[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7ADARRVTK5R62NCF05P5TWM":3,"skill-same:01M7ADARRVTK5R62NCF05P5TWM":49},{"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":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":32,"ownerHtmlUrl":33,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":34,"install":35,"resources":48},"01M7ADARRVTK5R62NCF05P5TWM","ai-tools",null,"Use when running local, offline-first ML tooling (mechanistic interpretability PoCs like refusal-direction ablation on local safetensors or the HF cache), when adding self-contained model experiments that print JSON, or when post-training, distilling (s1K\u002FR1-style), or RLVR\u002FGRPO-tuning a reasoning or domain model.","Leostruka\u002Fdevin-bundle","Leostruka","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F81370685?v=4","01M7AD8MAK63VDNAWRDRRRW04N","leandro ostruka","01M7AD8MBS65YDS1KMMV7YPM07",0,"MIT","https:\u002F\u002Fgithub.com\u002FLeostruka\u002Fdevin-bundle\u002Fblob\u002Fmain\u002Fskills\u002Fai-tools\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: ai-tools\ndescription: Use when running local, offline-first ML tooling (mechanistic interpretability PoCs like refusal-direction ablation on local safetensors or the HF cache), when adding self-contained model experiments that print JSON, or when post-training, distilling (s1K\u002FR1-style), or RLVR\u002FGRPO-tuning a reasoning or domain model.\ntriggers: [user, model]\n---\n\n# AI Tools & Model Post-Training Playbook\n\nCovers two surfaces:\n1. **Local offline-first ML experiments** in `extensions\u002Fai-tools\u002F` (mechanistic interpretability, weight inspection, deterministic verifiers).\n2. **Reasoning & domain model post-training playbook** (inference-time scaling, SFT distillation, and RLVR with GRPO\u002FDAPO).\n\n## 1. Local extension (`extensions\u002Fai-tools\u002F`)\n\n- Bundle source: `extensions\u002Fai-tools\u002F`\n- Installed: `%APPDATA%\\devin\\extensions\\ai-tools\\` (Windows) or `~\u002F.config\u002Fdevin\u002Fextensions\u002Fai-tools\u002F` (POSIX)\n- Full extension docs: `USAGE.md` inside that directory.\n\n### Conventions (all local tools)\n\n- Self-contained `\u003Ctool>.py` with `__main__` guard and JSON on stdout.\n- `--self-test` runs fully offline asserts; exit 1 on failure.\n- No network: real inputs come from explicit file paths or the local HF cache only.\n- Deps pinned in `requirements.txt`, installed on demand.\n\n### Current tools\n\n- `abliterator.py`: refusal-direction orthogonalization PoC (Arditi et al. 2024): difference-in-means direction, activation ablation, residual-weight orthogonalization. Dummy tensors by default; `--weights F.safetensors` or `--model \u003Ccached>` for real weights.\n- `prompt_compiler.py`: deterministic prompt compilation against `knowledge_bases\u002F`.\n\n### Reading HF model names (decoder)\n\nLong release filenames decode left-to-right as: maker \u002F size \u002F claims \u002F scores \u002F surgical-mod \u002F packaging. Example `Qwen3.8-27B-turbo-fable-coldfusion-735-882-heretic`:\n\n| Segment | Meaning |\n|---|---|\n| `Qwen3.8` \u002F `27B` | vendor + parameter count (weights) |\n| `turbo`, `fable`, `coldfusion` | training claims \u002F distillation source \u002F recipe name |\n| `735`, `882` | benchmark scores stitched into the name |\n| `heretic` | surgical mod - abliteration (refusal direction removed; see `abliterator.py`) |\n| trailing quants (`Q4_K_M`, `gguf`, `awq`) | packaging only - does not change what it knows |\n\nUse this before `hf-model-eval`: scores in the name are marketing claims, not measured facts - verify against the card and pick the quant variant after the eval.\n\n### Notable local-model watchlist\n\n- `nvidia\u002Fpersonaplex-7b` - full-duplex conversational speech-to-speech (real-time voice agent, not TTS). MIT repo; gated weights (NVIDIA Open Model License + HF token), ~16.7GB, GPU required. Different shape from `extensions\u002Fvoicebox` (one-shot TTS); evaluate only when a GPU target exists.