[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7FBMYVF3BFPBVFC5W2VYX14":3,"skill-same:01M7FBMYVF3BFPBVFC5W2VYX14":170},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":15,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":36,"commitSha":37,"skillDirectory":38,"forksCount":13,"repositoryHtmlUrl":39,"repositoryHomepageUrl":6,"ownerHtmlUrl":40,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":41,"install":42,"resources":55},"01M7FBMYVF3BFPBVFC5W2VYX14","spec-forge",null,"Specification system for creating, auditing, and propagating requirements docs (PRD, IEEE 830 SRS, tech design, feature specs, and test plans).","hienpham0344\u002FAgent-kit","hienpham0344","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F202117481?v=4","01M7FAXAHS9A2VJ6F0AF1W8EY7","01M7FAXAJDAEP8NEG7B0KFTS90",0,"MIT","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002FSKILL.md",[],[],"2026-11-16 10:05:34",false,"---\nname: spec-forge\ndescription: Specification system for creating, auditing, and propagating requirements docs (PRD, IEEE 830 SRS, tech design, feature specs, and test plans).\nversion: 1.0\nsubcommands:\n  - idea\n  - decompose\n  - tech-design\n  - prd\n  - srs\n  - test-cases\n  - review\n  - audit\n  - analyze\n  - propagate\n---\n\n# Spec-Forge Orchestrator\n\nSpec-forge is a documentation and specification workflow suite. This root skill is\nthe Codex-facing orchestrator: decide which child skill to load, keep the document\nchain coherent, and avoid loading large generation references until they are needed.\n\n## Trigger Patterns\n\nUse this skill when the user asks for any of the following:\n\n- Start or refine an idea before specs: use `skills\u002Fidea\u002FSKILL.md`\n- Split a large project into sub-features: use `skills\u002Fdecompose\u002FSKILL.md`\n- Write a PRD, business case, or product definition — \"is this worth building\": use `skills\u002Fprd-generation\u002FSKILL.md`\n- Write the requirements a team will be held to — a requirements spec, an SRS, functional\u002Fnon-functional requirements, **a requirements document to hand to an implementation team or vendor**: use `skills\u002Fsrs-generation\u002FSKILL.md`\n- Write a technical design, architecture doc, design doc, or RFC: use `skills\u002Ftech-design-generation\u002FSKILL.md`\n- Generate test cases, QA coverage, or a test case matrix: use `skills\u002Ftest-cases-generation\u002FSKILL.md`\n- Review spec-forge documents for quality and consistency: use `skills\u002Freview\u002FSKILL.md`\n- Audit project docs against code: use `skills\u002Faudit\u002FSKILL.md`\n- Analyze a document ecosystem for themes, conflicts, gaps, redundancy, or staleness: use `skills\u002Fanalyze\u002FSKILL.md`\n- Propagate an upstream doc change downstream: use `skills\u002Fpropagate\u002FSKILL.md`\n- Run the full chain for a feature: orchestrate `[idea] -> [decompose] -> [prd] -> srs -> tech-design -> review -> handoff to code-forge`\n\n**Routing the ambiguous \"write me a requirements document\".** When the user says\n\"需求文档\", \"requirements doc\", or \"spec\" without qualifying it, the deciding\nquestion is whether the decision to build has already been made. If it has — the\nuser knows what they want and needs it specified precisely enough to implement\nagainst — route to **SRS**. If it has not — the user is making a case, seeking\napproval, or sizing an opportunity — route to **PRD**. When genuinely unclear,\nask that one question rather than guessing; producing the wrong document wastes\nthe entire generation.\n\nAlso treat slash-like requests as aliases:\n\n| User request | Route |\n|---|---|\n| `\u002Fspec-forge \u003Cname>` | Full chain |\n| `\u002Fspec-forge:idea \u003Cname>` or `\u002Fidea \u003Cname>` | Idea |\n| `\u002Fspec-forge:decompose \u003Cname>` or `\u002Fdecompose \u003Cname>` | Decompose |\n| `\u002Fspec-forge:tech-design \u003Cname>` or `\u002Ftech-design \u003Cname>` | Tech Design |\n| `\u002Fspec-forge:prd \u003Cname>` or `\u002Fprd \u003Cname>` | PRD |\n| `\u002Fspec-forge:srs \u003Cname>` or `\u002Fsrs \u003Cname>` | SRS |\n| `\u002Fspec-forge:test-cases \u003Ctarget>` or `\u002Ftest-cases \u003Ctarget>` | Test Cases |\n| `\u002Fspec-forge:review \u003Cname>` or `\u002Freview \u003Cname>` | Review |\n| `\u002Fspec-forge:audit [path]` or `\u002Faudit [path]` | Audit |\n| `\u002Fspec-forge:analyze [path]` or `\u002Fanalyze [path]` | Analyze |\n| `\u002Fspec-forge:propagate ...` or `\u002Fpropagate ...` | Propagate |\n\nThe `commands\u002F*.md` files are compatibility command definitions from the original\nClaude-oriented implementation. In Codex, prefer reading the child `skills\u002F*\u002FSKILL.md`\nfiles directly. Use `commands\u002Fspec-forge.md` only as an additional orchestration\nreference if you need more detail about the full chain.\n\n## Progressive Loading\n\nKeep context small.\n\n1. Read this root file first.\n2. Read only the child `SKILL.md` matching the user's request.\n3. When a child skill references `skills\u002Fshared\u002Fproject-context.md`, read it if the workflow needs code\u002Fproject awareness.\n4. When a child skill references `skills\u002Fshared\u002Fdoc-first.md`, apply it before generating or editing docs.\n5. When a child skill invokes the script layer, read `skills\u002Fshared\u002Fscripts.md` once to resolve `\u003Csf_scripts>`; do not read the scripts' source unless you need to debug them.\n6. Read generation references, templates, or checklists only when that child skill explicitly needs them.\n\nDo not bulk-load every child skill. Do not read every reference folder just because\nthe suite is complex.