Skill-Verzeichnis
Jede Empfehlung bleibt mit ihrem Repository, Audit und Installationspfad nachvollziehbar.
给纯文本 LLM agent 装上眼睛：图片问答、OCR、截图分析、视觉定位等一套视觉工具箱 + skill，并可无缝接入 Codex、Claude Code、OpenCode、Pi ｜ Give text-only LLM agents vision: image Q&A, OCR, screenshot understanding, visual grounding, image-to-SVG - a vision toolkit & skill, with drop-in integration for Codex, Claude Code, OpenCode, Pi
A DeepSeek Harness plugin that equips text-only agents with vision capabilities for image Q&A, long-screenshot OCR, UI restoration, grounding, and pixel diff, packaged as an installable skill with a Web UI.
Autonomously improve a real artifact (code, training recipe, agent harness, data pipeline, prompt) against an objective and an evaluator, using Hypothesis Tree Refinement (HTR) from the Arbor paper. Use this whenever someone wants to iteratively optimize something over many experiments without overfitting — e.g. "get my model's eval score up", "improve this agent/harness", "tune this pipeline", "beat the baseline on this benchmark", "run a search over approaches and keep the best", "do an MLE-bench / Kaggle-style optimization", or any long-horizon "make this artifact better and don't just memorize the dev set" task. Trigger it even when the user doesn't say "Arbor" or "hypothesis tree" but describes repeated experiment-and-evaluate loops, branching exploration of competing ideas, or worries about a dev/test gap. Runs Claude itself as the coordinator with subagent executors in isolated git worktrees; for the standalone `arbor` CLI tool see references/arbor-upstream.md.
awesome grounding: A curated list of research papers in visual grounding
[ICLR 2023] Official implementation of the paper "DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection"
Read when the user asks what OpenKnowledge is, wants to install it on a repository, wants to open or preview a single markdown file that is not part of an OpenKnowledge project, wants to share an OpenKnowledge project with collaborators, asks whether OpenKnowledge supports a particular capability, or asks how `ok init` / `ok cowork` / OK Desktop set up a project. Do NOT load to perform OpenKnowledge reads/writes — the runtime guidance for editing markdown inside an initialized OK project ships as a separate project-local skill installed into each detected agent's skills dir (for example `.claude/skills/open-knowledge/`) whenever `ok init` runs.
A Claude Code harness that forces agents to perform discovery and weigh evidence before writing code, ensuring they earn the right to start building.
A Claude-compatible academic research agent skill for Master/PhD students covering literature grounding, novelty gates, math formalization, experiment planning, reviewer simulation, and claim verification.
Vision-grounding plugin for browser-use agents with SoM, Florence-2, Vision–DOM alignment, adaptive visual context, and objective evaluation.
Standard writing workflow in which the main Agent drafts and revises while the built-in general-purpose SubAgent reviews, balancing quality and speed.
Nine quality-canary skills for AI coding agents covering code health, rule completeness, grounding, supply chain, resilience, and drift, with auto-cadence hooks for session start/end.
Lightweight TDD skill suite for AI coding agents enforcing test-first workflow.
Skill-Verzeichnis
Jede Empfehlung bleibt mit ihrem Repository, Audit und Installationspfad nachvollziehbar.
给纯文本 LLM agent 装上眼睛：图片问答、OCR、截图分析、视觉定位等一套视觉工具箱 + skill，并可无缝接入 Codex、Claude Code、OpenCode、Pi ｜ Give text-only LLM agents vision: image Q&A, OCR, screenshot understanding, visual grounding, image-to-SVG - a vision toolkit & skill, with drop-in integration for Codex, Claude Code, OpenCode, Pi
A DeepSeek Harness plugin that equips text-only agents with vision capabilities for image Q&A, long-screenshot OCR, UI restoration, grounding, and pixel diff, packaged as an installable skill with a Web UI.
