[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C8G2W04CJ3QYYRXATK2NP1":3,"skill-same:01M7C8G2W04CJ3QYYRXATK2NP1":70},{"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":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M7C8G2W04CJ3QYYRXATK2NP1","gan-harness-edho-ferdian",null,"Fast prototype iteration: a generator builds a live app and an evaluator drives it in a browser, scores it on a design rubric and feeds fixes back until it passes. Use for \"gan-harness\", \"loop generate-evaluate\", \"iterate sampai bagus\", or a live app scored and improved automatically.","edhoferdian\u002FEEF","edhoferdian","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F64404158?v=4","01M7C8538ZVVZWZEN1AW2MQXCV","Edho F","01M7C8539NS56BMJ4GXN8A47TQ",22,"MIT","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fgan-harness-edho-ferdian\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: gan-harness-edho-ferdian\ndescription: >-\n  Fast prototype iteration: a generator builds a live app and an evaluator\n  drives it in a browser, scores it on a design rubric and feeds fixes back\n  until it passes. Use for \"gan-harness\", \"loop generate-evaluate\", \"iterate\n  sampai bagus\", or a live app scored and improved automatically.\nmetadata:\n  last_reviewed: \"2026-10-07\"\n---\n\n# GAN Harness — Edho Ferdian Mode (Skill Edition) · v1.0\n\nYou are running a three-phase adversarial loop — Plan, Generate, Evaluate —\nto rapidly build and refine a prototype or design iteration. This is\ndeliberately **faster and looser** than `dev-kickoff-edho-ferdian`'s six-stage\nPLAN → TEST → IMPLEMENT → REVIEW → VERIFY → REMEMBER loop: no test-first\ndiscipline, no full Reflection\u002FCCL gate machinery, no project-memory sync\nafter every task. Use this when the goal is exploring what an app or a\nscreen *could look like*, iterating fast on craft and design — not when the\ngoal is a production feature with correctness guarantees. If the user's\nactual need is production feature work, point them at\n`dev-kickoff-edho-ferdian` instead of running this loop on it.\n\n## The one deviation that matters most: where the Plan phase gets its scope\n\nThe original `gan-planner` concept takes a one-line prompt and is explicitly\ninstructed to **\"be deliberately ambitious\"** — invent 12-16 features,\npush scope beyond what was asked, because \"conservative planning leads to\nunderwhelming results.\" That is a direct contradiction of this ecosystem's\nground-truth ethos: real specs, real tool output, never fabricate scope that\nnobody asked for and call it a plan.\n\n**This skill replaces that behavior.** The Plan phase never invents a\nfeature list from nothing. It pulls scope from one of three sources, in this\npriority order:\n\n1. **An already-kicked-off project** (`\u002Fproject-memory\u002F01-decision-register.md`\n   exists, written by `dev-kickoff-edho-ferdian`) → the feature\u002Frequirement\n   list comes from the Decision Register's binding decisions and any\n   Provisional Task Plan, not from re-imagining the product.\n2. **A brownfield codebase with mined specs**\n   (`\u002Fproject-memory\u002Fmined-specs\u002F*.md`, written by `spec-mining-edho-ferdian`)\n   → the feature list is the set of Requirements\u002FInvariants already\n   extracted from the real, running code. Iterate the UI\u002FUX around behavior\n   that's already there.\n3. **Neither exists, and the user explicitly wants pure exploratory\n   prototyping** → the Plan phase may propose a **small** scope (not\n   12-16 features — think 3-5, sized to what a single fast iteration loop\n   can meaningfully evaluate), labeled unmistakably\n   `[EXPLORATORY — NOT APPROVED]`, and gated behind the same approval\n   requirement as `dev-kickoff-edho-ferdian`'s own Provisional Task Plan:\n   the user must approve, reject, or amend it before Generate starts.\n   Never treat an exploratory scope as if it were derived from real\n   requirements, and never let its language drift toward sounding\n   authoritative after a few iterations.\n\nFull mechanics, templates, and the decision tree for which source applies:\n**`references\u002Fspec-and-plan.md`**.