exposes blockers, warnings, and fixes; control-map carries the matching fix hints through its transition-risk output; and tests cover the fail-closed paths. - closeout-report: surface fix hints and propagate next safe action alongside blockers and warnings (38 lines, 16 tests) - control-map: expand transition-risk coverage and fix-hint attachment so consumers see actionable repair guidance (113 lines, 4 tests) - health: align fix-hint wording with closeout consumer - docs: README.md, distilled/README.md, docs/USER-GUIDE.md pick up the new closeout-report capability line - .gitignore: ignore stray .worktrees/ scratch dir from earlier worktree experiments - .gitattributes: enforce eol=lf across the repo to end the Windows CRLF phantom-diff problem that was hiding the real Phase 64 surface area behind 150+ ghost-modified files
feat: close v1.9 phase 64 — regression sweep with LF normalization
One knob deciding how much the assistant does on its own vs. shows you and asks. Four levels (low/medium/high/max), each adding two workflow flags (showCode, askBeforeDecide). Legacy quick/balanced/thorough alias silently so existing configs keep working; medium == the old balanced default, so default behavior is unchanged. Config carries rigorProfile + optional rigorOverrides{plan,execute,verify}; resolveStepRigor/effectiveRigorLevel resolve per step. Adds 'gsdd rigor [show | <level> | <step> <level>]'; init wizard lists the four levels and the summary prints them. All new behavior ships OFF by default (gates read === true, so a missing key means off). Tests: profile shape, legacy aliases, per-step resolution, CLI mutations, and updated init expectations.
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request adds benchmarkable browser-runtime capture annotations to Workspine's UI proof contract, allowing execution and verification to record browser evidence, provider paths, costs, and budgets without requiring browser infrastructure. It introduces optional runtime_capture_requirements for planned slots and runtime_capture metadata for observed bundles, alongside comprehensive validation, comparison, and verification logic. The documentation, agent roles, and tests have been updated to support these changes. One medium-severity logic issue was found in bin/lib/ui-proof.mjs where the aggregation of screenshot_count can produce incorrect totals when mixing explicit and implicit counts or when explicitly set to zero.
Important
The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.
| if (metric === 'screenshot_count') { | ||
| const explicitCount = captures.reduce((sum, capture) => sum + (typeof capture.screenshot_count === 'number' ? capture.screenshot_count : 0), 0); | ||
| return explicitCount > 0 ? explicitCount : captures.filter((capture) => capture.mode === 'screenshot').length; | ||
| } |
There was a problem hiding this comment.
There is a logic issue in how screenshot_count is aggregated when there is a mix of explicit and implicit screenshot counts, or when screenshot_count is explicitly set to 0.
Specifically:
- If a capture has
mode: 'screenshot'andscreenshot_count: 0explicitly,explicitCountwill be0. SinceexplicitCount > 0is false, the function falls back to counting the number of screenshot captures, returning1instead of0. - If one capture has no explicit count (implicit
1) and another hasscreenshot_count: 2,explicitCountis2. SinceexplicitCount > 0is true, the function returns2, completely ignoring the implicit screenshot from the first capture.
We can resolve this by cleanly accumulating the count for each capture: using screenshot_count if it is a number, otherwise defaulting to 1 if the mode is 'screenshot'.
| if (metric === 'screenshot_count') { | |
| const explicitCount = captures.reduce((sum, capture) => sum + (typeof capture.screenshot_count === 'number' ? capture.screenshot_count : 0), 0); | |
| return explicitCount > 0 ? explicitCount : captures.filter((capture) => capture.mode === 'screenshot').length; | |
| } | |
| if (metric === 'screenshot_count') { | |
| return captures.reduce((sum, capture) => { | |
| if (typeof capture.screenshot_count === 'number') { | |
| return sum + capture.screenshot_count; | |
| } | |
| return sum + (capture.mode === 'screenshot' ? 1 : 0); | |
| }, 0); | |
| } |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 417bee7c96
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".
| const explicitCount = captures.reduce((sum, capture) => sum + (typeof capture.screenshot_count === 'number' ? capture.screenshot_count : 0), 0); | ||
| return explicitCount > 0 ? explicitCount : captures.filter((capture) => capture.mode === 'screenshot').length; |
There was a problem hiding this comment.
When a slot has screenshot_count_max and the observed captures mix explicit and implicit screenshot counts, this returns only the sum of explicit screenshot_count values and ignores additional mode: "screenshot" captures without that field. For example, one screenshot capture with screenshot_count: 1 plus two screenshot captures that omit the field is counted as 1 instead of 3, so an over-budget proof can still compare as satisfied.
Useful? React with 👍 / 👎.
1 participant
Summary
Public-safety notes
Verification