\n\n## Global Rules\n\n- Preserve the user's requested language. If the user writes in Chinese, respond and write user-facing explanations in Chinese unless the target document convention requires English.\n- Treat docs as current truth, not history. Before generating a new spec, scan for existing related docs and decide whether to reuse, edit, or create.\n- Do not create parallel versions such as `prd-v2.md`, `new-tech-design.md`, or addendum files. Edit the canonical file in place unless no canonical document exists.\n- Keep traceability stable: preserve existing requirement IDs, feature names, component names, and section titles unless the user explicitly asks to rename them.\n- When changing upstream docs, identify downstream impact. Either update affected downstream docs or recommend\u002Frun the propagate workflow.\n- Every finding in review, audit, analyze, or propagate output must cite concrete files\u002Fsections. Do not invent issues.\n- Prefer project-local conventions and existing docs over generic templates.\n- Ask only necessary questions. If existing docs or code provide the answer, use them and state the assumption.\n\n## Shared Project Context\n\nFor generation workflows that need grounding in the actual codebase, use:\n\n`skills\u002Fshared\u002Fproject-context.md`\n\nApply it before SRS, PRD, tech-design, decompose, and test-cases work when the\nproject exists. It produces a concise summary of:\n\n- Project profile\n- Language, framework, database, auth, testing, and CI signals\n- Existing documents and upstream specs\n- Architecture and module boundaries\n- Existing test infrastructure when relevant\n\nFor the full chain, scan project context once and reuse the summary across stages\ninstead of repeatedly rescanning the same repository.\n\n## Script Layer\n\nspec-forge ships a small layer of deterministic Python 3 (standard-library-only)\nhelpers under `skills\u002Fshared\u002Fscripts\u002F` that absorb the mechanical, token-heavy\nbookkeeping the skills used to do by hand — project + document scanning\n(`sf-scan.py`), configuration\u002Fdirectory resolution (`sf-config.py`), document\nstructure\u002FID validation (`sf-verify-doc.py`), and traceability\u002Fcoverage math\n(`sf-trace.py`). The division of labour: **the scripts keep the record-keeping;\nthe model keeps the reasoning** (requirement quality, architecture trade-offs,\nconflict judgement). Every skill that calls a script keeps a silent manual\nfallback for when `python3` is unavailable.\n\nHow to locate and invoke the scripts, and which one does what:\n\n`skills\u002Fshared\u002Fscripts.md`\n\nPrefer the fast path (run the script, use its JSON) over re-deriving the same\nfacts by hand. Never restate a script's algorithm in prose — the script is the\nsingle source of truth so the two cannot drift.\n\n## Document Discipline\n\nFor any document creation or edit, apply:\n\n`skills\u002Fshared\u002Fdoc-first.md`\n\nMinimum pre-write checklist:\n\n1. List relevant existing docs under `docs\u002F`, `specs\u002F`, `design\u002F`, `architecture\u002F`, `ideas\u002F`, and the project root.\n2. Grep for the feature name, domain terms, requirement IDs, component names, and aliases.\n3. Decide for each topic: reuse existing content, extend existing content, or create new content.\n4. Preserve stable IDs and terminology.\n5. Note downstream docs that will need updates.\n\n## Full Chain\n\nThe chain has one mandatory spine — **a delivery-grade requirements specification, then a technical design** — with everything else conditional on what the project actually needs. When the user asks for `\u002Fspec-forge \u003Cname>` or \"run the full spec-forge chain\", execute this sequence:\n\n1. **Idea** *(conditional)*: Use `skills\u002Fidea\u002FSKILL.md`.\n   - Run when the demand itself is unvalidated: the user is exploring, the problem is fuzzy, or there is no evidence anyone needs this.\n   - Skip when the user already knows what they want built.\n   - Goal: validate demand and produce `ideas\u002F{name}\u002Fdraft.md`.\n   - If a ready or graduated idea already exists, reuse it. If the idea is still exploring\u002Frefining, warn the user before continuing.\n\n2. **Decompose** *(conditional)*: Use `skills\u002Fdecompose\u002FSKILL.md`.\n   - Run when the scope plausibly covers several independently specifiable sub-features.\n   - Skip for a single coherent feature.\n   - Multi-split output: `docs\u002Fproject-{name}.md` with a first-line `FEATURE_MANIFEST` block.\n\n3. **PRD** *(conditional)*: Use `skills\u002Fprd-generation\u002FSKILL.md`.\n   - Run for a new product, a new product line, or anything needing a go\u002Fno-go decision, budget approval, or stakeholder alignment before commitment.\n   - Skip when the decision to build is already made and the work is a well-understood feature — go straight to the SRS. A PRD written after the decision is ceremony.\n   - Output: `docs\u002F{name}\u002Fprd.md` — business case and product definition.\n\n4. **SRS** *(mandatory)*: Use `skills\u002Fsrs-generation\u002FSKILL.md`.\n   - **This is the spine of the chain.** It is the delivery contract: the document an implementation team — an external vendor, another team, or code-forge — is held to.\n   - Output: `docs\u002F{name}\u002Fsrs.md` with `FR-*`\u002F`NFR-*` requirements, input field rules, state machines, permission matrix, error catalogue, acceptance criteria, and change control.