Autonomously improve a real artifact (code, training recipe, agent harness, data pipeline, prompt) against an objective and an evaluator, using Hypothesis Tree Refinement (HTR) from the Arbor paper. Use this whenever someone wants to iteratively optimize something over many experiments without overfitting — e.g. "get my model's eval score up", "improve this agent/harness", "tune this pipeline", "beat the baseline on this benchmark", "run a search over approaches and keep the best", "do an MLE-bench / Kaggle-style optimization", or any long-horizon "make this artifact better and don't just memorize the dev set" task. Trigger it even when the user doesn't say "Arbor" or "hypothesis tree" but describes repeated experiment-and-evaluate loops, branching exploration of competing ideas, or worries about a dev/test gap. Runs Claude itself as the coordinator with subagent executors in isolated git worktrees; for the standalone `arbor` CLI tool see references/arbor-upstream.md.
awesome grounding: A curated list of research papers in visual grounding
[ICLR 2023] Official implementation of the paper "DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection"
Read when the user asks what OpenKnowledge is, wants to install it on a repository, wants to open or preview a single markdown file that is not part of an OpenKnowledge project, wants to share an OpenKnowledge project with collaborators, asks whether OpenKnowledge supports a particular capability, or asks how `ok init` / `ok cowork` / OK Desktop set up a project. Do NOT load to perform OpenKnowledge reads/writes — the runtime guidance for editing markdown inside an initialized OK project ships as a separate project-local skill installed into each detected agent's skills dir (for example `.claude/skills/open-knowledge/`) whenever `ok init` runs.
A Claude Code harness that forces agents to perform discovery and weigh evidence before writing code, ensuring they earn the right to start building.
A Claude-compatible academic research agent skill for Master/PhD students covering literature grounding, novelty gates, math formalization, experiment planning, reviewer simulation, and claim verification.
Vision-grounding plugin for browser-use agents with SoM, Florence-2, Vision–DOM alignment, adaptive visual context, and objective evaluation.
Standard writing workflow in which the main Agent drafts and revises while the built-in general-purpose SubAgent reviews, balancing quality and speed.
Nine quality-canary skills for AI coding agents covering code health, rule completeness, grounding, supply chain, resilience, and drift, with auto-cadence hooks for session start/end.
Lightweight TDD skill suite for AI coding agents enforcing test-first workflow.
Skill-Verzeichnis
Jede Empfehlung bleibt mit ihrem Repository, Audit und Installationspfad nachvollziehbar.
给纯文本 LLM agent 装上眼睛：图片问答、OCR、截图分析、视觉定位等一套视觉工具箱 + skill，并可无缝接入 Codex、Claude Code、OpenCode、Pi ｜ Give text-only LLM agents vision: image Q&A, OCR, screenshot understanding, visual grounding, image-to-SVG - a vision toolkit & skill, with drop-in integration for Codex, Claude Code, OpenCode, Pi
A DeepSeek Harness plugin that equips text-only agents with vision capabilities for image Q&A, long-screenshot OCR, UI restoration, grounding, and pixel diff, packaged as an installable skill with a Web UI.
Autonomously improve a real artifact (code, training recipe, agent harness, data pipeline, prompt) against an objective and an evaluator, using Hypothesis Tree Refinement (HTR) from the Arbor paper. Use this whenever someone wants to iteratively optimize something over many experiments without overfitting — e.g. "get my model's eval score up", "improve this agent/harness", "tune this pipeline", "beat the baseline on this benchmark", "run a search over approaches and keep the best", "do an MLE-bench / Kaggle-style optimization", or any long-horizon "make this artifact better and don't just memorize the dev set" task. Trigger it even when the user doesn't say "Arbor" or "hypothesis tree" but describes repeated experiment-and-evaluate loops, branching exploration of competing ideas, or worries about a dev/test gap. Runs Claude itself as the coordinator with subagent executors in isolated git worktrees; for the standalone `arbor` CLI tool see references/arbor-upstream.md.
awesome grounding: A curated list of research papers in visual grounding
[ICLR 2023] Official implementation of the paper "DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection"
Read when the user asks what OpenKnowledge is, wants to install it on a repository, wants to open or preview a single markdown file that is not part of an OpenKnowledge project, wants to share an OpenKnowledge project with collaborators, asks whether OpenKnowledge supports a particular capability, or asks how `ok init` / `ok cowork` / OK Desktop set up a project. Do NOT load to perform OpenKnowledge reads/writes — the runtime guidance for editing markdown inside an initialized OK project ships as a separate project-local skill installed into each detected agent's skills dir (for example `.claude/skills/open-knowledge/`) whenever `ok init` runs.
