[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M76Z1QB4V7K1THJBD0V2THK8":3,"skill-same:01M76Z1QB4V7K1THJBD0V2THK8":47},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M76Z1QB4V7K1THJBD0V2THK8","alevel-physics-cie",null,"Generate structured answer templates for CIE A-Level Physics (9702) exam questions. Fine-tuned Qwen3-4B LoRA model: question type, given\u002Frequired, formulae, answer frame, checks. Primary use: local MLX inference (skill\u002Fscripts\u002Finference.py) — loads HF base weights and local adapters; no API key and no web scraping in that path. Optional maintainer-only: scraper (cie.fraft.org) and DeepSeek API for rebuilding training data; see SECURITY.md. MANDATORY orchestrator: plain-text math only (no LaTeX dollar delimiters).","Lord1Egypt\u002Fawesome-skill-forge","Lord1Egypt","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F99746359?v=4","01M7592YDNT0XK9WJ0T2MQVPDN","Mohamed Mounir Ali AL Keshita","01M75GRFNHC8VZNWEB8DKDE433",3,"MIT","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fcommunity\u002Fclawhub\u002Fa\u002Falevel-physics-cie\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: alevel-physics-cie\ndescription: \"Generate structured answer templates for CIE A-Level Physics (9702) exam questions. Fine-tuned Qwen3-4B LoRA model: question type, given\u002Frequired, formulae, answer frame, checks. Primary use: local MLX inference (skill\u002Fscripts\u002Finference.py) — loads HF base weights and local adapters; no API key and no web scraping in that path. Optional maintainer-only: scraper (cie.fraft.org) and DeepSeek API for rebuilding training data; see SECURITY.md. MANDATORY orchestrator: plain-text math only (no LaTeX dollar delimiters).\"\n---\n\n# A-Level Physics CIE (9702) Answer Template Generator\n\nGenerate structured answer templates for Cambridge International A-Level Physics (9702) questions using a fine-tuned Qwen3-4B model with LoRA adapters trained on 1652 real past papers.\n\n## Skill contract (runtime vs optional tooling)\n\nThis section clarifies **what runs for normal skill use** vs **what exists only for dataset rebuild \u002F retraining**, so automated reviewers (e.g. OpenClaw) and humans can align expectations with the code.\n\n| | **Primary path — inference** | **Optional — training \u002F data pipeline** |\n|---|------------------------------|----------------------------------------|\n| **Entrypoints** | `skill\u002Fscripts\u002Finference.py`, `generate_template` \u002F `generate_template_verified` in that module | `scraper\u002F*`, `scripts\u002Fbuild_sft.py`, `scripts\u002Frun_full_pipeline.py`, `scripts\u002Ftrain.sh`, etc. |\n| **Remote APIs** | None for generation | **DeepSeek API** when `--teacher deepseek` or full pipeline teacher mode (`DEEPSEEK_API_KEY`) |\n| **Web \u002F HTTP** | Hugging Face (typical) to **download** base model `Qwen\u002FQwen3-4B-MLX-4bit` on first run; no user question leaves your machine as HTTP payload | **cie.fraft.org** when running the scraper; HF again for training stack as configured |\n| **Secrets** | **No** `DEEPSEEK_API_KEY` required by inference | `DEEPSEEK_API_KEY` only if you regenerate SFT via DeepSeek |\n\nInference does **not** scrape past papers, does **not** call DeepSeek, and does **not** exfiltrate prompts to a third-party LLM API. Maintainer scripts may; they are separate.\n\nFull detail: [`SECURITY.md`](SECURITY.md) in the repository root.\n\n## Mandatory rule for the orchestrator (plain-text math)\n\nWhen you produce **any** final answer, template, or paraphrase for the user—whether you ran `skill\u002Fscripts\u002Finference.py` or answered from general knowledge—you **must**:\n\n1. Write formulae in **plain text** (e.g. `v² = u² + 2as`, `E = hf`, `λ = h\u002Fp`, `P = IV`).\n2. **Never** wrap math in `$...$`, `$$...