Conversation
- Add hatchling build-system and tool.uv.package so the `shots` entry point installs correctly - Move shots.yaml to shots.yaml.example and gitignore the real config - Add .env.example for OPENAI_API_KEY - Support non-interactive login via Playwright inspector (page.pause) when stdin is not a TTY - Update README with setup instructions Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add per-module logging with file + stderr output (shots.log) - Separate action_timeout_ms (clicks/typing, fail fast) from nav timeout - Detect and re-query when LLM repeats a previously failed action - Feed failure history into LLM prompt to prevent loops - Tighten stability wait caps for faster runs Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Stop downscaling screenshots to 1000px before sending to the vision model — the squished text caused garbled OCR (wrong emails, wrong URLs). Add accessibility tree + link href extraction so the LLM sees actual clickable elements instead of guessing from pixels. Also add PAGE ELEMENTS rule to system prompt and reduce history window to last 5 steps. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Pass shot description to crop LLM so it knows what content to include. Add 8% padding around crop coordinates to prevent clipping field labels and content at edges. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace the 8% padding hack with a self-correcting crop validation loop:
after cropping, send the result back to the LLM to verify all content is
visible. On failure, re-crop with the rejection reason (max 2 retries).
Restructure output into per-shot folders: {out_dir}/{shot_id}/output.png
with sources/ and previous/ subdirectories. Add out_dir and per-shot
folder override to shots.yaml config.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Support grouping screenshots in config with per-group output mode
(png/pdf). Labels render as white banners below screenshots with
template variables ({url}, {id}, {title}) and URL desensitization.
Backward-compatible: flat shots lists auto-wrap into single-shot groups.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add label_date boolean to group config for appending UTC timestamp. Draw black separator line between screenshot and label banner. Increase padding and line spacing in label for better readability. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
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
1 participant
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.
Summary
shots.log) + stderr outputaction_timeout_ms(clicks/typing, fail fast) from navigation timeoutTest plan
shots runwith--use-llmand verifyshots.logis created with debug output--action-timeout-msflag works and element interactions fail fast🤖 Generated with Claude Code