\n   - If `docs\u002F{name}\u002Fprd.md` exists, trace every capability into requirements. If only `ideas\u002F{name}\u002Fdraft.md` exists, derive scope and requirements from it. If neither exists, work from the user's request and state the assumptions explicitly.\n   - Never skip this stage to reach the tech design faster. A design built on unstated requirements produces code whose correctness nobody can settle.\n\n5. **Tech Design** *(mandatory)*: Use `skills\u002Ftech-design-generation\u002FSKILL.md`.\n   - Single feature output: `docs\u002F{name}\u002Ftech-design.md`.\n   - Multi-split output: one tech design per sub-feature.\n   - Also generates one implementation-facing feature spec under `docs\u002Ffeatures\u002F` per component in §8.1, plus `docs\u002Ffeatures\u002Foverview.md` with the dependency graph and execution order.\n   - Every component must trace to the `FR-*`\u002F`NFR-*` requirements it satisfies, and every requirement must be satisfied by at least one component.\n\n6. **Review** *(mandatory)*: Use `skills\u002Freview\u002FSKILL.md`.\n   - Review the SRS, the tech design, and any `docs\u002Ffeatures\u002F*.md`.\n   - Auto-fix only where the review skill permits it. Leave review comments where domain knowledge is missing.\n\n7. **Handoff**: Report the handoff command rather than running it.\n   - `\u002Fcode-forge:plan @docs\u002Ffeatures\u002F` when feature specs were generated.\n   - `\u002Fcode-forge:plan @docs\u002F{name}\u002Ftech-design.md` for a single-component design.\n   - Implementation and test generation belong to code-forge. spec-forge owns requirements and design; code-forge owns implementation and verification. Tests are derived from the SRS acceptance criteria by `code-forge:tdd` — spec-forge does not produce a separate test document in the chain.\n\nTest cases, audit, analyze, and propagate are on-demand outputs, not stages in the default chain.\n\n## Single-Skill Routes\n\n### Idea\n\nRead `skills\u002Fidea\u002FSKILL.md`.\n\nUse for early exploration, demand validation, competitive research, and\nanti-pseudo-requirement checks. Output lives in top-level `ideas\u002F{idea-name}\u002F`,\nnot under `docs\u002F`.\n\n### Decompose\n\nRead `skills\u002Fdecompose\u002FSKILL.md`.\n\nUse for deciding whether a project is one feature or several independently\nspecifiable sub-features. If multi-split, generate `docs\u002Fproject-{name}.md`.\n\n### PRD — the business case\n\nRead `skills\u002Fprd-generation\u002FSKILL.md`.\n\nAnswers **\"is this worth building, for whom, and at what priority\"**. Market\ncontext, competitive landscape, demand validation, feasibility go\u002Fno-go,\npersonas, user stories, capability scope, success metrics, milestones, risks.\nRead by stakeholders *before* commitment and largely frozen once the decision is\nmade. Optional: skip it when the decision to build is already settled.\n\nDeliberately **not** in the PRD: field-level rules, state machines, error codes,\nGiven\u002FWhen\u002FThen acceptance criteria, solution architecture. Those are the SRS's\nand tech design's job. A PRD row that needs more than two sentences is a system\nbehavior in the wrong document.\n\n### SRS — the delivery contract\n\nRead `skills\u002Fsrs-generation\u002FSKILL.md`.\n\nAnswers **\"exactly what must be built, and how do we settle whether it was\"**.\nThis is the document an implementation team — an external vendor, another team,\nor code-forge — is held to, and the mandatory spine of the chain. Formal\n`FR-*`\u002F`NFR-*` requirements with main\u002Falternative flows, **input field rules**,\n**state machines including illegal-transition handling**, **permission matrix**,\n**error catalogue with degradation behavior**, machine-verifiable acceptance\ncriteria, interface and data requirements, traceability, and **acceptance and\nchange control**.\n\nUnlike the PRD, it keeps evolving through delivery — which is exactly why the\ntwo are separate files. A requirement change must not require touching market\nsizing, and internal go\u002Fno-go reasoning must not ship to an implementation\nvendor.\n\n### Tech Design\n\nRead `skills\u002Ftech-design-generation\u002FSKILL.md`.\n\nUse for architecture\u002Fdesign docs, RFCs, API design, data model design,\ndeployment, observability, security, performance, alternatives, and generated\nfeature specs.\n\n### Test Cases\n\nRead `skills\u002Ftest-cases-generation\u002FSKILL.md`.\n\nUse for test case generation, coverage matrices, test strategy, testable unit\nextraction, and code\u002Fspec-driven QA planning. Honor `--formal` when requested.\n\n### Review\n\nRead `skills\u002Freview\u002FSKILL.md`.\n\nUse after spec generation to check completeness, specificity, internal\nconsistency, traceability, and actionability. Findings first; fixes only where\nallowed.\n\n### Audit\n\nRead `skills\u002Faudit\u002FSKILL.md`.\n\nUse when auditing one project that has docs and code. It checks documentation\nquality, completeness, consistency, staleness, and alignment with the codebase.\n\n### Analyze\n\nRead `skills\u002Fanalyze\u002FSKILL.md`.\n\nUse for docs-only or cross-repo document landscapes. It maps themes, conflicts,\ngaps, redundancies, and stale areas without assuming a single codebase.\n\n### Propagate\n\nRead `skills\u002Fpropagate\u002FSKILL.md`.\n\nUse after upstream document changes. It finds downstream docs affected by changed\nconcepts and updates them surgically or reports unresolved stale references.