[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6C46WHZC5DT0SC96XSZ71QH":3,"skill-same:01M6C46WHZC5DT0SC96XSZ71QH":85},{"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":6,"sourceHtmlUrl":15,"categories":16,"keywords":22,"publishedAt":33,"linkOnly":34,"markdownContent":35,"markdownBody":36,"frontMatter":37,"refName":38,"commitSha":39,"skillDirectory":40,"forksCount":41,"repositoryHtmlUrl":42,"repositoryHomepageUrl":6,"ownerHtmlUrl":43,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":44,"install":45,"resources":58},"01M6C46WHZC5DT0SC96XSZ71QH","routing-tables",null,"Concise reference for deterministic specialist dispatch, review mode routing, CE surface mapping, enum values, and gate criteria extracted from current orchestration agents. Use when an agent or script needs the canonical current routing data without re-deriving it from duplicated prompt prose. DO NOT USE FOR: taking orchestration decisions away from the calling agent, replacing judgment on uncovered cases, or storing mutable runtime state.","Grimblaz\u002Fagent-orchestra","Grimblaz","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F74755849?v=4","01M6C430M9XZDHGHPWQVTQ3KMA","01M6C430RJY3MMX0QAZ3X8DR3Y",3,"MIT","https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Frouting-tables\u002FSKILL.md",[17],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[23,24,25,26,27,28,29,30,31,32],"routing","routing tables","review","orchestration","state","data","prompt","decisions","use","dispatch","2026-11-10 10:12:30",false,"---\nname: routing-tables\ndescription: \"Concise reference for deterministic specialist dispatch, review mode routing, CE surface mapping, enum values, and gate criteria extracted from current orchestration agents. Use when an agent or script needs the canonical current routing data without re-deriving it from duplicated prompt prose. DO NOT USE FOR: taking orchestration decisions away from the calling agent, replacing judgment on uncovered cases, or storing mutable runtime state.\"\n---\n\n# Routing Tables\n\nThis skill is the human-readable companion to the routing data in `assets\u002Frouting-config.json` and `assets\u002Fgate-criteria.json`. It centralizes current routing tables and enum values from Code-Conductor and Code-Critic so agents can load concise prompt context while future deterministic consumers read the same data from JSON.\n\n## When to Use\n\n- When an agent needs the canonical current specialist-dispatch, review-mode, CE-surface, enum, or gate data\n- When a deterministic consumer should read the shared routing assets instead of copying inline tables from an agent prompt\n- When reviewing or updating routing metadata and you need one human-readable summary aligned with the JSON assets\n\n## Schema Overview\n\n`assets\u002Frouting-config.json` contains six top-level sections:\n\n- `specialist_dispatch`: file and task routing entries from Code-Conductor's Agent Selection table; deterministic `FilePattern` consumers should use the optional `file_patterns` arrays rather than parsing `file_type_or_task`\n- `review_mode_routing`: selector-line marker to review-mode mapping from Code-Critic, including conflict resolution\n- `surface_identification`: CE Gate surface to tool mapping from Code-Conductor; `No customer surface` carries the exact status marker template in `status_result_template`\n- `skill_mapping`: delegation guidance from Code-Conductor's Skill Mapping table; this is a reference list, not a strict deterministic router\n- `nl_intent_routing`: natural-language intent keys and patterns mapped to Claude and Copilot slash-command surfaces via `claude_command` and `copilot_command`; `null` means no platform equivalent. This supports Phase 1 directive prose and Phase 2 forward-compatible lookup, not runtime hook wiring\n- `enums`: canonical current enum values used by review and routing outputs\n\nFor deterministic consumers, `Invoke-RoutingLookup` preserves the legacy first-match behavior. Use `Invoke-RoutingLookupAll` when a Pattern lookup needs to detect ambiguous natural-language matches and present every matched intent to the user.