$$`, `\\(...\\)`, `\\[...\\]`, or similar TeX delimiters. Raw `$$` is unreadable for users in Clawhub\u002FOpenClaw-style clients.\n3. If tool output still contains stray `$` signs, **strip or rewrite** those segments into plain text before showing them to the user.\n\nLocal inference already applies the same rule via its system prompt and post-processing; the orchestrator must follow it **even when not calling the script**.\n\n## Quick Start\n\nRun inference on a physics question:\n\n```bash\npython skill\u002Fscripts\u002Finference.py \"Define specific heat capacity.\"\n```\n\nOr in Python:\n\n```python\nfrom skill.scripts.inference import generate_template\nresult = generate_template(\"Calculate the maximum height reached by a ball thrown upward at 20 m\u002Fs.\")\nprint(result)\n```\n\n## Output Format\n\nThe model produces structured answer templates:\n\n- **Question type** — calculation \u002F definition \u002F explain \u002F describe \u002F derive \u002F analyse \u002F practical\n- **Given** — quantities and conditions from the question\n- **Required** — what the student must find or state\n- **Formulae \u002F principles** — relevant equations and physics laws\n- **Answer frame** — numbered step-by-step approach\n- **Check** — unit\u002Fsign\u002Fdirection\u002Fsignificant-figure verification\n\n**Display note (Clawhub \u002F chat clients — applies to orchestrator and model):** Present equations in **plain text** (ASCII and Unicode, e.g. `v²`, `λ`, `×`, fractions with `\u002F`). Do not use LaTeX delimiters (`$`, `$$`, `\\(…\\)`, `\\[…\\]`) in final user-facing output — many clients do not render math, so those tokens look garbled. The inference script enforces this with a system prompt and post-processing when you run it; **if you answer without the script, you must still follow this rule.**\n\n## Model Details\n\n- **Base model**: `Qwen\u002FQwen3-4B-MLX-4bit`\n- **Adapter**: LoRA rank 8, 16 layers, trained 1000 iterations\n- **Training data**: 414 question–template pairs from 9702 Papers 2\u002F4\u002F5 (2001–2025), templates generated by DeepSeek with mark-scheme context\n- **Peak memory**: 4 GB (runs on any 8GB+ Apple Silicon Mac)\n\n## Retraining\n\nTo retrain or extend with more data:\n\n```bash\npython scripts\u002Frun_full_pipeline.py --teacher deepseek\n```\n\nSee `skill\u002Freferences\u002Ftraining.md` for the full pipeline details.\n\n## Adversarial Robustness Evaluation\n\nTest the model's robustness using three physics-adapted attack strategies from Xie et al. (2024):\n\n```bash\npython skill\u002Fscripts\u002Fadversarial_eval.py\npython skill\u002Fscripts\u002Fadversarial_eval.py --strategies numeric --variants 5 --max-questions 10\n```\n\nReports OA (Original Accuracy), AA (Adversarial Accuracy), and ASR (Attack Success Rate) per strategy.\n\n## References\n\n- `skill\u002Freferences\u002Ftraining.md` — Full scraping, extraction, SFT, and training pipeline\n- `skill\u002Freferences\u002Fanswer_template_format.md` — Detailed output format specification\n- `skill\u002Fscripts\u002Finference.py` — Standalone inference script\n- `skill\u002Fscripts\u002Fadversarial_eval.py` — Adversarial robustness evaluation (numeric perturbation, context swap, question-type adversarial)\n- `SECURITY.md` — Network, secrets, and trust boundaries\n","# A-Level Physics CIE (9702) Answer Template Generator\n\nGenerate structured answer templates for Cambridge International A-Level Physics (9702) questions using a fine-tuned Qwen3-4B model with LoRA adapters trained on 1652 real past papers.\n\n## Skill contract (runtime vs optional tooling)\n\nThis section clarifies **what runs for normal skill use** vs **what exists only for dataset rebuild \u002F retraining**, so automated reviewers (e.g. OpenClaw) and humans can align expectations with the code.