\n- bloc `Dia` - speech-to-speech model with no text stage (hears\u002Freasons\u002Fanswers speech; preserves emotion, pauses, sarcasm). Trained from scratch on recorded conversation; requires M4 Max-class Apple silicon. Watch only - no integration until an accessible target + license exists.\n- `Bonside-2` 27B (ternary) - 27B reasoning model, 272k ctx, ternary quantization claimed local-capable. Verify open weights\u002Ftooling + license before any eval.\n\n---\n\n## 2. Reasoning & Domain Model Post-Training Playbook\n\nUse this playbook when building, adapting, or distilling a custom reasoning or domain-specialist model (source architecture: Raschka *Build a Reasoning Model From Scratch*; validated against primary literature below).\n\n### Method selector (evidence-based decision tree)\n\n| Target & Regime | Method | Empirical Basis |\n|---|---|---|\n| Base\u002Finstruct model already solves the task sometimes; weights fixed | **Inference-time scaling (Self-Consistency \u002F Majority Voting)** over $N$ sampled paths (`temp > 0`, `top-p`) + external verifier tie-break | +17.9% GSM8K (`arXiv:2203.11171`). Allocate sample budget per prompt difficulty for 4x efficiency over uniform best-of-N (`arXiv:2408.03314`). |\n| Student model `\u003C7B` parameters (e.g. Qwen3-0.6B\u002F1.7B\u002F4B) | **SFT Reasoning Distillation** from a strong teacher (e.g. DeepSeek-R1) on 1k-10k curated traces + optional budget forcing | Distilling teacher traces into small models beats direct RLVR on the small model by a wide margin (`arXiv:2501.12948`). 1,000 curated examples filtered by difficulty, diversity, and quality suffice (`s1`, `arXiv:2501.19393`). |\n| Base\u002FSFT model `>=7B` on a verifiable domain (math, code tests, SQL, formal rules) | **RLVR with GRPO \u002F DAPO** (group-relative advantages from a deterministic verifier, no critic network) | Eliminates PPO value network (~50% less memory, `arXiv:2402.03300`); elicits emergent self-verification (`arXiv:2501.12948`, `arXiv:2411.15124`). |\n| Full frontier reasoning recipe (`>=7B`) | **Cold-start SFT (thousands of clean traces) -> RLVR (GRPO\u002FDAPO) -> Rejection-sampled SFT -> final RL** | Prevents the readability and language-mixing failures of pure zero-SFT RL (`DeepSeek-R1` vs `R1-Zero`, `arXiv:2501.12948`). |\n\n### Phase route & gates\n\n| Gate | Stage | Exit condition |\n|---|---|---|\n| **G1** | **Deterministic Verifier & Decontaminated Eval** | Answer extractor + normalizer + equivalence checker (math symbolic equivalence, unit test runner, SQL\u002FAST checker) passes unit tests; held-out benchmark (e.g. MATH-500 \u002F domain suite) is decontaminated against train split (`arXiv:2411.15124`) and baseline `pass@1` recorded. |\n| **G2** | **Inference Baseline (KV Cache + Self-Consistency)** | Greedy `pass@1` and `maj@k` (e.g. $k=8, 16$) measured with KV caching + `torch.compile`; latency and token\u002Fanswer length distribution logged. |\n| **G3** | **Distillation (SFT on Teacher Traces)** | Teacher traces filtered by G1 verifier (keep only verified-correct trajectories, balanced by domain diversity and difficulty per `s1K`); cross-entropy loss computed on assistant reasoning+answer tokens only; held-out `pass@1` beats G2 baseline. |\n| **G4** | **RLVR Training Loop (GRPO \u002F DAPO)** | For each prompt $q$, sample group $\\{o_i\\}_{i=1}^G$, score with G1 verifier, compute group-normalized advantage $\\hat{A}_{i} = (r_i - \\mu_R) \u002F (\\sigma_R + \\epsilon)$, update policy with clipped ratio. Checkpoints evaluated on held-out split. |\n\n### Iron rules for model post-training\n\n1. **Verifier before trainer:** never write a GRPO or distillation loop before the deterministic verifier (G1) is unit-tested on edge cases (formatting variations, symbolic equivalence, timeouts).\n2. **Never use intrinsic self-refinement without an external verifier:** asking an LLM to critique and revise its own reasoning without an external verifier or dedicated RL training degrades or stalls accuracy (`arXiv:2310.01798`). Pair refinement loops with a deterministic check or verifier score.