{"id":50783176,"url":"https://github.com/interesting-vibe-coding/agentshot","last_synced_at":"2026-06-12T05:02:58.880Z","repository":{"id":363036095,"uuid":"1261714111","full_name":"interesting-vibe-coding/agentshot","owner":"interesting-vibe-coding","description":"📸 The screenshot tool built for AI agents — auto-compress to vision's sweet spot, paste anywhere with up to 81% fewer image tokens, no loss in comprehension. Native macOS, zero deps.","archived":false,"fork":false,"pushed_at":"2026-06-07T06:07:26.000Z","size":56,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-06-07T06:23:54.129Z","etag":null,"topics":["ai","ai-agents","claude","clipboard","gpt","llm","macos","menubar","objective-c","productivity","screenshot","token-optimization","vision"],"latest_commit_sha":null,"homepage":"https://interesting-vibe-coding.github.io/agentshot-site/","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/interesting-vibe-coding.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-06-07T03:58:10.000Z","updated_at":"2026-06-07T06:20:00.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/interesting-vibe-coding/agentshot","commit_stats":null,"previous_names":["misterbrookt/agentshot"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/interesting-vibe-coding/agentshot","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/interesting-vibe-coding%2Fagentshot","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/interesting-vibe-coding%2Fagentshot/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/interesting-vibe-coding%2Fagentshot/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/interesting-vibe-coding%2Fagentshot/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/interesting-vibe-coding","download_url":"https://codeload.github.com/interesting-vibe-coding/agentshot/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/interesting-vibe-coding%2Fagentshot/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":34229625,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-12T02:00:06.859Z","response_time":109,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai","ai-agents","claude","clipboard","gpt","llm","macos","menubar","objective-c","productivity","screenshot","token-optimization","vision"],"created_at":"2026-06-12T05:02:57.714Z","updated_at":"2026-06-12T05:02:58.840Z","avatar_url":"https://github.com/interesting-vibe-coding.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# AgentShot\n\nThe screenshot tool built for Codex and GPT vision agents. Snip a region and it's auto-compressed to the vision model's optimal size, then copied to your clipboard.\n\n**Why it matters for Codex:** OpenAI bills vision input by the image's pixel dimensions, tiled. A full-screen Retina grab burns thousands of tokens before the model reads a thing. AgentShot caps the long edge at 1568px and JPEG-encodes it, so you stop paying for pixels the model never uses.\n\n**Measured, not theoretical** — same screenshot, real OpenRouter API calls (`bench/run_openrouter_tokens.py`):\n\n| Model | Original (757 KB, 3024×1964) | AgentShot (174 KB, 1568×1018) | Saved |\n|-------|------|------|------|\n| **GPT-5.5** | 7083 input tokens | 1896 input tokens | **−73%** |\n\n[中文](README.zh.md)\n\n\u003cimg src=\"assets/before-after.png\" width=\"100%\"\u003e\n\n## Install\n\n```bash\ncurl -fsSL https://raw.githubusercontent.com/interesting-vibe-coding/agentshot/main/install.sh | bash\n```\n\nInstalls to `/Applications` and launches. Grant Accessibility once in onboarding. The first capture asks for Screen Recording — allow it, then reopen AgentShot.\n\n## Usage\n\nPress `⌘⇧2` to start. Click once for full screen, or drag to select a region. A preview appears:\n\n- `C` / `↩` — copy the compressed image (label shows `original KB → compressed KB`)\n- `⇧C` — copy the original\n- `Esc` — cancel\n\nPaste into Codex or any model billed by image resolution. Shortcut, quality tier, and launch-at-login live in the menubar (📸).\n\n## How it works\n\nOpenAI bills vision by pixel area (tiled). AgentShot's pipeline: downscale to ≤1568px long edge → JPEG q0.82 → enforce ≤⅓ the original size → hard cap at 1000 KB. Reading accuracy is unaffected — only token count drops ([benchmarks](bench/RESULTS.md)).\n\n\u003e **Note on Claude:** Anthropic normalizes images server-side before billing, so for Claude the token cost is already fixed regardless of what you send — AgentShot saves no tokens there. The win is real on resolution-billed models (GPT/Codex).\n\n---\n\nHomepage \u0026 benchmarks: **https://interesting-vibe-coding.github.io/agentshot-site/** · MIT License · part of [interesting-vibe-coding](https://github.com/interesting-vibe-coding)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Finteresting-vibe-coding%2Fagentshot","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Finteresting-vibe-coding%2Fagentshot","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Finteresting-vibe-coding%2Fagentshot/lists"}