[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75BCACCJ2AB3XN6JWA14CZN":3,"skill-same:01M75BCACCJ2AB3XN6JWA14CZN":87},{"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":43,"commitSha":44,"skillDirectory":45,"forksCount":14,"repositoryHtmlUrl":46,"repositoryHomepageUrl":6,"ownerHtmlUrl":47,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":48,"install":49,"resources":62},"01M75BCACCJ2AB3XN6JWA14CZN","topic-channel-soul-routing",null,"Use when configuring Telegram topics, Discord channels\u002Fthreads, or other messaging surfaces so each one has a specialist agent soul\u002Fprofile while keeping normal conversation separate from explicit Kanban\u002Ftask intake.","fmercurio\u002Fclawdia-custom-skills","fmercurio","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F15571697?v=4","01M75B9TGVV3RPB40SY7AT0VNH","Felippe Mercurio","01M75B9THMAWYGFT755CR1PT9A",1,"MIT","https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Fskills\u002Fproductivity\u002Ftopic-channel-soul-routing\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: topic-channel-soul-routing\ndescription: \"Use when configuring Telegram topics, Discord channels\u002Fthreads, or other messaging surfaces so each one has a specialist agent soul\u002Fprofile while keeping normal conversation separate from explicit Kanban\u002Ftask intake.\"\nversion: 1.0.0\nauthor: Skills Lab\nlicense: MIT\nplatforms: [linux, macos]\nmetadata:\n  hermes:\n    tags: [topic-routing, profiles, souls, telegram, discord, kanban, tenant-onboarding]\n    related_skills: [hermes-agent, github-operations, software-development-workflows]\n---\n\n# Topic\u002FChannel Soul Routing\n\n## Overview\n\nThis skill turns a messaging workspace into a set of specialist agent surfaces: a Telegram forum topic, Discord channel, Discord thread, or equivalent surface can have its own soul\u002Fprofile, behavior contract, and default task board.\n\nThe core rule is **conversation first, explicit tasks only**. A configured topic\u002Fchannel should answer in the specialist agent's style for normal messages. It should create Kanban\u002Ftask items only when the user explicitly asks with prefixes such as `task:`, `tarefa:`, `kanban:`, `\u002Ftask`, or `\u002Fkanban`.\n\nUse this skill as a reusable implementation\u002Frunbook for any agent runtime that supports platform metadata and profile\u002Fsoul prompt injection.\n\n## When to Use\n\nUse when:\n\n- A user wants a Telegram topic or Discord channel to behave like a specialist agent.\n- A tenant needs separate agents for Research, Product, Admin, Support, Engineering, Finance, etc.\n- You are designing profile\u002Fsoul routing, topic-specific prompts, or channel-specific behavior.\n- You need to preserve conversational behavior while supporting explicit Kanban intake.\n- You are onboarding a non-technical operator who should ask their own agent to configure this.\n\nDo **not** use for:\n\n- A pure task inbox where every message should become a task.\n- One-off persona changes inside a single conversation.\n- Importing or executing unreviewed external skills.\n- Storing secrets, tokens, API keys, or credentials in SOUL\u002Fprofile files.\n\n## Customization Checklist\n\nBefore applying this skill, fill in or discover these values:\n\n| Placeholder | Meaning | Example |\n|---|---|---|\n| `\u003CPLATFORM>` | Messaging platform | `telegram`, `discord` |\n| `\u003CTENANT_KEY>` | Runtime tenant\u002Faccount\u002Fproject scope | `tenant-alpha` |\n| `\u003CENVIRONMENT>` | Runtime environment | `dev`, `staging`, `prod` |\n| `\u003CWORKSPACE_NAME>` | Human name of group\u002Fserver\u002Fworkspace | `Acme HQ` |\n| `\u003CWORKSPACE_ID>` | Stable platform workspace id | `-1001234567890` |\n| `\u003CSURFACE_NAME>` | Topic\u002Fchannel\u002Fthread name | `Research` |\n| `\u003CSURFACE_ID>` | Stable topic\u002Fchannel\u002Fthread id | `525` |\n| `\u003CSURFACE_TYPE>` | Surface kind | `topic`, `channel`, `thread` |\n| `\u003CPROJECT_KEY>` | Internal project key | `research` |\n| `\u003CBOARD_KEY>` | Kanban\u002Ftask board | `research` |\n| `\u003CPROFILE_NAME>` | Runtime profile\u002Fassignee | `researcher` |\n| `\u003CSOUL_ID>` | Persona\u002Fsoul identifier | `researcher` |\n| `\u003CSOUL_PATH>` | Optional SOUL\u002Fprofile file path | `~\u002F.hermes\u002Fprofiles\u002Fresearcher\u002FSOUL.md` |\n\nKeep company-specific values in deployment config, not inside the generic skill.