\n\n`assets\u002Fgate-criteria.json` contains four top-level sections:\n\n- `scope_classification`: abbreviated-vs-full pipeline gate criteria and tier table\n- `express_lane`: low-risk fix gate criteria for routing findings directly to a specialist\n- `review_completion`: pre-PR review-completion gate that requires prosecution, defense, and judgment completion to all be recorded as true for the current review cycle\n- `post_fix_trigger`: conditions that require post-fix targeted prosecution after accepted review fixes\n\n## Specialist Dispatch\n\nSpecialist dispatch stays file- and task-oriented. Test files and fixtures route to Test-Writer. New behavior in `src\u002F**\u002F*.ts` and `src\u002F**\u002F*.tsx` routes to Code-Smith, while restructuring those same source files routes to Refactor-Specialist. PowerShell production scripts (`*.ps1`) also route to Code-Smith. Visual polish work on UI source files routes to UI-Iterator, and documentation files route to Doc-Keeper.\n\nFor deterministic lookups keyed by `FilePattern`, use the structured `file_patterns` arrays in the asset. The prose-bearing `file_type_or_task` field remains the human-readable summary and should not be parsed by consumers.\n\nPlanning artifacts and plan markers route to Issue-Planner. CE Gate evidence capture, customer framing, journeys, and scenarios route to Experience-Owner. Read-only quality review goes to Code-Critic, scored judgment goes to Code-Review-Response, and systemic gap analysis goes to Process-Review.\n\n## Review Mode Routing\n\nReview mode routing is driven by explicit top-level selector lines of the form `Review mode selector: \"{marker}\"`. No selector line means normal code prosecution with five parallel passes. Design review perspectives switch to design or plan prosecution with three parallel passes. Defense, CE prosecution, GitHub proxy prosecution, post-fix code prosecution, and lite code prosecution each run as single-pass modes.\n\nWhen multiple selector lines appear, the conflict rule applies in strict priority order: defense, then CE, then proxy, then design, then post-fix code prosecution, then lite code prosecution, then standard code prosecution. One explicit override exists: `Review mode selector: \"Use code review perspectives\"` beats `Review mode selector: \"Use design review perspectives\"` and forces normal code prosecution.\n\n### Handshake disposition by command\n\nMoved here from the root guidance file (`CLAUDE.md`) by issue #998, which needed the space for the standing statement of what a finished run is true of. The dispositions are unchanged by the move; the root file keeps the one-line summary and points here.\n\n| Command | Handshake |\n| --- | --- |\n| `\u002Forchestra:review` | Required |\n| `\u002Forchestra:review-lite` | Required |\n| `\u002Forchestra:review-prosecute` | Required |\n| `\u002Forchestra:review-defend` | Required |\n| `\u002Forchestra:review-judge` | Optional |\n\n## Surface Identification\n\nCE Gate surface identification maps the customer surface to the expected validation tool. Web UI uses native browser tools first, with Playwright MCP as fallback. REST or GraphQL uses terminal HTTP calls. CLI and SDK surfaces use representative terminal invocations. Batch and pipeline surfaces use representative test data. No customer surface returns the documented status marker `⏭️ CE Gate not applicable — {reason}` from structured data so consumers do not need a separate hardcoded case.\n\n## Skill Mapping\n\nThe skill mapping table is guidance for delegation prompts rather than a pure lookup table. It links each reusable skill to the kind of work where Code-Conductor should explicitly instruct a specialist to load it. The mappings cover review, customer experience, design exploration, documentation, frontend design, implementation discipline, parallel execution, planning, property-based testing, refactoring, research, review judgment, skill creation, software architecture, debugging, TDD, UI iteration, UI testing, and web app testing.\n\nConsumers should treat this section as a reference list that explains when a skill is relevant, not as a hard decision engine that replaces agent judgment.