\n3. **Distill `\u003C7B`, RLVR `>=7B`:** do not spend GPU hours running pure RLVR from scratch on sub-3B models expecting emergent reasoning; distill from a strong teacher first (`arXiv:2501.12948`, `arXiv:2501.19393`).\n4. **Condition format rewards on correctness:** if granting a reward bonus for `\u003Cthink>...\u003C\u002Fthink>` or `\\boxed{}` tags, gate or scale it by answer correctness; otherwise the policy hacks the format reward with empty or repetitive reasoning.\n5. **Drop or decouple KL penalty for long-CoT RLVR (DAPO rules):** when training long chain-of-thought from a non-reasoning base, the policy distribution must diverge from the reference model; omit the KL penalty (`beta = 0`) or keep it minimal, use token-level policy gradient loss, and skip prompts where all $G$ group samples have identical reward ($\\sigma_R = 0$ yields zero gradient; `arXiv:2503.14476`).\n6. **Track 4 collapse signals every N steps:** (a) policy entropy (sudden drop = mode collapse), (b) mean\u002Fmax response length (length inflation \u002F reward hacking), (c) zero-variance group ratio ($\\sigma_R = 0$), and (d) held-out `pass@1` vs training reward.\n\n### Primary references\n\n- `arXiv:2203.11171`: Wang et al. 2022, *Self-Consistency Improves Chain of Thought Reasoning in Language Models* (ICLR 2023).\n- `arXiv:2310.01798`: Huang et al. 2023, *Large Language Models Cannot Self-Correct Reasoning Yet* (ICLR 2024).\n- `arXiv:2402.03300`: Shao et al. 2024, *DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models* (GRPO).\n- `arXiv:2408.03314`: Snell et al. 2024, *Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters* (ICLR 2025).\n- `arXiv:2411.15124`: Lambert et al. 2024, *Tulu 3: Pushing Frontiers in Open Language Model Post-Training* (RLVR & decontamination).\n- `arXiv:2501.12948`: DeepSeek-AI 2025, *DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning*.\n- `arXiv:2501.19393`: Muennighoff et al. 2025, *s1: Simple test-time scaling* (`s1K` curation & budget forcing).\n- `arXiv:2503.14476`: Yu et al. 2025, *DAPO: An Open-Source LLM Reinforcement Learning System at Scale*.\n\n### External tools worth knowing\n\n- MiniMind (jingyaogong\u002Fminimind): trains a 26M model from scratch in\n  ~2h for ~$0.43; useful for teaching\u002Fdemos before real post-training.\n- gputaskman \u002F GPU Taskman (Scott Hanselman, MIT): GPU task manager\n  showing per-process VRAM\u002FAI usage on NVIDIA\u002FAMD\u002FIntel\u002FApple; quick\n  observability when local inference competes for VRAM.\n- rasbt\u002FLLMs-from-scratch + eriklindernoren\u002FML-From-Scratch: step-by-step\n  PyTorch\u002FNumPy implementations for verifying model mechanics before\n  post-training; see `.devin\u002Fresearch\u002Fresource-index.md`.\n","# AI Tools & Model Post-Training Playbook\n\nCovers two surfaces:\n1. **Local offline-first ML experiments** in `extensions\u002Fai-tools\u002F` (mechanistic interpretability, weight inspection, deterministic verifiers).\n2. **Reasoning & domain model post-training playbook** (inference-time scaling, SFT distillation, and RLVR with GRPO\u002FDAPO).\n\n## 1. Local extension (`extensions\u002Fai-tools\u002F`)\n\n- Bundle source: `extensions\u002Fai-tools\u002F`\n- Installed: `%APPDATA%\\devin\\extensions\\ai-tools\\` (Windows) or `~\u002F.config\u002Fdevin\u002Fextensions\u002Fai-tools\u002F` (POSIX)\n- Full extension docs: `USAGE.md` inside that directory.\n\n### Conventions (all local tools)\n\n- Self-contained `\u003Ctool>.py` with `__main__` guard and JSON on stdout.\n- `--self-test` runs fully offline asserts; exit 1 on failure.\n- No network: real inputs come from explicit file paths or the local HF cache only.\n- Deps pinned in `requirements.txt`, installed on demand.\n\n### Current tools\n\n- `abliterator.py`: refusal-direction orthogonalization PoC (Arditi et al. 2024): difference-in-means direction, activation ablation, residual-weight orthogonalization. Dummy tensors by default; `--weights F.safetensors` or `--model \u003Ccached>` for real weights.