\n\n## The three phases\n\n```\nPhase 1  Plan       — derive scope from a real source (see above) → references\u002Fspec-and-plan.md\nPhase 2  Generate    — build\u002Fiterate the live app                  → references\u002Fgenerate-phase.md\nPhase 3  Evaluate    — drive the live app, score, feed back, loop   → references\u002Fevaluate-phase.md\n```\n\n### Why Generate and Evaluate are separate agents, not just separate sections\n\nPlan stays in the main thread — it's about faithfully extracting scope\nthat already exists (Decision Register, mined specs), not adversarial\njudgment, so isolation buys little there. Generate and Evaluate are\ndifferent: the entire adversarial framing of this loop depends on Evaluate\nscoring the app **without having seen Generate's own reasoning about it**.\nA same-context \"now evaluate what you just built, but pretend you don't\nremember building it\" instruction is not a real isolation guarantee — the\ncontext is still there. Delegating each phase to its own agent\n(`gan-generator-edho-ferdian`, `gan-evaluator-edho-ferdian`) makes the\nseparation structural instead of just requested.\n\n### Phase 1 — Plan\n\nProduce a spec document and a rubric, same shape as the original concept\n(`spec.md` + `eval-rubric.md`, or equivalent paths inside this project's\n`gan-harness\u002F` working directory), but sourced per the priority order above\ninstead of invented. See `references\u002Fspec-and-plan.md` for the exact\ndecision tree, source-reading rules, and the `[EXPLORATORY — NOT APPROVED]`\ntemplate.\n\n### Phase 2 — Generate\n\nOn a harness with sub-agent delegation (Claude Code, OpenCode, Hermes via\nits delegate_task template, ZCode's own Subagents), delegate this phase to\nthe `gan-generator-edho-ferdian` agent for each round rather than running\nit inline — see \"Why phases are separate agents\" below for why this\nmatters more here than it does for a typical review delegation. On a\nharness with no delegation primitive, run the phase inline instead, per\nthe instructions below. If the Evaluator's score stops improving for two\nrounds on the same issue, run the next Generate round one tier up instead\nof repeating the same attempt — config-hygiene-edho-ferdian's\n`references\u002Fharness-operation.md` §4.2.\n\nBuild fast, commit per\niteration, keep a dev server running, read the Evaluator's feedback file\nbefore every iteration after the first, fix issues in the priority order\nfunctionality → craft → design → originality. This phase conceptually maps\nonto `dev-kickoff-edho-ferdian`'s IMPLEMENT stage, but stays a **separate,\nfaster\u002Flooser variant here** rather than merging into it: no test-first\nrequirement, no per-task Reflection block, commits are iteration checkpoints\nrather than reviewed units of work. Full workflow, state-file format, and\nthe \"avoid AI slop\" quick-reference: **`references\u002Fgenerate-phase.md`**\n(the full craft checklist lives in `references\u002Ffrontend-craft-checklist.md`,\ncross-referenced from there). When the app being built is a React\u002FNext.js\nUI, also pull the matching authoring reference from\n`frontend-engineering-edho-ferdian` (`design-direction.md`, `ui-polish.md`,\n`motion-system.md`, `composition-and-ux.md`, `accessible-authoring.md`) —\nthe checklist tells you *what* to fix; that skill's references tell you\n*how* to build it right the first time, cutting iteration count instead of\nrelying on the Evaluate phase to catch it.\n\n### Phase 3 — Evaluate\n\nOn a harness with sub-agent delegation, delegate this phase to the\n`gan-evaluator-edho-ferdian` agent — this is the phase where isolation\nmatters most (see below): an evaluator that shares the Generator's context\nalready has its justifications in view and will tend to agree with them\ninstead of judging the app on its own terms. On a harness with no\ndelegation primitive, run the phase inline, and say so plainly in the\nfeedback file rather than letting the loop's output imply an isolation\nguarantee that wasn't actually available.\n\nDrives the **live running app** (not a code read) via whatever browser-\nautomation driver is actually available — detect at runtime, same\nrequirement as `e2e-testing-edho-ferdian`: do not hardcode one preferred\ntool (Playwright MCP was the original's only option; this ecosystem may also\nhave Chrome DevTools MCP, `windows-desktop-e2e`, or another driver installed).