\n\n## Enums\n\nThe enum section captures the canonical current values used by the review pipeline:\n\n- `severity`: `critical`, `high`, `medium`, `low`\n- `points`: `critical` and `high` map to `10`; `medium` maps to `5`; `low` maps to `1`\n- `confidence`: `high`, `medium`, `low`\n- `category`: `architecture`, `security`, `performance`, `pattern`, `implementation-clarity`, `script-automation`, `documentation-audit`\n- `blast_radius`: `localized`, `module`, `cross-module`, `system-wide`\n- `authority_needed`: `yes`, `no`\n- `systemic_fix_type`: `instruction`, `skill`, `agent-prompt`, `plan-template`, `none`\n\nThe category list is canonical-only. Legacy `simplicity` is intentionally excluded.\n\n## Gate Criteria\n\nScope classification is an AND gate: all five abbreviated-tier criteria must hold, otherwise the issue stays on the full pipeline. The stored tier table captures which phases run in full and abbreviated execution.\n\nExpress lane is also an AND gate. A finding must be low severity, strictly mechanical, non-logic-changing, test-neutral, backward-compatible with respect to stored IDs and schema, and limited to one file. It only applies to main review and post-fix review routing, and Tier 1 re-validation is still required after the fix lands.\n\nReview completion is an AND gate. It fails closed unless the current review cycle records all three stage booleans as true: `prosecution_complete`, `defense_complete`, and `judgment_complete`. `review_mode` is tracked alongside the state as label metadata, but it is not part of the gate criteria.\n\nPost-fix targeted prosecution uses OR semantics for its trigger conditions: one accepted critical or high finding is enough, and any accepted fix that modifies control flow is also enough. If no findings were accepted and applied, the post-fix review is skipped.\n\n## Interpretation Rules\n\nUse exact current values from the JSON assets when a mode, enum, or routing value needs to be preserved across prompts or scripts. Match review markers literally. For specialist dispatch file routing, use explicit `file_patterns` keys instead of parsing prose. Treat `skill_mapping` as explanatory metadata, not a replacement for human judgment. Treat `scope_classification`, `express_lane`, and `review_completion` as all-criteria gates, while `post_fix_trigger` activates when any trigger condition is met.\n\nWhen an entry includes prose qualifiers, preserve them rather than compressing them away. The goal of these files is stable extraction of current rules, not reinterpretation of the orchestration policy.\n\n## Fallback Guidance\n\nFuture deterministic consumers should prefer the JSON assets first. If a later script is unavailable, agents can still read the JSON directly. If a case is not covered by the structured data or a prompt needs human-readable context, fall back to this skill summary.\n\nRouting authority remains with the calling agent. This skill provides shared data and summaries; it does not own orchestration decisions.\n\n## Intent Routing Mechanics\n\nPlugin processes are the default chat experience. Natural-language requests matching the `nl_intent_routing` table route to the corresponding slash command with a visible confirmation; `\u002Fraw` opts out.\n\n**Activation order**: (1) VS Code dropdown for VS Code users; (2) slash commands for both platforms; (3) natural-language with auto-routing confirmation; (4) @-mention is NOT recommended (unreliable in every plugin surface tested).\n\nSlash commands diverge between Claude (`commands\u002F*.md`) and Copilot (`.github\u002Fprompts\u002F*.prompt.md`); the `nl_intent_routing` table carries both column names so the canonical command name is platform-portable.\n\n**First match per command-family** per conversation asks explicitly, with options `Run \u002FX for this (Recommended)`, `Continue as raw chat`, and `Don't ask again for this command-family this conversation`; Claude confirmation phrasing should use `Run \u002FX?`. Subsequent same-family matches use inline confirmation: `Routing to \u002FX — say \u002Fraw to opt out, otherwise proceed.`\n\n**Tier hint**: When a proposed command's `model:` frontmatter differs from the user-session model, append a one-line tier hint, e.g. `Will run on sonnet + high per command frontmatter.`\n\n**No-match and disambiguation**: No-match answers normally; first no-match per conversation appends `Tip: type \u002Fhelp for plugin slash commands, or \u002Fraw to suppress these hints.` Ambiguous-match uses a text-only disambiguation prompt, e.g. `Did you mean \u002Forchestra:review (local code) or \u002Freview-github (GitHub PR)?