\n- `prompt_compiler.py`: deterministic prompt compilation against `knowledge_bases\u002F`.\n\n### Reading HF model names (decoder)\n\nLong release filenames decode left-to-right as: maker \u002F size \u002F claims \u002F scores \u002F surgical-mod \u002F packaging. Example `Qwen3.8-27B-turbo-fable-coldfusion-735-882-heretic`:\n\n| Segment | Meaning |\n|---|---|\n| `Qwen3.8` \u002F `27B` | vendor + parameter count (weights) |\n| `turbo`, `fable`, `coldfusion` | training claims \u002F distillation source \u002F recipe name |\n| `735`, `882` | benchmark scores stitched into the name |\n| `heretic` | surgical mod - abliteration (refusal direction removed; see `abliterator.py`) |\n| trailing quants (`Q4_K_M`, `gguf`, `awq`) | packaging only - does not change what it knows |\n\nUse this before `hf-model-eval`: scores in the name are marketing claims, not measured facts - verify against the card and pick the quant variant after the eval.\n\n### Notable local-model watchlist\n\n- `nvidia\u002Fpersonaplex-7b` - full-duplex conversational speech-to-speech (real-time voice agent, not TTS). MIT repo; gated weights (NVIDIA Open Model License + HF token), ~16.7GB, GPU required. Different shape from `extensions\u002Fvoicebox` (one-shot TTS); evaluate only when a GPU target exists.\n- bloc `Dia` - speech-to-speech model with no text stage (hears\u002Freasons\u002Fanswers speech; preserves emotion, pauses, sarcasm). Trained from scratch on recorded conversation; requires M4 Max-class Apple silicon. Watch only - no integration until an accessible target + license exists.\n- `Bonside-2` 27B (ternary) - 27B reasoning model, 272k ctx, ternary quantization claimed local-capable. Verify open weights\u002Ftooling + license before any eval.\n\n---\n\n## 2. Reasoning & Domain Model Post-Training Playbook\n\nUse this playbook when building, adapting, or distilling a custom reasoning or domain-specialist model (source architecture: Raschka *Build a Reasoning Model From Scratch*; validated against primary literature below).\n\n### Method selector (evidence-based decision tree)\n\n| Target & Regime | Method | Empirical Basis |\n|---|---|---|\n| Base\u002Finstruct model already solves the task sometimes; weights fixed | **Inference-time scaling (Self-Consistency \u002F Majority Voting)** over $N$ sampled paths (`temp > 0`, `top-p`) + external verifier tie-break | +17.9% GSM8K (`arXiv:2203.11171`). Allocate sample budget per prompt difficulty for 4x efficiency over uniform best-of-N (`arXiv:2408.03314`). |\n| Student model `\u003C7B` parameters (e.g. Qwen3-0.6B\u002F1.7B\u002F4B) | **SFT Reasoning Distillation** from a strong teacher (e.g. DeepSeek-R1) on 1k-10k curated traces + optional budget forcing | Distilling teacher traces into small models beats direct RLVR on the small model by a wide margin (`arXiv:2501.12948`). 1,000 curated examples filtered by difficulty, diversity, and quality suffice (`s1`, `arXiv:2501.19393`). |\n| Base\u002FSFT model `>=7B` on a verifiable domain (math, code tests, SQL, formal rules) | **RLVR with GRPO \u002F DAPO** (group-relative advantages from a deterministic verifier, no critic network) | Eliminates PPO value network (~50% less memory, `arXiv:2402.03300`); elicits emergent self-verification (`arXiv:2501.12948`, `arXiv:2411.15124`). |\n| Full frontier reasoning recipe (`>=7B`) | **Cold-start SFT (thousands of clean traces) -> RLVR (GRPO\u002FDAPO) -> Rejection-sampled SFT -> final RL** | Prevents the readability and language-mixing failures of pure zero-SFT RL (`DeepSeek-R1` vs `R1-Zero`, `arXiv:2501.12948`). |\n\n### Phase route & gates\n\n| Gate | Stage | Exit condition |\n|---|---|---|\n| **G1** | **Deterministic Verifier & Decontaminated Eval** | Answer extractor + normalizer + equivalence checker (math symbolic equivalence, unit test runner, SQL\u002FAST checker) passes unit tests; held-out benchmark (e.g. MATH-500 \u002F domain suite) is decontaminated against train split (`arXiv:2411.15124`) and baseline `pass@1` recorded. |\n| **G2** | **Inference Baseline (KV Cache + Self-Consistency)** | Greedy `pass@1` and `maj@k` (e.g. $k=8, 16$) measured with KV caching + `torch.compile`; latency and token\u002Fanswer length distribution logged. |\n| **G3** | **Distillation (SFT on Teacher Traces)** | Teacher traces filtered by G1 verifier (keep only verified-correct trajectories, balanced by domain diversity and difficulty per `s1K`); cross-entropy loss computed on assistant reasoning+answer tokens only; held-out `pass@1` beats G2 baseline. |\n| **G4** | **RLVR Training Loop (GRPO \u002F DAPO)** | For each prompt $q$, sample group $\\{o_i\\}_{i=1}^G$, score with G1 verifier, compute group-normalized advantage $\\hat{A}_{i} = (r_i - \\mu_R) \u002F (\\sigma_R + \\epsilon)$, update policy with clipped ratio. Checkpoints evaluated on held-out split. |\n\n### Iron rules for model post-training\n\n1. **Verifier before trainer:** never write a GRPO or distillation loop before the deterministic verifier (G1) is unit-tested on edge cases (formatting variations, symbolic equivalence, timeouts).\n2. **Never use intrinsic self-refinement without an external verifier:** asking an LLM to critique and revise its own reasoning without an external verifier or dedicated RL training degrades or stalls accuracy (`arXiv:2310.01798`). Pair refinement loops with a deterministic check or verifier score.\n3. **Distill `\u003C7B`, RLVR `>=7B`:** do not spend GPU hours running pure RLVR from scratch on sub-3B models expecting emergent reasoning; distill from a strong teacher first (`arXiv:2501.12948`, `arXiv:2501.19393`).\n4. **Condition format rewards on correctness:** if granting a reward bonus for `\u003Cthink>...\u003C\u002Fthink>` or `\\boxed{}` tags, gate or scale it by answer correctness; otherwise the policy hacks the format reward with empty or repetitive reasoning.\n5. **Drop or decouple KL penalty for long-CoT RLVR (DAPO rules):** when training long chain-of-thought from a non-reasoning base, the policy distribution must diverge from the reference model; omit the KL penalty (`beta = 0`) or keep it minimal, use token-level policy gradient loss, and skip prompts where all $G$ group samples have identical reward ($\\sigma_R = 0$ yields zero gradient; `arXiv:2503.14476`).\n6. **Track 4 collapse signals every N steps:** (a) policy entropy (sudden drop = mode collapse), (b) mean\u002Fmax response length (length inflation \u002F reward hacking), (c) zero-variance group ratio ($\\sigma_R = 0$), and (d) held-out `pass@1` vs training reward.\n\n### Primary references\n\n- `arXiv:2203.11171`: Wang et al. 2022, *Self-Consistency Improves Chain of Thought Reasoning in Language Models* (ICLR 2023).\n- `arXiv:2310.01798`: Huang et al. 2023, *Large Language Models Cannot Self-Correct Reasoning Yet* (ICLR 2024).\n- `arXiv:2402.03300`: Shao et al. 2024, *DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models* (GRPO).\n- `arXiv:2408.03314`: Snell et al. 2024, *Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters* (ICLR 2025).\n- `arXiv:2411.15124`: Lambert et al. 2024, *Tulu 3: Pushing Frontiers in Open Language Model Post-Training* (RLVR & decontamination).\n- `arXiv:2501.12948`: DeepSeek-AI 2025, *DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning*.\n- `arXiv:2501.19393`: Muennighoff et al. 2025, *s1: Simple test-time scaling* (`s1K` curation & budget forcing).\n- `arXiv:2503.14476`: Yu et al. 2025, *DAPO: An Open-Source LLM Reinforcement Learning System at Scale*.\n\n### External tools worth knowing\n\n- MiniMind (jingyaogong\u002Fminimind): trains a 26M model from scratch in\n  ~2h for ~$0.43; useful for teaching\u002Fdemos before real post-training.\n- gputaskman \u002F GPU Taskman (Scott Hanselman, MIT): GPU task manager\n  showing per-process VRAM\u002FAI usage on NVIDIA\u002FAMD\u002FIntel\u002FApple; quick\n  observability when local inference competes for VRAM.\n- rasbt\u002FLLMs-from-scratch + eriklindernoren\u002FML-From-Scratch: step-by-step\n  PyTorch\u002FNumPy implementations for verifying model mechanics before\n  post-training; see `.devin\u002Fresearch\u002Fresource-index.md`.