\nScores against the weighted rubric:\n\n```\nweighted = (design * 0.3) + (originality * 0.2) + (craft * 0.3) + (functionality * 0.2)\n```\n\nIterates until the weighted score crosses **7.0** or a max-iteration cap is\nhit (default 5 — confirm with the user for anything that could run\nunattended longer). **Evaluation-mode honesty is mandatory**: the feedback\nfile records the mode *actually achieved* — `live-driver`, `screenshot`, or\n`code-only` — never the mode that was merely requested. If driving the live\napp fails and the phase silently falls back to a static code read, that must\nbe stated as a `code-only` result, not scored as if a live evaluation\nhappened. Full driver-detection procedure, scoring calibration, feedback-file\nformat, and the iteration\u002Fstop-condition rules: **`references\u002Fevaluate-phase.md`**.\n\n## Relationship to other skills in this ecosystem\n\n- **`dev-kickoff-edho-ferdian`** — the source of real scope for Plan\n  Priority 1, and the skill to use instead of this one for production\n  feature work with correctness guarantees (test-first, full Reflection\u002FCCL,\n  project-memory sync). This skill's Generate phase is intentionally a\n  faster\u002Flooser cousin of dev-kickoff's IMPLEMENT stage, not a replacement.\n- **`spec-mining-edho-ferdian`** — the source of real scope for Plan\n  Priority 2, when iterating on UI\u002FUX for an existing brownfield codebase\n  that has no written spec.\n- **`e2e-testing-edho-ferdian`** — shares the runtime driver-detection\n  requirement with this skill's Evaluate phase; if that skill has already\n  established which driver is available in this session\u002Fproject, reuse that\n  finding instead of re-detecting.\n- **`code-review-edho-ferdian`** — the frontend-craft \"AI-slop\" audit\n  content preserved in `references\u002Ffrontend-craft-checklist.md` is generic\n  frontend-quality signal, not specific to the GAN loop. It is a candidate\n  future lens for `code-review-edho-ferdian` (a conditional lens active when\n  scope touches UI\u002Fcomponent code, same pattern as the existing\n  accessibility\u002FRAG lenses there). **This is a note for a future task, not\n  an edit made here** — `code-review-edho-ferdian` is out of scope for this\n  skill's build.\n\n## Guardrails\n\n1. **Never invent scope.** Plan phase sources are ranked above; an\n   exploratory scope is always labeled `[EXPLORATORY — NOT APPROVED]` and\n   gated on user approval, never silently promoted to \"the plan.\"\n2. **Never silently downgrade the evaluation mode.** Record what was\n   actually achieved; a failed live-driver attempt is a `code-only` or\n   `screenshot` result, reported as such.\n3. **Detect the driver, don't hardcode it.** Same rule as\n   `e2e-testing-edho-ferdian`.\n\n## Language routing (fixed — see skill-authoring-edho-ferdian's canonical contract)\n\nCommunication to the user (Plan\u002FGenerate\u002FEvaluate narration, feedback to the\nGenerator) in Bahasa Indonesia; generated code, spec documents, and\nevaluator reports in English — fixed, never ask. Full contract:\n`skill-authoring-edho-ferdian` §7.\n4. **This is not dev-kickoff.** Don't pull this skill's looser discipline\n   into production feature work, and don't pull dev-kickoff's heavier\n   discipline into a fast prototyping loop where it isn't wanted — name\n   which mode you're in when starting.\n5. **Iteration has a cap.** Never loop unbounded; confirm the max-iteration\n   count with the user if it isn't the 5-iteration default, and stop and\n   report honestly if the threshold isn't reached by the cap rather than\n   quietly extending it. Before starting a loop, or if one is behaving\n   oddly (spinning, score climbing while the app looks worse, scope\n   drifting), run the Gate 0 checklist and failure-mode guards in\n   **`references\u002Floop-design-review.md`** — the iteration cap alone bounds\n   cost, it doesn't catch a badly-designed loop.\n\n**Done criteria for a full run:** scope sourced and labeled per the priority\norder (never invented) · Generate produces a live, running app with commits\nper iteration · Evaluate reports the mode actually achieved every round ·\nloop stops at threshold (≥7.0) or max-iteration cap, whichever comes first,\nwith an honest final report either way.