AI & website directory - LinkWord

colleague.skill "You AI guys are traitors to the codebase — you've already killed frontend, now you're coming for backend, QA, ops, infosec, chip design, and eventually yourselves and all of humanity"     Your colleague quit, leaving behind a mountain of unmaintained docs? Your intern left, nothing but an empty desk and a half-finished project? Your mentor graduated, taking all the context and experience with them? Your partner transferred, and the chemistry you built reset to zero overnight? Your predecessor handed over, trying to condense three years into three pages? Turn cold goodbyes into warm Skills — welcome to cyber-immortality! Provide source materials (Feishu messages, DingTalk docs, Slack messages, emails, screenshots) plus your subjective description of the person and get an AI Skill that actually works like them Supported Sources · Install · Usage · Demo · Detailed Install 中文 · Español · Deutsch · 日本語 · Русский · Português --- 🆕 2025.04.07 Update — The community's enthusiasm for dot-skill remixes has been incredible! I've built a community gallery — PRs welcome! Share any skill or meta-skill, and drive traffic directly to your own GitHub repo. No middleman. 👉 titanwings.github.io/colleague-skill-site Now listed: 户晨风.skill · 峰哥亡命天涯.skill · 罗翔.skill and more ⏳ PRs are manually reviewed for now — may be a bit slow, thanks for your patience!/ PRs are manually reviewed for now — may be a bit slow, thanks for your patience! --- Created by @titanwings | Powered by Shanghai AI Lab · AI Safety Center Supported Data Sources This is still a beta version of colleague.skill — more sources coming soon, stay tuned! | Source | Messages | Docs / Wiki | Spreadsheets | Notes | |--------|:--------:|:-----------:|:------------:|-------| | Feishu (auto) | ✅ API | ✅ | ✅ | Just enter a name, fully automatic | | DingTalk (auto) | ⚠️ Browser | ✅ | ✅ | DingTalk API doesn't support message history | | Slack (auto) | ✅ API | — | — | Requires admin to install Bot; free plan limited to 90 days | | WeChat chat history | ✅ SQLite | — | — | Currently unstable, recommend using open-source tools below | | PDF | — | ✅ | — | Manual upload | | Images / Screenshots | ✅ | — | — | Manual upload | | Feishu JSON export | ✅ | ✅ | — | Manual upload | | Email .eml / .mbox | ✅ | — | — | Manual upload | | Markdown | ✅ | ✅ | — | Manual upload | | Paste text directly | ✅ | — | — | Manual input | Recommended WeChat Chat Export Tools These are independent open-source projects — this project does not include their code, but our parsers are compatible with their export formats. WeChat auto-decryption is currently unstable, so we recommend using these open-source tools to export chat history, then paste or import into this project: | Tool | Platform | Description | |------|----------|-------------| | WeChatMsg | Windows | WeChat chat history export, supports multiple formats | | PyWxDump | Windows | WeChat database decryption & export | | 留痕 (Liuhen) | macOS | WeChat chat history export (recommended for Mac users) | Tool recommendations from @therealXiaomanChu. Thanks to all the open-source authors — together for cyber-immortality! --- Install Claude Code Important: Claude Code looks for skills in .claude/skills/ at the git repo root. Make sure you run this in the right place. OpenClaw Dependencies (optional) Feishu/DingTalk/Slack auto-collection requires App credentials. See INSTALL.md for details. --- Usage In Claude Code, type: Follow the prompts: enter an alias, company/level (e.g. ByteDance L2-1 backend engineer), personality tags, then choose a data source. All fields can be skipped — even a description alone can generate a Skill. Once created, invoke the colleague Skill with /{slug}. Commands | Command | Description | |---------|-------------| | /list-colleagues | List all colleague Skills | | /{slug} | Invoke full Skill (Persona + Work) | | /{slug}-work | Work capabilities only | | /{slug}-persona | Persona only | | /colleague-rollback {slug} {version} | Rollback to a previous version | | /delete-colleague {slug} | Delete | --- Demo Input: ByteDance L2-1 backend engineer, INTJ, blame-shifter, ByteDance-style Scenario 1: Code Review Scenario 2: Blame game --- Features Generated Skill Structure Each colleague Skill has two parts that work together: | Part | Content | |------|---------| | Part A — Work Skill | Systems, tech standards, workflows, experience | | Part B — Persona | 5-layer personality: hard rules → identity → expression → decisions → interpersonal | Execution: Receive task → Persona decides attitude → Work Skill executes → Output in their voice Supported Tags Personality: Responsible · Blame-shifter · Perfectionist · Good-enough · Procrastinator · PUA master · Office politician · Managing-up expert · Passive-aggressive · Flip-flopper · Quiet · Read-no-reply … Corporate culture: ByteDance-style · Alibaba-style · Tencent-style · Huawei-style · Baidu-style · Meituan-style · First-principles · OKR-obsessed · Big-corp-pipeline · Startup-mode Levels: ByteDance 2-1~3-3+ · Alibaba P5~P11 · Tencent T1~T4 · Baidu T5~T9 · Meituan P4~P8 · Huawei 13~21 · NetEase · JD · Xiaomi … Evolution - Append files → auto-analyze delta → merge into relevant sections, never overwrite existing conclusions - Conversation correction → say "he wouldn't do that, he should be xxx" → writes to Correction layer, takes effect immediately - Version control → auto-archive on every update, rollback to any previous version --- Project Structure This project follows the AgentSkills open standard. The entire repo is a skill directory: --- Notes - Source material quality = Skill quality: chat logs + long docs > manual description only - Prioritize collecting: long-form writing by them > decision-making replies > casual messages - Feishu auto-collection requires adding the App bot to relevant group chats - This is still a demo version — please file issues if you find bugs! --- 📄 Technical Report Colleague.Skill: Automated AI Skill Generation via Expert Knowledge Distillation We wrote a paper detailing the system design of colleague.skill — the two-part architecture (Work Skill + Persona), multi-source data collection, Skill generation & evolution mechanisms, and evaluation results in real-world scenarios. Check it out if you're interested! --- Star History --- MIT License © titanwings