`\n\n## Gotchas\n\n| Trigger                                               | Gotcha                                                                                                                  | Fix                                                                                         |\n| ----------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------- |\n| Treating a routing table entry as binding policy      | The skill centralizes canonical data, but the calling agent still owns the final routing decision and fallback behavior | Use the skill for shared data, then keep decision authority and override logic in the agent |\n| Editing only the prose summary or only the JSON asset | Human-readable guidance and deterministic consumers drift apart                                                         | Update the JSON assets and the corresponding SKILL.md summary together in the same change   |\n","# Routing Tables\n\nThis skill is the human-readable companion to the routing data in `assets\u002Frouting-config.json` and `assets\u002Fgate-criteria.json`. It centralizes current routing tables and enum values from Code-Conductor and Code-Critic so agents can load concise prompt context while future deterministic consumers read the same data from JSON.\n\n## When to Use\n\n- When an agent needs the canonical current specialist-dispatch, review-mode, CE-surface, enum, or gate data\n- When a deterministic consumer should read the shared routing assets instead of copying inline tables from an agent prompt\n- When reviewing or updating routing metadata and you need one human-readable summary aligned with the JSON assets\n\n## Schema Overview\n\n`assets\u002Frouting-config.json` contains six top-level sections:\n\n- `specialist_dispatch`: file and task routing entries from Code-Conductor's Agent Selection table; deterministic `FilePattern` consumers should use the optional `file_patterns` arrays rather than parsing `file_type_or_task`\n- `review_mode_routing`: selector-line marker to review-mode mapping from Code-Critic, including conflict resolution\n- `surface_identification`: CE Gate surface to tool mapping from Code-Conductor; `No customer surface` carries the exact status marker template in `status_result_template`\n- `skill_mapping`: delegation guidance from Code-Conductor's Skill Mapping table; this is a reference list, not a strict deterministic router\n- `nl_intent_routing`: natural-language intent keys and patterns mapped to Claude and Copilot slash-command surfaces via `claude_command` and `copilot_command`; `null` means no platform equivalent. This supports Phase 1 directive prose and Phase 2 forward-compatible lookup, not runtime hook wiring\n- `enums`: canonical current enum values used by review and routing outputs\n\nFor deterministic consumers, `Invoke-RoutingLookup` preserves the legacy first-match behavior. Use `Invoke-RoutingLookupAll` when a Pattern lookup needs to detect ambiguous natural-language matches and present every matched intent to the user.\n\n`assets\u002Fgate-criteria.json` contains four top-level sections:\n\n- `scope_classification`: abbreviated-vs-full pipeline gate criteria and tier table\n- `express_lane`: low-risk fix gate criteria for routing findings directly to a specialist\n- `review_completion`: pre-PR review-completion gate that requires prosecution, defense, and judgment completion to all be recorded as true for the current review cycle\n- `post_fix_trigger`: conditions that require post-fix targeted prosecution after accepted review fixes\n\n## Specialist Dispatch\n\nSpecialist dispatch stays file- and task-oriented. Test files and fixtures route to Test-Writer. New behavior in `src\u002F**\u002F*.ts` and `src\u002F**\u002F*.tsx` routes to Code-Smith, while restructuring those same source files routes to Refactor-Specialist. PowerShell production scripts (`*.ps1`) also route to Code-Smith. Visual polish work on UI source files routes to UI-Iterator, and documentation files route to Doc-Keeper.\n\nFor deterministic lookups keyed by `FilePattern`, use the structured `file_patterns` arrays in the asset. The prose-bearing `file_type_or_task` field remains the human-readable summary and should not be parsed by consumers.