A Claude Code harness that forces agents to perform discovery and weigh evidence before writing code, ensuring they earn the right to start building.
A Claude-compatible academic research agent skill for Master/PhD students covering literature grounding, novelty gates, math formalization, experiment planning, reviewer simulation, and claim verification.
Vision-grounding plugin for browser-use agents with SoM, Florence-2, Vision–DOM alignment, adaptive visual context, and objective evaluation.
Standard writing workflow in which the main Agent drafts and revises while the built-in general-purpose SubAgent reviews, balancing quality and speed.
Nine quality-canary skills for AI coding agents covering code health, rule completeness, grounding, supply chain, resilience, and drift, with auto-cadence hooks for session start/end.
Lightweight TDD skill suite for AI coding agents enforcing test-first workflow.
Skill-Verzeichnis
Jede Empfehlung bleibt mit ihrem Repository, Audit und Installationspfad nachvollziehbar.
给纯文本 LLM agent 装上眼睛：图片问答、OCR、截图分析、视觉定位等一套视觉工具箱 + skill，并可无缝接入 Codex、Claude Code、OpenCode、Pi ｜ Give text-only LLM agents vision: image Q&A, OCR, screenshot understanding, visual grounding, image-to-SVG - a vision toolkit & skill, with drop-in integration for Codex, Claude Code, OpenCode, Pi
A DeepSeek Harness plugin that equips text-only agents with vision capabilities for image Q&A, long-screenshot OCR, UI restoration, grounding, and pixel diff, packaged as an installable skill with a Web UI.
Autonomously improve a real artifact (code, training recipe, agent harness, data pipeline, prompt) against an objective and an evaluator, using Hypothesis Tree Refinement (HTR) from the Arbor paper. Use this whenever someone wants to iteratively optimize something over many experiments without overfitting — e.g. "get my model's eval score up", "improve this agent/harness", "tune this pipeline", "beat the baseline on this benchmark", "run a search over approaches and keep the best", "do an MLE-bench / Kaggle-style optimization", or any long-horizon "make this artifact better and don't just memorize the dev set" task. Trigger it even when the user doesn't say "Arbor" or "hypothesis tree" but describes repeated experiment-and-evaluate loops, branching exploration of competing ideas, or worries about a dev/test gap. Runs Claude itself as the coordinator with subagent executors in isolated git worktrees; for the standalone `arbor` CLI tool see references/arbor-upstream.md.
awesome grounding: A curated list of research papers in visual grounding
[ICLR 2023] Official implementation of the paper "DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection"
Read when the user asks what OpenKnowledge is, wants to install it on a repository, wants to open or preview a single markdown file that is not part of an OpenKnowledge project, wants to share an OpenKnowledge project with collaborators, asks whether OpenKnowledge supports a particular capability, or asks how `ok init` / `ok cowork` / OK Desktop set up a project. Do NOT load to perform OpenKnowledge reads/writes — the runtime guidance for editing markdown inside an initialized OK project ships as a separate project-local skill installed into each detected agent's skills dir (for example `.claude/skills/open-knowledge/`) whenever `ok init` runs.
A Claude Code harness that forces agents to perform discovery and weigh evidence before writing code, ensuring they earn the right to start building.
A Claude-compatible academic research agent skill for Master/PhD students covering literature grounding, novelty gates, math formalization, experiment planning, reviewer simulation, and claim verification.
Vision-grounding plugin for browser-use agents with SoM, Florence-2, Vision–DOM alignment, adaptive visual context, and objective evaluation.
Standard writing workflow in which the main Agent drafts and revises while the built-in general-purpose SubAgent reviews, balancing quality and speed.
Nine quality-canary skills for AI coding agents covering code health, rule completeness, grounding, supply chain, resilience, and drift, with auto-cadence hooks for session start/end.
Lightweight TDD skill suite for AI coding agents enforcing test-first workflow.