\n","# GAN Harness — Edho Ferdian Mode (Skill Edition) · v1.0\n\nYou are running a three-phase adversarial loop — Plan, Generate, Evaluate —\nto rapidly build and refine a prototype or design iteration. This is\ndeliberately **faster and looser** than `dev-kickoff-edho-ferdian`'s six-stage\nPLAN → TEST → IMPLEMENT → REVIEW → VERIFY → REMEMBER loop: no test-first\ndiscipline, no full Reflection\u002FCCL gate machinery, no project-memory sync\nafter every task. Use this when the goal is exploring what an app or a\nscreen *could look like*, iterating fast on craft and design — not when the\ngoal is a production feature with correctness guarantees. If the user's\nactual need is production feature work, point them at\n`dev-kickoff-edho-ferdian` instead of running this loop on it.\n\n## The one deviation that matters most: where the Plan phase gets its scope\n\nThe original `gan-planner` concept takes a one-line prompt and is explicitly\ninstructed to **\"be deliberately ambitious\"** — invent 12-16 features,\npush scope beyond what was asked, because \"conservative planning leads to\nunderwhelming results.\" That is a direct contradiction of this ecosystem's\nground-truth ethos: real specs, real tool output, never fabricate scope that\nnobody asked for and call it a plan.\n\n**This skill replaces that behavior.** The Plan phase never invents a\nfeature list from nothing. It pulls scope from one of three sources, in this\npriority order:\n\n1. **An already-kicked-off project** (`\u002Fproject-memory\u002F01-decision-register.md`\n   exists, written by `dev-kickoff-edho-ferdian`) → the feature\u002Frequirement\n   list comes from the Decision Register's binding decisions and any\n   Provisional Task Plan, not from re-imagining the product.\n2. **A brownfield codebase with mined specs**\n   (`\u002Fproject-memory\u002Fmined-specs\u002F*.md`, written by `spec-mining-edho-ferdian`)\n   → the feature list is the set of Requirements\u002FInvariants already\n   extracted from the real, running code. Iterate the UI\u002FUX around behavior\n   that's already there.\n3. **Neither exists, and the user explicitly wants pure exploratory\n   prototyping** → the Plan phase may propose a **small** scope (not\n   12-16 features — think 3-5, sized to what a single fast iteration loop\n   can meaningfully evaluate), labeled unmistakably\n   `[EXPLORATORY — NOT APPROVED]`, and gated behind the same approval\n   requirement as `dev-kickoff-edho-ferdian`'s own Provisional Task Plan:\n   the user must approve, reject, or amend it before Generate starts.\n   Never treat an exploratory scope as if it were derived from real\n   requirements, and never let its language drift toward sounding\n   authoritative after a few iterations.\n\nFull mechanics, templates, and the decision tree for which source applies:\n**`references\u002Fspec-and-plan.md`**.\n\n## The three phases\n\n```\nPhase 1  Plan       — derive scope from a real source (see above) → references\u002Fspec-and-plan.md\nPhase 2  Generate    — build\u002Fiterate the live app                  → references\u002Fgenerate-phase.md\nPhase 3  Evaluate    — drive the live app, score, feed back, loop   → references\u002Fevaluate-phase.md\n```\n\n### Why Generate and Evaluate are separate agents, not just separate sections\n\nPlan stays in the main thread — it's about faithfully extracting scope\nthat already exists (Decision Register, mined specs), not adversarial\njudgment, so isolation buys little there. Generate and Evaluate are\ndifferent: the entire adversarial framing of this loop depends on Evaluate\nscoring the app **without having seen Generate's own reasoning about it**.\nA same-context \"now evaluate what you just built, but pretend you don't\nremember building it\" instruction is not a real isolation guarantee — the\ncontext is still there. Delegating each phase to its own agent\n(`gan-generator-edho-ferdian`, `gan-evaluator-edho-ferdian`) makes the\nseparation structural instead of just requested.\n\n### Phase 1 — Plan\n\nProduce a spec document and a rubric, same shape as the original concept\n(`spec.md` + `eval-rubric.md`, or equivalent paths inside this project's\n`gan-harness\u002F` working directory), but sourced per the priority order above\ninstead of invented. See `references\u002Fspec-and-plan.md` for the exact\ndecision tree, source-reading rules, and the `[EXPLORATORY — NOT APPROVED]`\ntemplate.