\n\nPlanning artifacts and plan markers route to Issue-Planner. CE Gate evidence capture, customer framing, journeys, and scenarios route to Experience-Owner. Read-only quality review goes to Code-Critic, scored judgment goes to Code-Review-Response, and systemic gap analysis goes to Process-Review.\n\n## Review Mode Routing\n\nReview mode routing is driven by explicit top-level selector lines of the form `Review mode selector: \"{marker}\"`. No selector line means normal code prosecution with five parallel passes. Design review perspectives switch to design or plan prosecution with three parallel passes. Defense, CE prosecution, GitHub proxy prosecution, post-fix code prosecution, and lite code prosecution each run as single-pass modes.\n\nWhen multiple selector lines appear, the conflict rule applies in strict priority order: defense, then CE, then proxy, then design, then post-fix code prosecution, then lite code prosecution, then standard code prosecution. One explicit override exists: `Review mode selector: \"Use code review perspectives\"` beats `Review mode selector: \"Use design review perspectives\"` and forces normal code prosecution.\n\n### Handshake disposition by command\n\nMoved here from the root guidance file (`CLAUDE.md`) by issue #998, which needed the space for the standing statement of what a finished run is true of. The dispositions are unchanged by the move; the root file keeps the one-line summary and points here.\n\n| Command | Handshake |\n| --- | --- |\n| `\u002Forchestra:review` | Required |\n| `\u002Forchestra:review-lite` | Required |\n| `\u002Forchestra:review-prosecute` | Required |\n| `\u002Forchestra:review-defend` | Required |\n| `\u002Forchestra:review-judge` | Optional |\n\n## Surface Identification\n\nCE Gate surface identification maps the customer surface to the expected validation tool. Web UI uses native browser tools first, with Playwright MCP as fallback. REST or GraphQL uses terminal HTTP calls. CLI and SDK surfaces use representative terminal invocations. Batch and pipeline surfaces use representative test data. No customer surface returns the documented status marker `⏭️ CE Gate not applicable — {reason}` from structured data so consumers do not need a separate hardcoded case.\n\n## Skill Mapping\n\nThe skill mapping table is guidance for delegation prompts rather than a pure lookup table. It links each reusable skill to the kind of work where Code-Conductor should explicitly instruct a specialist to load it. The mappings cover review, customer experience, design exploration, documentation, frontend design, implementation discipline, parallel execution, planning, property-based testing, refactoring, research, review judgment, skill creation, software architecture, debugging, TDD, UI iteration, UI testing, and web app testing.\n\nConsumers should treat this section as a reference list that explains when a skill is relevant, not as a hard decision engine that replaces agent judgment.\n\n## Enums\n\nThe enum section captures the canonical current values used by the review pipeline:\n\n- `severity`: `critical`, `high`, `medium`, `low`\n- `points`: `critical` and `high` map to `10`; `medium` maps to `5`; `low` maps to `1`\n- `confidence`: `high`, `medium`, `low`\n- `category`: `architecture`, `security`, `performance`, `pattern`, `implementation-clarity`, `script-automation`, `documentation-audit`\n- `blast_radius`: `localized`, `module`, `cross-module`, `system-wide`\n- `authority_needed`: `yes`, `no`\n- `systemic_fix_type`: `instruction`, `skill`, `agent-prompt`, `plan-template`, `none`\n\nThe category list is canonical-only. Legacy `simplicity` is intentionally excluded.\n\n## Gate Criteria\n\nScope classification is an AND gate: all five abbreviated-tier criteria must hold, otherwise the issue stays on the full pipeline. The stored tier table captures which phases run in full and abbreviated execution.\n\nExpress lane is also an AND gate. A finding must be low severity, strictly mechanical, non-logic-changing, test-neutral, backward-compatible with respect to stored IDs and schema, and limited to one file. It only applies to main review and post-fix review routing, and Tier 1 re-validation is still required after the fix lands.\n\nReview completion is an AND gate. It fails closed unless the current review cycle records all three stage booleans as true: `prosecution_complete`, `defense_complete`, and `judgment_complete`. `review_mode` is tracked alongside the state as label metadata, but it is not part of the gate criteria.