Update to Matt Skills Curated

{
  "description": "Author, refine, evaluate, and package agent skills across their full lifecycle. Use when building a new skill from scratch, improving an existing skill, fixing a skill that triggers unreliably, running skill evals, benchmarking skill performance, or packaging skills for distribution — even if they don't explicitly say \"skill conductor\". Do NOT use for general software coding tasks or using pre-existing skills.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 113
    }
  ],
  "name": "skill-conductor",
  "skill_md_contents": "---\nname: skill-conductor\ndescription: \"Author, refine, evaluate, and package agent skills across their full lifecycle. Use when building a new skill from scratch, improving an existing skill, fixing a skill that triggers unreliably, running skill evals, benchmarking skill performance, or packaging skills for distribution — even if they don't explicitly say \\\"skill conductor\\\". Do NOT use for general software coding tasks or using pre-existing skills.\"\n---\n\n# Skill Conductor\n\nFull lifecycle management for agent skills: **draft → test → review → improve → package**.\n\nOne master discipline to govern skill creation and optimization, rooted in the 10 canonical authoring principles.\n\n---\n\n## Core Invariants & The 10 Canonical Authoring Principles\n\n1. **Pre-flight verification**: Check dependencies and environment before executing workflows.\n2. **No process in descriptions**: Descriptions define triggering boundaries (`Use when...`, `Do NOT use for...`), never workflow recipes.\n3. **Map of Content (MOC)**: `SKILL.md` is a clear architectural map (< 500 lines) pointing to modular references.\n4. **Fresh practitioner empathy**: Explain the rationale behind steps so agents understand context.\n5. **Training Within Industry (TWI)**: Structure critical instructions as `Action`, `Key Point`, and `Why`.\n6. **Blind agent testability**: Ensure instructions are unambiguous when executed without prior conversation history.\n7. **Inline risk checklists**: Place verification checklists directly at high-risk seams.\n8. **One term per concept**: Use consistent domain terminology throughout.\n9. **Zero secrets / environment cleanliness**: Never hardcode credentials, tokens, or absolute user home paths.\n10. **Match form to failure**: Counter specific failure modes with targeted structural constraints and anti-rationalization tables.\n\n---\n\n## Core Lifecycle Modes\n\n| Mode | Trigger / Context | Key Output |\n|---|---|---|\n| **1. CREATE** | \"build a skill\", \"new skill for...\" | Full lifecycle: intent → architecture → scaffold → write → eval |\n| **2. IMPROVE** | \"fix this skill\", \"it doesn't trigger\" | Diagnose → eval loop → gated self-update → iterate |\n| **3. VALIDATE** | \"test this skill\", \"run evals\" | Structural checks + trigger testing + BinEval scoring |\n| **4. REVIEW** | \"review this skill\", quality audit | 11-point quality gate assessment |\n| **5. OPTIMIZE** | \"improve triggering\", \"optimize description\" | Automated description optimization with train/test splits |\n| **6. PACKAGE** | \"package for distribution\" | Validation + bundle into release artifact |\n\n---\n\n## Step-by-Step Procedure (TWI)\n\n### Step 1: Capture Intent & Define Triggers\n- **Action**: Extract 2–3 concrete user scenarios and establish positive and negative trigger boundaries.\n- **Key Point**: Specify exact phrases users say and adjacent domains the skill must reject.\n- **Why**: Clear boundaries prevent undertriggering and false-positive overtriggering.\n\n### Step 2: Architecture & Progressive Disclosure\n- **Action**: Select the architectural pattern (sequential, iterative, context-aware) and structure files.\n- **Key Point**: Keep `SKILL.md` concise (< 500 lines) and push heavy schemas or tables into references.\n- **Why**: Bloated instruction files exhaust attention budgets and degrade execution quality.\n\n### Step 3: Write Frontmatter & Body\n- **Action**: Draft YAML frontmatter with kebab-case name and trigger-rich description, followed by MOC, TWI steps, and guardrails.\n- **Inline Checklist**:\n  - [ ] Frontmatter name matches directory name\n  - [ ] Description is $\\le 1024$ characters and contains no process steps\n  - [ ] Negative triggers specified (`Do NOT use for...`)\n  - [ ] Anti-rationalization table included\n  - [ ] No hardcoded tokens, passwords, or machine-specific paths\n\n### Step 4: Validate with BinEval 5 Dimensions\n- **Action**: Evaluate across **Discovery, Clarity, Structure, Robustness, Completeness**.\n- **Key Point**: Pass every critical gate check before declaring release readiness.\n- **Why**: Multi-dimensional evaluation catches hidden failure modes before distribution.\n\n---\n\n## Anti-Rationalization Guardrails\n\n| Tempting Rationalization | Binding Rule | Engineering Rationale |\n|---|---|---|\n| *\"Putting steps in the description helps the model.\"* | **Forbidden: zero process in description.** | Models follow description steps and skip the comprehensive body instructions. |\n| *\"The skill is small, so negative triggers are unneeded.\"* | **Mandatory negative triggers in description.** | Without negative boundaries, skills trigger falsely on loosely related queries. |\n| *\"One big 1200-line markdown file is easier to manage.\"* | **Progressive disclosure: SKILL.md < 500 lines.** | Overloaded contexts dilute attention and lead to instruction-skipping. |\n| *\"We can skip testing if the markdown looks clean.\"* | **Trigger evaluation on positive and negative test cases.** | Clean prose can still fail in actual agent routing. |\n"
}