\n\n## Output Locations\n\nDefault locations:\n\n- Ideas: `ideas\u002F{name}\u002F`\n- PRD (business case, optional): `docs\u002F{name}\u002Fprd.md`\n- SRS (delivery contract, mandatory): `docs\u002F{name}\u002Fsrs.md`\n- Tech design: `docs\u002F{name}\u002Ftech-design.md`\n- Feature specs: `docs\u002Ffeatures\u002F*.md`\n- Test cases: `docs\u002F{name}\u002Ftest-cases.md` or the path specified by the child skill\n- Project manifest: `docs\u002Fproject-{name}.md`\n- Audit report: `{target-docs-path}\u002Faudit-report.md`\n- Analysis report: `{target-root}\u002Fanalysis-report.md`\n\nIf existing project conventions differ, follow the existing convention and state\nthe location you used.\n\n## Completion Criteria\n\nBefore finishing a spec-forge task:\n\n- Confirm the requested document or report exists, or explain why it could not be written.\n- Mention the exact files changed or generated.\n- For generation tasks, note whether project context and existing docs were scanned.\n- For review\u002Faudit\u002Fanalyze tasks, summarize findings by severity and cite the report path.\n- For chain tasks, summarize which stages completed, which conditional stages were skipped and why, and state the `\u002Fcode-forge:plan` handoff command.\n- For a full chain, confirm the SRS exists and passed its structural gate. The SRS is the chain's spine; a chain that reached a tech design without one is incomplete, not merely abbreviated.\n","# Spec-Forge Orchestrator\n\nSpec-forge is a documentation and specification workflow suite. This root skill is\nthe Codex-facing orchestrator: decide which child skill to load, keep the document\nchain coherent, and avoid loading large generation references until they are needed.\n\n## Trigger Patterns\n\nUse this skill when the user asks for any of the following:\n\n- Start or refine an idea before specs: use `skills\u002Fidea\u002FSKILL.md`\n- Split a large project into sub-features: use `skills\u002Fdecompose\u002FSKILL.md`\n- Write a PRD, business case, or product definition — \"is this worth building\": use `skills\u002Fprd-generation\u002FSKILL.md`\n- Write the requirements a team will be held to — a requirements spec, an SRS, functional\u002Fnon-functional requirements, **a requirements document to hand to an implementation team or vendor**: use `skills\u002Fsrs-generation\u002FSKILL.md`\n- Write a technical design, architecture doc, design doc, or RFC: use `skills\u002Ftech-design-generation\u002FSKILL.md`\n- Generate test cases, QA coverage, or a test case matrix: use `skills\u002Ftest-cases-generation\u002FSKILL.md`\n- Review spec-forge documents for quality and consistency: use `skills\u002Freview\u002FSKILL.md`\n- Audit project docs against code: use `skills\u002Faudit\u002FSKILL.md`\n- Analyze a document ecosystem for themes, conflicts, gaps, redundancy, or staleness: use `skills\u002Fanalyze\u002FSKILL.md`\n- Propagate an upstream doc change downstream: use `skills\u002Fpropagate\u002FSKILL.md`\n- Run the full chain for a feature: orchestrate `[idea] -> [decompose] -> [prd] -> srs -> tech-design -> review -> handoff to code-forge`\n\n**Routing the ambiguous \"write me a requirements document\".** When the user says\n\"需求文档\", \"requirements doc\", or \"spec\" without qualifying it, the deciding\nquestion is whether the decision to build has already been made. If it has — the\nuser knows what they want and needs it specified precisely enough to implement\nagainst — route to **SRS**. If it has not — the user is making a case, seeking\napproval, or sizing an opportunity — route to **PRD**. When genuinely unclear,\nask that one question rather than guessing; producing the wrong document wastes\nthe entire generation.\n\nAlso treat slash-like requests as aliases:\n\n| User request | Route |\n|---|---|\n| `\u002Fspec-forge \u003Cname>` | Full chain |\n| `\u002Fspec-forge:idea \u003Cname>` or `\u002Fidea \u003Cname>` | Idea |\n| `\u002Fspec-forge:decompose \u003Cname>` or `\u002Fdecompose \u003Cname>` | Decompose |\n| `\u002Fspec-forge:tech-design \u003Cname>` or `\u002Ftech-design \u003Cname>` | Tech Design |\n| `\u002Fspec-forge:prd \u003Cname>` or `\u002Fprd \u003Cname>` | PRD |\n| `\u002Fspec-forge:srs \u003Cname>` or `\u002Fsrs \u003Cname>` | SRS |\n| `\u002Fspec-forge:test-cases \u003Ctarget>` or `\u002Ftest-cases \u003Ctarget>` | Test Cases |\n| `\u002Fspec-forge:review \u003Cname>` or `\u002Freview \u003Cname>` | Review |\n| `\u002Fspec-forge:audit [path]` or `\u002Faudit [path]` | Audit |\n| `\u002Fspec-forge:analyze [path]` or `\u002Fanalyze [path]` | Analyze |\n| `\u002Fspec-forge:propagate ...` or `\u002Fpropagate ...` | Propagate |\n\nThe `commands\u002F*.md` files are compatibility command definitions from the original\nClaude-oriented implementation. In Codex, prefer reading the child `skills\u002F*\u002FSKILL.md`\nfiles directly. Use `commands\u002Fspec-forge.md` only as an additional orchestration\nreference if you need more detail about the full chain.\n\n## Progressive Loading\n\nKeep context small.\n\n1. Read this root file first.\n2. Read only the child `SKILL.md` matching the user's request.\n3. When a child skill references `skills\u002Fshared\u002Fproject-context.md`, read it if the workflow needs code\u002Fproject awareness.\n4. When a child skill references `skills\u002Fshared\u002Fdoc-first.md`, apply it before generating or editing docs.\n5. When a child skill invokes the script layer, read `skills\u002Fshared\u002Fscripts.md` once to resolve `\u003Csf_scripts>`; do not read the scripts' source unless you need to debug them.\n6. Read generation references, templates, or checklists only when that child skill explicitly needs them.\n\nDo not bulk-load every child skill. Do not read every reference folder just because\nthe suite is complex.