\n\nPost-fix targeted prosecution uses OR semantics for its trigger conditions: one accepted critical or high finding is enough, and any accepted fix that modifies control flow is also enough. If no findings were accepted and applied, the post-fix review is skipped.\n\n## Interpretation Rules\n\nUse exact current values from the JSON assets when a mode, enum, or routing value needs to be preserved across prompts or scripts. Match review markers literally. For specialist dispatch file routing, use explicit `file_patterns` keys instead of parsing prose. Treat `skill_mapping` as explanatory metadata, not a replacement for human judgment. Treat `scope_classification`, `express_lane`, and `review_completion` as all-criteria gates, while `post_fix_trigger` activates when any trigger condition is met.\n\nWhen an entry includes prose qualifiers, preserve them rather than compressing them away. The goal of these files is stable extraction of current rules, not reinterpretation of the orchestration policy.\n\n## Fallback Guidance\n\nFuture deterministic consumers should prefer the JSON assets first. If a later script is unavailable, agents can still read the JSON directly. If a case is not covered by the structured data or a prompt needs human-readable context, fall back to this skill summary.\n\nRouting authority remains with the calling agent. This skill provides shared data and summaries; it does not own orchestration decisions.\n\n## Intent Routing Mechanics\n\nPlugin processes are the default chat experience. Natural-language requests matching the `nl_intent_routing` table route to the corresponding slash command with a visible confirmation; `\u002Fraw` opts out.\n\n**Activation order**: (1) VS Code dropdown for VS Code users; (2) slash commands for both platforms; (3) natural-language with auto-routing confirmation; (4) @-mention is NOT recommended (unreliable in every plugin surface tested).\n\nSlash commands diverge between Claude (`commands\u002F*.md`) and Copilot (`.github\u002Fprompts\u002F*.prompt.md`); the `nl_intent_routing` table carries both column names so the canonical command name is platform-portable.\n\n**First match per command-family** per conversation asks explicitly, with options `Run \u002FX for this (Recommended)`, `Continue as raw chat`, and `Don't ask again for this command-family this conversation`; Claude confirmation phrasing should use `Run \u002FX?`. Subsequent same-family matches use inline confirmation: `Routing to \u002FX — say \u002Fraw to opt out, otherwise proceed.`\n\n**Tier hint**: When a proposed command's `model:` frontmatter differs from the user-session model, append a one-line tier hint, e.g. `Will run on sonnet + high per command frontmatter.`\n\n**No-match and disambiguation**: No-match answers normally; first no-match per conversation appends `Tip: type \u002Fhelp for plugin slash commands, or \u002Fraw to suppress these hints.` Ambiguous-match uses a text-only disambiguation prompt, e.g. `Did you mean \u002Forchestra:review (local code) or \u002Freview-github (GitHub PR)?`\n\n## Gotchas\n\n| Trigger                                               | Gotcha                                                                                                                  | Fix                                                                                         |\n| ----------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------- |\n| Treating a routing table entry as binding policy      | The skill centralizes canonical data, but the calling agent still owns the final routing decision and fallback behavior | Use the skill for shared data, then keep decision authority and override logic in the agent |\n| Editing only the prose summary or only the JSON asset | Human-readable guidance and deterministic consumers drift apart                                                         | Update the JSON assets and the corresponding SKILL.md summary together in the same change   |\n",{"name":5,"description":7},"main","d5486aacc32c9ca1b07173cfd530f8fe75c29e58","skills\u002Frouting-tables",0,"https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra","https:\u002F\u002Fgithub.com\u002FGrimblaz","2026-08-16 