\n\n| | **Primary path — inference** | **Optional — training \u002F data pipeline** |\n|---|------------------------------|----------------------------------------|\n| **Entrypoints** | `skill\u002Fscripts\u002Finference.py`, `generate_template` \u002F `generate_template_verified` in that module | `scraper\u002F*`, `scripts\u002Fbuild_sft.py`, `scripts\u002Frun_full_pipeline.py`, `scripts\u002Ftrain.sh`, etc. |\n| **Remote APIs** | None for generation | **DeepSeek API** when `--teacher deepseek` or full pipeline teacher mode (`DEEPSEEK_API_KEY`) |\n| **Web \u002F HTTP** | Hugging Face (typical) to **download** base model `Qwen\u002FQwen3-4B-MLX-4bit` on first run; no user question leaves your machine as HTTP payload | **cie.fraft.org** when running the scraper; HF again for training stack as configured |\n| **Secrets** | **No** `DEEPSEEK_API_KEY` required by inference | `DEEPSEEK_API_KEY` only if you regenerate SFT via DeepSeek |\n\nInference does **not** scrape past papers, does **not** call DeepSeek, and does **not** exfiltrate prompts to a third-party LLM API. Maintainer scripts may; they are separate.\n\nFull detail: [`SECURITY.md`](SECURITY.md) in the repository root.\n\n## Mandatory rule for the orchestrator (plain-text math)\n\nWhen you produce **any** final answer, template, or paraphrase for the user—whether you ran `skill\u002Fscripts\u002Finference.py` or answered from general knowledge—you **must**:\n\n1. Write formulae in **plain text** (e.g. `v² = u² + 2as`, `E = hf`, `λ = h\u002Fp`, `P = IV`).\n2. **Never** wrap math in `$...$`, `$$...$$`, `\\(...\\)`, `\\[...\\]`, or similar TeX delimiters. Raw `$$` is unreadable for users in Clawhub\u002FOpenClaw-style clients.\n3. If tool output still contains stray `$` signs, **strip or rewrite** those segments into plain text before showing them to the user.\n\nLocal inference already applies the same rule via its system prompt and post-processing; the orchestrator must follow it **even when not calling the script**.\n\n## Quick Start\n\nRun inference on a physics question:\n\n```bash\npython skill\u002Fscripts\u002Finference.py \"Define specific heat capacity.\"\n```\n\nOr in Python:\n\n```python\nfrom skill.scripts.inference import generate_template\nresult = generate_template(\"Calculate the maximum height reached by a ball thrown upward at 20 m\u002Fs.\")\nprint(result)\n```\n\n## Output Format\n\nThe model produces structured answer templates:\n\n- **Question type** — calculation \u002F definition \u002F explain \u002F describe \u002F derive \u002F analyse \u002F practical\n- **Given** — quantities and conditions from the question\n- **Required** — what the student must find or state\n- **Formulae \u002F principles** — relevant equations and physics laws\n- **Answer frame** — numbered step-by-step approach\n- **Check** — unit\u002Fsign\u002Fdirection\u002Fsignificant-figure verification\n\n**Display note (Clawhub \u002F chat clients — applies to orchestrator and model):** Present equations in **plain text** (ASCII and Unicode, e.g. `v²`, `λ`, `×`, fractions with `\u002F`). Do not use LaTeX delimiters (`$`, `$$`, `\\(…\\)`, `\\[…\\]`) in final user-facing output — many clients do not render math, so those tokens look garbled. The inference script enforces this with a system prompt and post-processing when you run it; **if you answer without the script, you must still follow this rule.**\n\n## Model Details\n\n- **Base model**: `Qwen\u002FQwen3-4B-MLX-4bit`\n- **Adapter**: LoRA rank 8, 16 layers, trained 1000 iterations\n- **Training data**: 414 question–template pairs from 9702 Papers 2\u002F4\u002F5 (2001–2025), templates generated by DeepSeek with mark-scheme context\n- **Peak memory**: 4 GB (runs on any 8GB+ Apple Silicon Mac)\n\n## Retraining\n\nTo retrain or extend with more data:\n\n```bash\npython scripts\u002Frun_full_pipeline.py --teacher deepseek\n```\n\nSee `skill\u002Freferences\u002Ftraining.md` for the full pipeline details.