Core Features MiniMax-M2.7 is a 230-billion parameter text-to-text AI model launched by MiniMax, featuring an innovative self-evolving architecture. It can achieve self-iteration and optimization through the Agent Harness software infrastructure. This model is open-sourced on GitHub, allowing developers to access and use it. Key Advantages - Outstanding Performance: Achieves a 56.22% accuracy rate on the SWE-bench Pro benchmark, approaching the performance of top-tier closed-source models. - Efficient Parameter Utilization: Reaches Tier-1 level performance using only 10 billion active parameters, offering exceptional cost-effectiveness. - Powerful Tool Calling Capability: Specifically optimized Agent and tool calling functions, supporting the execution of complex tasks. - Multi-Domain Application: Excels in various fields including programming, reasoning, and office tasks. Application Scenarios - Code Generation and Debugging: Handles real-world programming issues in GitHub repositories, supporting code completion and bug fixing. - Agent Development: Builds AI agent applications based on tool calling. - Office Automation: Office tasks such as document processing, data analysis, and report generation. - Reasoning and Problem-Solving: Complex logical reasoning and mathematical problem-solving. Technical Characteristics - Model Architecture: A large-scale language model with 230B parameters, utilizing a Mixture of Experts (MoE) design. - Deployment Options: Supports various deployment solutions including NVIDIA NIM, Ollama, and vLLM. - Open-Source Ecosystem: Provides model weights and configurations on platforms like GitHub, Hugging Face, and ModelScope. - API Service: Offers commercial API services through the MiniMax Open Platform at competitive prices.

Phosphene

Github

Phosphene is an open-source video wallpaper engine for macOS Tahoe. It reverse-engineered Apple's video wallpaper technology, allowing you to use custom videos as dynamic wallpapers on your Mac. Supports multiple video formats, fully open-source, perfect for developers interested in macOS graphics internals.

Needle — 26M-Param Tool-Calling Model

Github

Needle is a 26M-parameter open-source model by Cactus Compute, distilled from Gemini for single-shot function/tool calling. Achieves 6000 tok/s prefill on edge devices, targeting AI assistants on phones, watches, and glasses. 775pts on HN, 2400+ GitHub stars.

gepa-viz — Prompt Optimization Visualizer

Github

gepa-viz by Modaic AI is an open-source prompt optimization visualizer that renders GEPA candidate trees as force-directed graphs. Supports embedded, remote, and static modes with real-time accept/reject status, diffs, and Pareto frontier. Install via pip install gepa-viz.