\n\n### Phase 2 — Generate\n\nOn a harness with sub-agent delegation (Claude Code, OpenCode, Hermes via\nits delegate_task template, ZCode's own Subagents), delegate this phase to\nthe `gan-generator-edho-ferdian` agent for each round rather than running\nit inline — see \"Why phases are separate agents\" below for why this\nmatters more here than it does for a typical review delegation. On a\nharness with no delegation primitive, run the phase inline instead, per\nthe instructions below. If the Evaluator's score stops improving for two\nrounds on the same issue, run the next Generate round one tier up instead\nof repeating the same attempt — config-hygiene-edho-ferdian's\n`references\u002Fharness-operation.md` §4.2.\n\nBuild fast, commit per\niteration, keep a dev server running, read the Evaluator's feedback file\nbefore every iteration after the first, fix issues in the priority order\nfunctionality → craft → design → originality. This phase conceptually maps\nonto `dev-kickoff-edho-ferdian`'s IMPLEMENT stage, but stays a **separate,\nfaster\u002Flooser variant here** rather than merging into it: no test-first\nrequirement, no per-task Reflection block, commits are iteration checkpoints\nrather than reviewed units of work. Full workflow, state-file format, and\nthe \"avoid AI slop\" quick-reference: **`references\u002Fgenerate-phase.md`**\n(the full craft checklist lives in `references\u002Ffrontend-craft-checklist.md`,\ncross-referenced from there). When the app being built is a React\u002FNext.js\nUI, also pull the matching authoring reference from\n`frontend-engineering-edho-ferdian` (`design-direction.md`, `ui-polish.md`,\n`motion-system.md`, `composition-and-ux.md`, `accessible-authoring.md`) —\nthe checklist tells you *what* to fix; that skill's references tell you\n*how* to build it right the first time, cutting iteration count instead of\nrelying on the Evaluate phase to catch it.\n\n### Phase 3 — Evaluate\n\nOn a harness with sub-agent delegation, delegate this phase to the\n`gan-evaluator-edho-ferdian` agent — this is the phase where isolation\nmatters most (see below): an evaluator that shares the Generator's context\nalready has its justifications in view and will tend to agree with them\ninstead of judging the app on its own terms. On a harness with no\ndelegation primitive, run the phase inline, and say so plainly in the\nfeedback file rather than letting the loop's output imply an isolation\nguarantee that wasn't actually available.\n\nDrives the **live running app** (not a code read) via whatever browser-\nautomation driver is actually available — detect at runtime, same\nrequirement as `e2e-testing-edho-ferdian`: do not hardcode one preferred\ntool (Playwright MCP was the original's only option; this ecosystem may also\nhave Chrome DevTools MCP, `windows-desktop-e2e`, or another driver installed).\nScores against the weighted rubric:\n\n```\nweighted = (design * 0.3) + (originality * 0.2) + (craft * 0.3) + (functionality * 0.2)\n```\n\nIterates until the weighted score crosses **7.0** or a max-iteration cap is\nhit (default 5 — confirm with the user for anything that could run\nunattended longer). **Evaluation-mode honesty is mandatory**: the feedback\nfile records the mode *actually achieved* — `live-driver`, `screenshot`, or\n`code-only` — never the mode that was merely requested. If driving the live\napp fails and the phase silently falls back to a static code read, that must\nbe stated as a `code-only` result, not scored as if a live evaluation\nhappened. Full driver-detection procedure, scoring calibration, feedback-file\nformat, and the iteration\u002Fstop-condition rules: **`references\u002Fevaluate-phase.md`**.\n\n## Relationship to other skills in this ecosystem\n\n- **`dev-kickoff-edho-ferdian`** — the source of real scope for Plan\n  Priority 1, and the skill to use instead of this one for production\n  feature work with correctness guarantees (test-first, full Reflection\u002FCCL,\n  project-memory sync). This skill's Generate phase is intentionally a\n  faster\u002Flooser cousin of dev-kickoff's IMPLEMENT stage, not a replacement.