\n\n## Global Rules\n\n- Preserve the user's requested language. If the user writes in Chinese, respond and write user-facing explanations in Chinese unless the target document convention requires English.\n- Treat docs as current truth, not history. Before generating a new spec, scan for existing related docs and decide whether to reuse, edit, or create.\n- Do not create parallel versions such as `prd-v2.md`, `new-tech-design.md`, or addendum files. Edit the canonical file in place unless no canonical document exists.\n- Keep traceability stable: preserve existing requirement IDs, feature names, component names, and section titles unless the user explicitly asks to rename them.\n- When changing upstream docs, identify downstream impact. Either update affected downstream docs or recommend\u002Frun the propagate workflow.\n- Every finding in review, audit, analyze, or propagate output must cite concrete files\u002Fsections. Do not invent issues.\n- Prefer project-local conventions and existing docs over generic templates.\n- Ask only necessary questions. If existing docs or code provide the answer, use them and state the assumption.\n\n## Shared Project Context\n\nFor generation workflows that need grounding in the actual codebase, use:\n\n`skills\u002Fshared\u002Fproject-context.md`\n\nApply it before SRS, PRD, tech-design, decompose, and test-cases work when the\nproject exists. It produces a concise summary of:\n\n- Project profile\n- Language, framework, database, auth, testing, and CI signals\n- Existing documents and upstream specs\n- Architecture and module boundaries\n- Existing test infrastructure when relevant\n\nFor the full chain, scan project context once and reuse the summary across stages\ninstead of repeatedly rescanning the same repository.\n\n## Script Layer\n\nspec-forge ships a small layer of deterministic Python 3 (standard-library-only)\nhelpers under `skills\u002Fshared\u002Fscripts\u002F` that absorb the mechanical, token-heavy\nbookkeeping the skills used to do by hand — project + document scanning\n(`sf-scan.py`), configuration\u002Fdirectory resolution (`sf-config.py`), document\nstructure\u002FID validation (`sf-verify-doc.py`), and traceability\u002Fcoverage math\n(`sf-trace.py`). The division of labour: **the scripts keep the record-keeping;\nthe model keeps the reasoning** (requirement quality, architecture trade-offs,\nconflict judgement). Every skill that calls a script keeps a silent manual\nfallback for when `python3` is unavailable.\n\nHow to locate and invoke the scripts, and which one does what:\n\n`skills\u002Fshared\u002Fscripts.md`\n\nPrefer the fast path (run the script, use its JSON) over re-deriving the same\nfacts by hand. Never restate a script's algorithm in prose — the script is the\nsingle source of truth so the two cannot drift.\n\n## Document Discipline\n\nFor any document creation or edit, apply:\n\n`skills\u002Fshared\u002Fdoc-first.md`\n\nMinimum pre-write checklist:\n\n1. List relevant existing docs under `docs\u002F`, `specs\u002F`, `design\u002F`, `architecture\u002F`, `ideas\u002F`, and the project root.\n2. Grep for the feature name, domain terms, requirement IDs, component names, and aliases.\n3. Decide for each topic: reuse existing content, extend existing content, or create new content.\n4. Preserve stable IDs and terminology.\n5. Note downstream docs that will need updates.\n\n## Full Chain\n\nThe chain has one mandatory spine — **a delivery-grade requirements specification, then a technical design** — with everything else conditional on what the project actually needs. When the user asks for `\u002Fspec-forge \u003Cname>` or \"run the full spec-forge chain\", execute this sequence:\n\n1. **Idea** *(conditional)*: Use `skills\u002Fidea\u002FSKILL.md`.\n   - Run when the demand itself is unvalidated: the user is exploring, the problem is fuzzy, or there is no evidence anyone needs this.\n   - Skip when the user already knows what they want built.\n   - Goal: validate demand and produce `ideas\u002F{name}\u002Fdraft.md`.\n   - If a ready or graduated idea already exists, reuse it. If the idea is still exploring\u002Frefining, warn the user before continuing.\n\n2. **Decompose** *(conditional)*: Use `skills\u002Fdecompose\u002FSKILL.md`.\n   - Run when the scope plausibly covers several independently specifiable sub-features.\n   - Skip for a single coherent feature.\n   - Multi-split output: `docs\u002Fproject-{name}.md` with a first-line `FEATURE_MANIFEST` block.\n\n3. **PRD** *(conditional)*: Use `skills\u002Fprd-generation\u002FSKILL.md`.\n   - Run for a new product, a new product line, or anything needing a go\u002Fno-go decision, budget approval, or stakeholder alignment before commitment.\n   - Skip when the decision to build is already made and the work is a well-understood feature — go straight to the SRS. A PRD written after the decision is ceremony.\n   - Output: `docs\u002F{name}\u002Fprd.md` — business case and product definition.\n\n4. **SRS** *(mandatory)*: Use `skills\u002Fsrs-generation\u002FSKILL.md`.\n   - **This is the spine of the chain.** It is the delivery contract: the document an implementation team — an external vendor, another team, or code-forge — is held to.\n   - Output: `docs\u002F{name}\u002Fsrs.md` with `FR-*`\u002F`NFR-*` requirements, input field rules, state machines, permission matrix, error catalogue, acceptance criteria, and change control.\n   - If `docs\u002F{name}\u002Fprd.md` exists, trace every capability into requirements. If only `ideas\u002F{name}\u002Fdraft.md` exists, derive scope and requirements from it. If neither exists, work from the user's request and state the assumptions explicitly.