01:12:46",{"source":8,"skillName":5,"commands":46,"prompt":56,"zipAvailable":57},[47,50,53],{"tool":48,"command":49},"npx","npx skills add Grimblaz\u002Fagent-orchestra --skill routing-tables",{"tool":51,"command":52},"bunx","bunx skills add Grimblaz\u002Fagent-orchestra --skill routing-tables",{"tool":54,"command":55},"pnpm","pnpm dlx skills add Grimblaz\u002Fagent-orchestra --skill routing-tables","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Frouting-tables\u002FSKILL.md",true,[59,65,69,75,81],{"path":60,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":63,"sourceHtmlUrl":64},"assets\u002Fgate-criteria.json","ASSET","资产文件",6117,"https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Frouting-tables\u002Fassets\u002Fgate-criteria.json",{"path":66,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":67,"sourceHtmlUrl":68},"assets\u002Frouting-config.json",14437,"https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Frouting-tables\u002Fassets\u002Frouting-config.json",{"path":70,"resourceType":71,"resourceTypeDesc":72,"sizeBytes":73,"sourceHtmlUrl":74},"references\u002Fmulti-issue-bundling.md","REFERENCE","参考文档",2485,"https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Frouting-tables\u002Freferences\u002Fmulti-issue-bundling.md",{"path":76,"resourceType":77,"resourceTypeDesc":78,"sizeBytes":79,"sourceHtmlUrl":80},"scripts\u002Freview-state-reader.ps1","SCRIPT","脚本",3730,"https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Frouting-tables\u002Fscripts\u002Freview-state-reader.ps1",{"path":82,"resourceType":77,"resourceTypeDesc":78,"sizeBytes":83,"sourceHtmlUrl":84},"scripts\u002Frouting-tables-core.ps1",10576,"https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Frouting-tables\u002Fscripts\u002Frouting-tables-core.ps1",[86,118,142,156,182,203],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":98,"publishedAt":33},"01M6C48YMJVFHM7DVGG1WWW5RJ","webapp-testing","Playwright end-to-end testing guidance for web apps, with practical patterns and setup steps. Use when creating or improving browser-based E2E coverage, test stability, and CI execution. DO NOT USE FOR: React component tests (use ui-testing), canvas game interaction (use browser-canvas-testing), TDD workflow (use test-driven-development), debugging failures (use systematic-debugging), or randomized property tests (use property-based-testing).","https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Fwebapp-testing\u002FSKILL.md",[92,95],{"categoryNo":93,"categoryName":94,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000008","测试与质量",{"categoryNo":96,"categoryName":97,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000009","前端开发",[99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,31],"testing","webapp testing","webapp","canvas","tdd","workflow","playwright","test","web","react","component","debugging","development","browser","ci","execution","e2e","game","coverage",{"skillNo":119,"name":120,"nameZh":6,"description":121,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":122,"categories":123,"keywords":124,"publishedAt":33},"01M6C48TNNP2KCSQXV8YK7SK49","verification-before-completion","Evidence-based verification checklist before marking work complete. Use before PRs, releases, marking tickets done, or any \"I'm finished\" declaration; also when naming a baseline commit for a differential suite claim, when designing a mutation campaign or red-run evidence, when making an absence claim, a universal claim, or a one-time demonstration offered as a standing guarantee, when attributing a measurement or claiming a sweep or verification loop is complete, and when stating what a check or a CI suite will do about a change. DO NOT USE FOR: post-merge cleanup or archival (use post-pr-review) or processing GitHub review comments (use code-review-intake).","https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Fverification-before-completion\u002FSKILL.md",[],[125,126,25,127,128,129,130,131,132,133,134,135,113,136,137,138,139,31,140,141],"verification","verification before completion","change","commit","merge","pr","cleanup","run","code","code-review","intake","loop","verification loop","work","check","evidence","campaign",{"skillNo":143,"name":144,"nameZh":6,"description":145,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":146,"categories":147,"keywords":148,"publishedAt":33},"01M6C48QZWNTP7JGKD32QZAKKX","validation-methodology","Reusable