\n- **`spec-mining-edho-ferdian`** — the source of real scope for Plan\n  Priority 2, when iterating on UI\u002FUX for an existing brownfield codebase\n  that has no written spec.\n- **`e2e-testing-edho-ferdian`** — shares the runtime driver-detection\n  requirement with this skill's Evaluate phase; if that skill has already\n  established which driver is available in this session\u002Fproject, reuse that\n  finding instead of re-detecting.\n- **`code-review-edho-ferdian`** — the frontend-craft \"AI-slop\" audit\n  content preserved in `references\u002Ffrontend-craft-checklist.md` is generic\n  frontend-quality signal, not specific to the GAN loop. It is a candidate\n  future lens for `code-review-edho-ferdian` (a conditional lens active when\n  scope touches UI\u002Fcomponent code, same pattern as the existing\n  accessibility\u002FRAG lenses there). **This is a note for a future task, not\n  an edit made here** — `code-review-edho-ferdian` is out of scope for this\n  skill's build.\n\n## Guardrails\n\n1. **Never invent scope.** Plan phase sources are ranked above; an\n   exploratory scope is always labeled `[EXPLORATORY — NOT APPROVED]` and\n   gated on user approval, never silently promoted to \"the plan.\"\n2. **Never silently downgrade the evaluation mode.** Record what was\n   actually achieved; a failed live-driver attempt is a `code-only` or\n   `screenshot` result, reported as such.\n3. **Detect the driver, don't hardcode it.** Same rule as\n   `e2e-testing-edho-ferdian`.\n\n## Language routing (fixed — see skill-authoring-edho-ferdian's canonical contract)\n\nCommunication to the user (Plan\u002FGenerate\u002FEvaluate narration, feedback to the\nGenerator) in Bahasa Indonesia; generated code, spec documents, and\nevaluator reports in English — fixed, never ask. Full contract:\n`skill-authoring-edho-ferdian` §7.\n4. **This is not dev-kickoff.** Don't pull this skill's looser discipline\n   into production feature work, and don't pull dev-kickoff's heavier\n   discipline into a fast prototyping loop where it isn't wanted — name\n   which mode you're in when starting.\n5. **Iteration has a cap.** Never loop unbounded; confirm the max-iteration\n   count with the user if it isn't the 5-iteration default, and stop and\n   report honestly if the threshold isn't reached by the cap rather than\n   quietly extending it. Before starting a loop, or if one is behaving\n   oddly (spinning, score climbing while the app looks worse, scope\n   drifting), run the Gate 0 checklist and failure-mode guards in\n   **`references\u002Floop-design-review.md`** — the iteration cap alone bounds\n   cost, it doesn't catch a badly-designed loop.\n\n**Done criteria for a full run:** scope sourced and labeled per the priority\norder (never invented) · Generate produces a live, running app with commits\nper iteration · Evaluate reports the mode actually achieved every round ·\nloop stops at threshold (≥7.0) or max-iteration cap, whichever comes first,\nwith an honest final report either way.\n",{"name":5,"metadata":24,"description":7},{"last_reviewed":25},"2026-10-07","main","533d32ad55c92d40fbabe93cbe21fe048598fe5a",".agents\u002Fskills\u002Fgan-harness-edho-ferdian",0,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF","https:\u002F\u002Fgithub.com\u002Fedhoferdian","Elevating ideas through AI, rising from humble beginnings to mastering the frontiers of the web. Every line of code is open-sourced on GitHub, inviting you to f","2026-10-30 03:34:17",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add edhoferdian\u002FEEF --skill gan-harness-edho-ferdian",{"tool":40,"command":41},"bunx","bunx skills add edhoferdian\u002FEEF --skill gan-harness-edho-ferdian",{"tool":43,"command":44},"pnpm","pnpm dlx skills add edhoferdian\u002FEEF --skill