\n   - Never skip this stage to reach the tech design faster. A design built on unstated requirements produces code whose correctness nobody can settle.\n\n5. **Tech Design** *(mandatory)*: Use `skills\u002Ftech-design-generation\u002FSKILL.md`.\n   - Single feature output: `docs\u002F{name}\u002Ftech-design.md`.\n   - Multi-split output: one tech design per sub-feature.\n   - Also generates one implementation-facing feature spec under `docs\u002Ffeatures\u002F` per component in §8.1, plus `docs\u002Ffeatures\u002Foverview.md` with the dependency graph and execution order.\n   - Every component must trace to the `FR-*`\u002F`NFR-*` requirements it satisfies, and every requirement must be satisfied by at least one component.\n\n6. **Review** *(mandatory)*: Use `skills\u002Freview\u002FSKILL.md`.\n   - Review the SRS, the tech design, and any `docs\u002Ffeatures\u002F*.md`.\n   - Auto-fix only where the review skill permits it. Leave review comments where domain knowledge is missing.\n\n7. **Handoff**: Report the handoff command rather than running it.\n   - `\u002Fcode-forge:plan @docs\u002Ffeatures\u002F` when feature specs were generated.\n   - `\u002Fcode-forge:plan @docs\u002F{name}\u002Ftech-design.md` for a single-component design.\n   - Implementation and test generation belong to code-forge. spec-forge owns requirements and design; code-forge owns implementation and verification. Tests are derived from the SRS acceptance criteria by `code-forge:tdd` — spec-forge does not produce a separate test document in the chain.\n\nTest cases, audit, analyze, and propagate are on-demand outputs, not stages in the default chain.\n\n## Single-Skill Routes\n\n### Idea\n\nRead `skills\u002Fidea\u002FSKILL.md`.\n\nUse for early exploration, demand validation, competitive research, and\nanti-pseudo-requirement checks. Output lives in top-level `ideas\u002F{idea-name}\u002F`,\nnot under `docs\u002F`.\n\n### Decompose\n\nRead `skills\u002Fdecompose\u002FSKILL.md`.\n\nUse for deciding whether a project is one feature or several independently\nspecifiable sub-features. If multi-split, generate `docs\u002Fproject-{name}.md`.\n\n### PRD — the business case\n\nRead `skills\u002Fprd-generation\u002FSKILL.md`.\n\nAnswers **\"is this worth building, for whom, and at what priority\"**. Market\ncontext, competitive landscape, demand validation, feasibility go\u002Fno-go,\npersonas, user stories, capability scope, success metrics, milestones, risks.\nRead by stakeholders *before* commitment and largely frozen once the decision is\nmade. Optional: skip it when the decision to build is already settled.\n\nDeliberately **not** in the PRD: field-level rules, state machines, error codes,\nGiven\u002FWhen\u002FThen acceptance criteria, solution architecture. Those are the SRS's\nand tech design's job. A PRD row that needs more than two sentences is a system\nbehavior in the wrong document.\n\n### SRS — the delivery contract\n\nRead `skills\u002Fsrs-generation\u002FSKILL.md`.\n\nAnswers **\"exactly what must be built, and how do we settle whether it was\"**.\nThis is the document an implementation team — an external vendor, another team,\nor code-forge — is held to, and the mandatory spine of the chain. Formal\n`FR-*`\u002F`NFR-*` requirements with main\u002Falternative flows, **input field rules**,\n**state machines including illegal-transition handling**, **permission matrix**,\n**error catalogue with degradation behavior**, machine-verifiable acceptance\ncriteria, interface and data requirements, traceability, and **acceptance and\nchange control**.\n\nUnlike the PRD, it keeps evolving through delivery — which is exactly why the\ntwo are separate files. A requirement change must not require touching market\nsizing, and internal go\u002Fno-go reasoning must not ship to an implementation\nvendor.\n\n### Tech Design\n\nRead `skills\u002Ftech-design-generation\u002FSKILL.md`.\n\nUse for architecture\u002Fdesign docs, RFCs, API design, data model design,\ndeployment, observability, security, performance, alternatives, and generated\nfeature specs.\n\n### Test Cases\n\nRead `skills\u002Ftest-cases-generation\u002FSKILL.md`.\n\nUse for test case generation, coverage matrices, test strategy, testable unit\nextraction, and code\u002Fspec-driven QA planning. Honor `--formal` when requested.\n\n### Review\n\nRead `skills\u002Freview\u002FSKILL.md`.\n\nUse after spec generation to check completeness, specificity, internal\nconsistency, traceability, and actionability. Findings first; fixes only where\nallowed.\n\n### Audit\n\nRead `skills\u002Faudit\u002FSKILL.md`.\n\nUse when auditing one project that has docs and code. It checks documentation\nquality, completeness, consistency, staleness, and alignment with the codebase.\n\n### Analyze\n\nRead `skills\u002Fanalyze\u002FSKILL.md`.\n\nUse for docs-only or cross-repo document landscapes. It maps themes, conflicts,\ngaps, redundancies, and stale areas without assuming a single codebase.\n\n### Propagate\n\nRead `skills\u002Fpropagate\u002FSKILL.md`.\n\nUse after upstream document changes. It finds downstream docs affected by changed\nconcepts and updates them surgically or reports unresolved stale references.