validation and review methodology for staged validation, failure triage, and prosecution-depth setup. Use when running validation ladders, triaging failures, or executing adversarial review passes. DO NOT USE FOR: CE Gate orchestration, specialist dispatch ownership, or step execution flow (keep those in Code-Conductor).","https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Fvalidation-methodology\u002FSKILL.md",[],[149,150,151,25,152,26,153,154,133,155,114,31,32],"validation","methodology","validation methodology","triage","setup","adversarial review","flow",{"skillNo":157,"name":158,"nameZh":6,"description":159,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":160,"categories":161,"keywords":168,"publishedAt":33},"01M6C48MCAHG5Z79BJ369CRYRZ","upstream-onboarding","Shared opening-phase protocol for upstream agents (Experience-Owner, Solution-Designer, Issue-Planner) and Code-Conductor when invoked on an existing GitHub issue. Renders a scaled context brief and runs a standards check on inherited work at each phase boundary. Use when a user-invocable agent receives a request referencing an existing GitHub issue and must load solution-authoring and upstream-onboarding at the opening phase. DO NOT USE FOR: subagent dispatches (which already operate within an assessed session context); post-merge review (use post-pr-review); research or non-tree subagents.","https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Fupstream-onboarding\u002FSKILL.md",[162,163],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},{"categoryNo":164,"categoryName":165,"taxonomyType":166,"taxonomyTypeDesc":167},"01JBSEEDCAT000000000000038","人力资源","PROFESSION","适用职业",[169,170,25,171,172,129,173,130,133,174,175,176,177,178,179,138,139,180,181,31],"onboarding","upstream onboarding","github","research","standards check","protocol","brief","boundary","designer","subagent","planner","user","session",{"skillNo":183,"name":184,"nameZh":6,"description":185,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":186,"categories":187,"keywords":196,"publishedAt":33},"01M6C48GPVE120TGPMD1CS365H","ui-testing","Resilient React component testing strategies focusing on user behavior. Use when writing or reviewing component-level React tests, fixing flaky component tests, or establishing React testing patterns. DO NOT USE FOR: Playwright E2E tests (use webapp-testing), canvas game interaction (use browser-canvas-testing), TDD workflow (use test-driven-development), debugging failures (use systematic-debugging), UI component design (use frontend-design), or randomized property tests (use property-based-testing).","https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Fui-testing\u002FSKILL.md",[188,189,190,193],{"categoryNo":93,"categoryName":94,"taxonomyType":20,"taxonomyTypeDesc":21},{"categoryNo":96,"categoryName":97,"taxonomyType":20,"taxonomyTypeDesc":21},{"categoryNo":191,"categoryName":192,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":194,"categoryName":195,"taxonomyType":166,"taxonomyTypeDesc":167},"01JBSEEDCAT000000000000032","设计师",[99,197,198,199,200,201,102,104,105,106,108,109,110,111,112,202,115,116,31,101],"ui","ui testing","frontend","design","writing","testing patterns",{"skillNo":204,"name":205,"nameZh":6,"description":206,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":207,"categories":208,"keywords":211,"publishedAt":33},"01M6C48DRJ9XBDR2FW0W37DY53","ui-iteration","Reusable screenshot-driven UI polish workflow for iterative visual refinement. Use when improving layout, hierarchy, spacing, feedback, or Tailwind-based presentation through repeated review-and-adjust passes. DO NOT USE FOR: functional bug fixing, accessibility-only audits, or major redesign strategy work (use Code-Smith or frontend-design).","https:\u002F\u002Fgithub.com\u002FGrimblaz\u002Fagent-orchestra\u002Fblob\u002Fmain\u002Fskills\u002Fui-iteration\u002FSKILL.md",[209,210],{"categoryNo":96,"categoryName":97,"taxonomyType":20,"taxonomyTypeDesc":21},{"categoryNo":194,"categoryName":195,"taxonomyType":166,"taxonomyTypeDesc":167},[197,212,199,200,25,104,213,133,214,215,216,217,218,138,219,220,31,221],"ui iteration","accessibility","screenshot","Tailwind","polish","ui polish","feedback","redesign","strategy","only"]