\n\n## Adversarial Robustness Evaluation\n\nTest the model's robustness using three physics-adapted attack strategies from Xie et al. (2024):\n\n```bash\npython skill\u002Fscripts\u002Fadversarial_eval.py\npython skill\u002Fscripts\u002Fadversarial_eval.py --strategies numeric --variants 5 --max-questions 10\n```\n\nReports OA (Original Accuracy), AA (Adversarial Accuracy), and ASR (Attack Success Rate) per strategy.\n\n## References\n\n- `skill\u002Freferences\u002Ftraining.md` — Full scraping, extraction, SFT, and training pipeline\n- `skill\u002Freferences\u002Fanswer_template_format.md` — Detailed output format specification\n- `skill\u002Fscripts\u002Finference.py` — Standalone inference script\n- `skill\u002Fscripts\u002Fadversarial_eval.py` — Adversarial robustness evaluation (numeric perturbation, context swap, question-type adversarial)\n- `SECURITY.md` — Network, secrets, and trust boundaries\n",{"name":5,"description":7},"master","4aa004da461d820f4ea5499e58eb8d196cf99016","community\u002Fclawhub\u002Fa\u002Falevel-physics-cie",0,"https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge","https:\u002F\u002Fskill-explorer-rho.vercel.app","https:\u002F\u002Fgithub.com\u002FLord1Egypt","Unbreakable. Unstoppable. Unmatched⚡.\r\n Speaker of machine tongues 🤖  | World-builder of decentralized dreams 🌐 | Defender of the digital realm🛡️.","2026-06-10 08:21:01",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add Lord1Egypt\u002Fawesome-skill-forge --skill alevel-physics-cie",{"tool":39,"command":40},"bunx","bunx skills add Lord1Egypt\u002Fawesome-skill-forge --skill alevel-physics-cie",{"tool":42,"command":43},"pnpm","pnpm dlx skills add Lord1Egypt\u002Fawesome-skill-forge --skill alevel-physics-cie","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fcommunity\u002Fclawhub\u002Fa\u002Falevel-physics-cie\u002FSKILL.md",true,[],[48,56,63,70,77,84],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":55},"01M785WKJB99G3STR1NF9NRTJW","youtube-content","YouTube transcripts to summaries, threads, blogs.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Fyoutube-content\u002FSKILL.md",[],[],"2026-11-15 14:19:17",{"skillNo":57,"name":58,"nameZh":6,"description":59,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":55},"01M785WEFHB0G2S9YA6EPR3H0E","xurl","X\u002FTwitter via xurl CLI: post, search, DM, media, v2 API.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Fxurl\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":55},"01M785W98EDN4EFJXJPMQEHCRC","touchdesigner-mcp","Control a running TouchDesigner instance via twozero MCP — create operators, set parameters, wire connections, execute Python, build real-time visuals. 36 native tools.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Ftouchdesigner-mcp\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":55},"01M785W5S9FP3QBH2V191E025P","test-driven-development","TDD: enforce RED-GREEN-REFACTOR, tests before code.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Ftest-driven-development\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":55},"01M785W2AS08NSSH9YATCV6FJB","teams-meeting-pipeline","Operate the Teams meeting summary pipeline via Hermes CLI — summarize meetings, inspect pipeline status, replay jobs, manage Microsoft Graph subscriptions.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Fteams-meeting-pipeline\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":55},"01M785VXKW4T34BTP1SCHW1R7F","systematic-debugging","4-phase root cause debugging: understand bugs before fixing.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Fsystematic-debugging\u002FSKILL.md",[],[]]