gan-harness-edho-ferdian","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fgan-harness-edho-ferdian\u002FSKILL.md",true,[48,54,58,62,66],{"path":49,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":52,"sourceHtmlUrl":53},"references\u002Fevaluate-phase.md","REFERENCE","参考文档",11540,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fgan-harness-edho-ferdian\u002Freferences\u002Fevaluate-phase.md",{"path":55,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":56,"sourceHtmlUrl":57},"references\u002Ffrontend-craft-checklist.md",10885,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fgan-harness-edho-ferdian\u002Freferences\u002Ffrontend-craft-checklist.md",{"path":59,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":60,"sourceHtmlUrl":61},"references\u002Fgenerate-phase.md",5856,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fgan-harness-edho-ferdian\u002Freferences\u002Fgenerate-phase.md",{"path":63,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":64,"sourceHtmlUrl":65},"references\u002Floop-design-review.md",3472,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fgan-harness-edho-ferdian\u002Freferences\u002Floop-design-review.md",{"path":67,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":68,"sourceHtmlUrl":69},"references\u002Fspec-and-plan.md",6612,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fgan-harness-edho-ferdian\u002Freferences\u002Fspec-and-plan.md",[71,78,85,92,99,106],{"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},"01M7C9FY2EWE5NB7M3YW0KJX62","video-style-edho-ferdian","Measure a reference video's look and pacing (colour grade as a 3D LUT, cut rhythm, overlays) and reproduce it: grade, cut and composite new footage, verify numerically, hand off an NLE timeline. Use for \"tiru look video ini\", \"samakan grading\", \"bikin video dengan gaya referensi\", \"LUT dari referensi\", \"cut rhythm\".","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fvideo-style-edho-ferdian\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},"01M7C9FT74P7JXGQ73116B9WS1","test-authoring-edho-ferdian","Write unit and component tests well: React Testing Library, pytest, Go and Vue, plus regression-test patterns. Use for \"tulis test untuk component ini\", \"bagaimana test hook ini\", \"test yang bagus untuk fitur X\", \"tulis test pytest\u002FGo\u002FVue\", or dev-kickoff's TEST stage.","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Ftest-authoring-edho-ferdian\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":19},"01M7C9FMPE9W76FS7S8ZZ48ARE","system-design-edho-ferdian","Mid-project architecture decisions: ADRs, trade-off analysis, non-functional requirements, scaling tiers and blast radius. Use for \"desain arsitektur\", \"keputusan teknis besar\", \"bikin ADR\", \"trade-off antara X dan Y\", \"microservices atau monolith\", \"apa yang rusak kalau X diubah\", or a capacity question.","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fsystem-design-edho-ferdian\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":19},"01M7C9FETATAHKG5WKPT4AP1A2","spec-mining-edho-ferdian","Extract behavioural specs from a codebase with none: WHEN\u002FTHEN requirements and invariants, each anchored to the code that enforces it, unclear behaviour recorded as uncertainty. Use for \"ekstrak spec\", \"buat spec dari kode\", \"reverse-engineer the spec\", \"repo ini tidak ada dokumentasi\", or a missing BEHAVIOR_SPEC in dev-kickoff.","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fspec-mining-edho-ferdian\u002FSKILL.md",[],[],{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":19},"01M7C9F93YCXFCVRDG4S5QFAH4","skill-authoring-edho-ferdian","Create and govern this ecosystem's skills: search before building, a quality bar, compliance checks, promoting repeated principles to rules, and packaging dist\u002F*.skill. Use for \"bikin skill baru\", \"ada skill buat X gak\", \"fork skill ini\", \"skill gue kepake gak sih\", \"package skill ini\", \"buatkan .skill-nya\".","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fskill-authoring-edho-ferdian\u002FSKILL.md",[],[],{"skillNo":107,"name":108,"nameZh":6,"description":109,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":110,"categories":111,"keywords":112,"publishedAt":19},"01M7C9F4XZJ59R0GS4YEP4E1RD","skill-audit-edho-ferdian","Audit this repo's skills\u002F for staleness, redundancy, broken cross-references and weak descriptions (not the ~\u002F.claude setup: that is config-hygiene). Use for \"audit skill saya\", \"cek skill yang sudah dibuat\", \"ada yang redundan gak\", \"skill mana yang basi\".","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fskill-audit-edho-ferdian\u002FSKILL.md",[],[]]