\n",{"name":5,"triggers":24,"description":7},[25,26],"user","model","main","02b95ddb62959a00b3c31442c2046aca0d19563c","skills\u002Fai-tools",1,"https:\u002F\u002Fgithub.com\u002FLeostruka\u002Fdevin-bundle","https:\u002F\u002Fleostruka.github.io\u002Fdevin-bundle\u002F","https:\u002F\u002Fgithub.com\u002FLeostruka","2026-11-15 05:19:58",{"source":8,"skillName":5,"commands":36,"prompt":46,"zipAvailable":47},[37,40,43],{"tool":38,"command":39},"npx","npx skills add Leostruka\u002Fdevin-bundle --skill ai-tools",{"tool":41,"command":42},"bunx","bunx skills add Leostruka\u002Fdevin-bundle --skill ai-tools",{"tool":44,"command":45},"pnpm","pnpm dlx skills add Leostruka\u002Fdevin-bundle --skill ai-tools","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FLeostruka\u002Fdevin-bundle\u002Fblob\u002Fmain\u002Fskills\u002Fai-tools\u002FSKILL.md",true,[],[50,57,64,71,78,85],{"skillNo":51,"name":52,"nameZh":6,"description":53,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":54,"categories":55,"keywords":56,"publishedAt":19},"01M7E0K3MG2XC4FSRV0MK6SFR1","montador-de-ideias","Use when assembling, expanding, or elevating a raw project idea or initial prompt into an ultra-detailed, first-principles dossier and compiled master prompt with journalistic OSINT market research, Tech Lead and Staff Architect justifications, atomic and modular hardening\u002Foptimization, and high-impact contact surfaces.","https:\u002F\u002Fgithub.com\u002FLeostruka\u002Fdevin-bundle\u002Fblob\u002Fmain\u002Fskills\u002Fmontador-de-ideias\u002FSKILL.md",[],[],{"skillNo":58,"name":59,"nameZh":6,"description":60,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":19},"01M7E0J7JJWZ8QMK01RZ3B3C9Z","git-github-adaptive-engineering","Use when any task touches git history, GitHub issues\u002FPRs\u002Freleases\u002FCI, repo governance, or branch and merge decisions. Runs the Fase Zero situational-awareness battery, picks one of five delivery flows via the adaptive decision engine, and enforces commit\u002FPR information-density standards. Orchestrates the gh, git-workflows, github-actions, pr, and debug\u002Ffinish skills.","https:\u002F\u002Fgithub.com\u002FLeostruka\u002Fdevin-bundle\u002Fblob\u002Fmain\u002Fskills\u002Fgit-github-adaptive-engineering\u002FSKILL.md",[],[],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":19},"01M7E0GJFRPFEVQQMXEM58EWDM","adaptive-agenda-calendar-routine-engine","Use when planning or executing work across hours, days, weeks, or sprints. Turns the calendar into a strict time-blocked agenda (HH:MM-HH:MM blocks carrying a routine profile, expected inputs, and a handoff output artifact), runs the 5-stage cognitive flow (Radar and Hunt -> Triage plus Steve Jobs conception -> workbench manifest -> >=32-cycle loop -> Close plus Open-Minded Scrum), and unfolds macro calendar milestones into micro agenda blocks through extensions\u002Fgcal\u002Fwrapper.py.","https:\u002F\u002Fgithub.com\u002FLeostruka\u002Fdevin-bundle\u002Fblob\u002Fmain\u002Fskills\u002Fadaptive-agenda-calendar-routine-engine\u002FSKILL.md",[],[],{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":19},"01M7AE2K9FTY0BVS48N0A9JFKF","demo","Use when demonstrating a skill in a fixture project.","https:\u002F\u002Fgithub.com\u002FLeostruka\u002Fdevin-bundle\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fdevin-config\u002Fproject\u002F.devin\u002Fskills\u002Fdemo\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":19},"01M7AE2G3DP38PTV3B1N7MW5TS","duplicate-skill","Use when testing duplicate skill name detection.","https:\u002F\u002Fgithub.com\u002FLeostruka\u002Fdevin-bundle\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fdevin-config\u002Fproject\u002F.devin\u002Fskills\u002Fbeta\u002FSKILL.md",[],[],{"skillNo":86,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M7AE2DBR8ZHT6T875MJ392GK","https:\u002F\u002Fgithub.com\u002FLeostruka\u002Fdevin-bundle\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fdevin-config\u002Fproject\u002F.devin\u002Fskills\u002Falpha\u002FSKILL.md",[],[]]