\n\n## Output Locations\n\nDefault locations:\n\n- Ideas: `ideas\u002F{name}\u002F`\n- PRD (business case, optional): `docs\u002F{name}\u002Fprd.md`\n- SRS (delivery contract, mandatory): `docs\u002F{name}\u002Fsrs.md`\n- Tech design: `docs\u002F{name}\u002Ftech-design.md`\n- Feature specs: `docs\u002Ffeatures\u002F*.md`\n- Test cases: `docs\u002F{name}\u002Ftest-cases.md` or the path specified by the child skill\n- Project manifest: `docs\u002Fproject-{name}.md`\n- Audit report: `{target-docs-path}\u002Faudit-report.md`\n- Analysis report: `{target-root}\u002Fanalysis-report.md`\n\nIf existing project conventions differ, follow the existing convention and state\nthe location you used.\n\n## Completion Criteria\n\nBefore finishing a spec-forge task:\n\n- Confirm the requested document or report exists, or explain why it could not be written.\n- Mention the exact files changed or generated.\n- For generation tasks, note whether project context and existing docs were scanned.\n- For review\u002Faudit\u002Fanalyze tasks, summarize findings by severity and cite the report path.\n- For chain tasks, summarize which stages completed, which conditional stages were skipped and why, and state the `\u002Fcode-forge:plan` handoff command.\n- For a full chain, confirm the SRS exists and passed its structural gate. The SRS is the chain's spine; a chain that reached a tech design without one is incomplete, not merely abbreviated.\n",{"name":5,"version":24,"description":7,"subcommands":25},1,[26,27,28,29,30,31,32,33,34,35],"idea","decompose","tech-design","prd","srs","test-cases","review","audit","analyze","propagate","main","98e910e6f821daa03a9eebb76d4c12b94fdb2699","skills\u002Fplanning-grilling\u002Fspec-forge","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit","https:\u002F\u002Fgithub.com\u002Fhienpham0344","2026-11-13 02:55:47",{"source":8,"skillName":5,"commands":43,"prompt":53,"zipAvailable":54},[44,47,50],{"tool":45,"command":46},"npx","npx skills add hienpham0344\u002FAgent-kit --skill spec-forge",{"tool":48,"command":49},"bunx","bunx skills add hienpham0344\u002FAgent-kit --skill spec-forge",{"tool":51,"command":52},"pnpm","pnpm dlx skills add hienpham0344\u002FAgent-kit --skill spec-forge","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002FSKILL.md",true,[56,62,68,72,76,80,84,88,92,96,100,104,108,112,116,120,124,128,132,136,140,146,150,154,158,162,166],{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},".gitignore","OTHER","其他",145,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002F.gitignore",{"path":63,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":66,"sourceHtmlUrl":67},"commands\u002Fanalyze.md","REFERENCE","参考文档",2217,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Fanalyze.md",{"path":69,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":70,"sourceHtmlUrl":71},"commands\u002Faudit.md",2021,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Faudit.md",{"path":73,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":74,"sourceHtmlUrl":75},"commands\u002Fdecompose.md",468,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Fdecompose.md",{"path":77,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":78,"sourceHtmlUrl":79},"commands\u002Fidea.md",603,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Fidea.md",{"path":81,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":82,"sourceHtmlUrl":83},"commands\u002Fprd.md",6650,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Fprd.md",{"path":85,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":86,"sourceHtmlUrl":87},"commands\u002Fpropagate.md",3131,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Fpropagate.md",{"path":89,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":90,"sourceHtmlUrl":91},"commands\u002Freview.md",2331,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Freview.md",{"path":93,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":94,"sourceHtmlUrl":95},"commands\u002Fspec-forge.md",26112,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Fspec-forge.md",{"path":97,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":98,"sourceHtmlUrl":99},"commands\u002Fsrs.md",9886,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Fsrs.md",{"path":101,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":102,"sourceHtmlUrl":103},"commands\u002Ftech-design.md",9334,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Ftech-design.md",{"path":105,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":106,"sourceHtmlUrl":107},"commands\u002Ftest-cases.md",10102,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Ftest-cases.md",{"path":109,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":110,"sourceHtmlUrl":111},"commands\u002Ftest-plan.md.deprecated",6751,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fcommands\u002Ftest-plan.md.deprecated",{"path":113,"resourceType":113,"resourceTypeDesc":114,"sizeBytes":115,"sourceHtmlUrl":15},"LICENSE","许可证",1063,{"path":117,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":118,"sourceHtmlUrl":119},"README.md",15813,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002FREADME.md",{"path":121,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":122,"sourceHtmlUrl":123},"RELEASE-NOTES.md",14769,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002FRELEASE-NOTES.md",{"path":125,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":126,"sourceHtmlUrl":127},"skills\u002Fshared\u002Fdoc-first.md",9646,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fdoc-first.md",{"path":129,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":130,"sourceHtmlUrl":131},"skills\u002Fshared\u002Fproject-context.md",5056,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fproject-context.md",{"path":133,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":134,"sourceHtmlUrl":135},"skills\u002Fshared\u002Fscripts.md",3223,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fscripts.md",{"path":137,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":138,"sourceHtmlUrl":139},"skills\u002Fshared\u002Fscripts\u002FREADME.md",5086,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fscripts\u002FREADME.md",{"path":141,"resourceType":142,"resourceTypeDesc":143,"sizeBytes":144,"sourceHtmlUrl":145},"skills\u002Fshared\u002Fscripts\u002Fsf_common.py","SCRIPT","脚本",13305,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fscripts\u002Fsf_common.py",{"path":147,"resourceType":142,"resourceTypeDesc":143,"sizeBytes":148,"sourceHtmlUrl":149},"skills\u002Fshared\u002Fscripts\u002Fsf-config.py",1886,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fscripts\u002Fsf-config.py",{"path":151,"resourceType":142,"resourceTypeDesc":143,"sizeBytes":152,"sourceHtmlUrl":153},"skills\u002Fshared\u002Fscripts\u002Fsf-scan.py",14447,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fscripts\u002Fsf-scan.py",{"path":155,"resourceType":142,"resourceTypeDesc":143,"sizeBytes":156,"sourceHtmlUrl":157},"skills\u002Fshared\u002Fscripts\u002Fsf-trace.py",8057,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fscripts\u002Fsf-trace.py",{"path":159,"resourceType":142,"resourceTypeDesc":143,"sizeBytes":160,"sourceHtmlUrl":161},"skills\u002Fshared\u002Fscripts\u002Fsf-verify-doc.py",8189,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fscripts\u002Fsf-verify-doc.py",{"path":163,"resourceType":142,"resourceTypeDesc":143,"sizeBytes":164,"sourceHtmlUrl":165},"skills\u002Fshared\u002Fscripts\u002Ftest_scripts.py",12488,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002Fskills\u002Fshared\u002Fscripts\u002Ftest_scripts.py",{"path":167,"resourceType":64,"resourceTypeDesc":65,"sizeBytes":168,"sourceHtmlUrl":169},"UPSTREAM.md",129,"https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fspec-forge\u002FUPSTREAM.md",[171,179,188,196,204,212],{"skillNo":172,"name":173,"nameZh":6,"description":174,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":175,"sourceHtmlUrl":176,"categories":177,"keywords":178,"publishedAt":19},"01M7FBMR12AAHSXDNZNDQ492F5","problem-based-srs","Complete Problem-Based Software Requirements Specification methodology following Gorski & Stadzisz research. Use when you need to perform requirements engineering from business problems to functional requirements with full traceability.","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fproblem-based-srs\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fplanning-grilling\u002Fproblem-based-srs\u002FSKILL.md",[],[],{"skillNo":180,"name":181,"nameZh":6,"description":182,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":183,"licenseFileUrl":184,"sourceHtmlUrl":185,"categories":186,"keywords":187,"publishedAt":19},"01M7FBD7K79CF2R0SFYK18M6TG","webapp-testing","Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.","Apache-2.0","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fengineering-core\u002Fwebapp-testing\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fengineering-core\u002Fwebapp-testing\u002FSKILL.md",[],[],{"skillNo":189,"name":190,"nameZh":6,"description":191,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":183,"licenseFileUrl":192,"sourceHtmlUrl":193,"categories":194,"keywords":195,"publishedAt":19},"01M7FBCCE5QWSQ1SG4NXJQX5V1","mcp-builder","Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node\u002FTypeScript (MCP SDK).","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fengineering-core\u002Fmcp-builder\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fengineering-core\u002Fmcp-builder\u002FSKILL.md",[],[],{"skillNo":197,"name":198,"nameZh":6,"description":199,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":183,"licenseFileUrl":200,"sourceHtmlUrl":201,"categories":202,"keywords":203,"publishedAt":19},"01M7FBBDDQNKYEXDGHSEKY94XN","web-artifacts-builder","Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn\u002Fui). Use for complex artifacts requiring state management, routing, or shadcn\u002Fui components - not for simple single-file HTML\u002FJSX artifacts.","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fdocuments-media\u002Fweb-artifacts-builder\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fdocuments-media\u002Fweb-artifacts-builder\u002FSKILL.md",[],[],{"skillNo":205,"name":206,"nameZh":6,"description":207,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":208,"sourceHtmlUrl":209,"categories":210,"keywords":211,"publishedAt":19},"01M7FB8SXSJ8XF8QD5HQHKWJT7","ui-styling","Build accessible user interfaces using shadcn\u002Fui (Radix UI + Tailwind CSS), utility-first styling, dark mode themes, and consistent design system tokens.","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fdesign-uiux\u002Fui-styling\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fdesign-uiux\u002Fui-styling\u002FSKILL.md",[],[],{"skillNo":213,"name":214,"nameZh":6,"description":215,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":216,"categories":217,"keywords":218,"publishedAt":19},"01M7FB7K5DEAQS82EWV9Y7E1NQ","design-system","Token architecture, component specifications, and slide generation. Three-layer tokens (primitive→semantic→component), CSS variables, spacing\u002Ftypography scales, component specs, strategic slide creation. Use for design tokens, systematic design, brand-compliant presentations.","https:\u002F\u002Fgithub.com\u002Fhienpham0344\u002FAgent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fdesign-uiux\u002Fdesign-system\u002FSKILL.md",[],[]]