\nTreat real tenant keys, workspace IDs, surface IDs, user IDs, and message links as operational metadata. Use placeholders in reusable docs, examples, tickets, screenshots, and public reports.\n\n## Operating Principles\n\n1. **Conversation first** — a mapped topic\u002Fchannel is still a place to talk.\n2. **Explicit task intake** — create tasks only for explicit prefixes or commands.\n3. **Platform-neutral registry** — represent Telegram and Discord with the same conceptual fields.\n4. **Soul\u002Fprofile separation** — soul describes identity\u002Fbehavior; profile describes runtime\u002Ftool\u002Fmemory isolation when available.\n5. **Read-only discovery before mutation** — inspect current config, profiles, boards, and behavior before edits.\n6. **No secrets in souls** — SOULs, registry entries, templates, and examples must never contain credentials.\n7. **Human confirmation before applying** — show a summary and ask the operator to confirm before writing config or restarting runtime.\n8. **Tenant-scoped routing** — in multi-tenant runtimes, match routes by tenant\u002Faccount\u002Fproject scope plus platform\u002Fworkspace\u002Fsurface IDs, never by surface name alone.\n9. **Minimal source references** — task creation should store a compact source reference, not raw message metadata, transcripts, permalinks, or user profile data by default.\n10. **Policy before prompt** — route lookup, authorization, and task-prefix detection are policy decisions; a SOUL\u002Fprofile prompt must not grant itself tools, secrets, broader route scope, or automatic task intake.\n\n## Data Model\n\nUse a platform-neutral registry. The actual filename depends on the runtime; common examples include `topic_souls.yaml`, `channel_souls.yaml`, or a section inside a larger routing config.\n\n```yaml\ntopic_souls:\n  version: \"1\"\n  entries:\n    - tenant_key: \"\u003CTENANT_KEY>\"\n      environment: \"\u003CENVIRONMENT>\"\n      platform: \"\u003CPLATFORM>\"\n      workspace_id: \"\u003CWORKSPACE_ID>\"\n      surface_id: \"\u003CSURFACE_ID>\"\n      surface_type: \"\u003CSURFACE_TYPE>\"\n      surface_name: \"\u003CSURFACE_NAME>\"\n      project: \"\u003CPROJECT_KEY>\"\n      board: \"\u003CBOARD_KEY>\"\n      profile: \"\u003CPROFILE_NAME>\"\n      soul_id: \"\u003CSOUL_ID>\"\n      mode: direct\n      kanban_intake: explicit\n      source_reference_policy: minimal\n      include_permalink: false\n      actor_reference: hashed\n      status: draft\n      activation:\n        requires_human_confirmation: true\n        approved_by: null\n        approved_at: null\n      prompt_summary: >-\n        Short description of what this specialist agent does, how it answers,\n        and what it must not do without approval.\n```\n\nRecommended `status` values:\n\n| Status | Meaning |\n|---|---|\n| `draft` | Collected but not active; missing IDs or approval |\n| `active` | Runtime may route matching messages to this soul\u002Fprofile |\n| `disabled` | Preserved for history but ignored by runtime |\n\n`active` entries must have a complete tenant\u002Fscope key when the runtime serves more than one tenant, account, project, or customer. If the runtime is intentionally single-tenant, document that assumption in the deployment config.\n\n## Wizard Workflow\n\nRun this as a guided wizard. Ask one question at a time unless the user explicitly requests bulk\u002FYAML mode.\n\n### 1. Explain the capability\n\nUse operator-facing wording:\n\n```text\nVamos configurar um agente especialista para um tópico\u002Fcanal.\n\nMensagens normais continuam sendo conversa. Tarefas só entram no Kanban quando alguém usar um prefixo explícito como `task:`, `tarefa:`, `kanban:`, `\u002Ftask` ou `\u002Fkanban`.\n```\n\n### 2. Choose platform\u002Fworkspace\n\nAsk:\n\n```text\nQual plataforma e workspace\u002Fgrupo\u002Fservidor você quer configurar?\nSe souber o ID técnico, envie também. Se não souber, posso orientar como descobrir.\n```\n\nCapture `platform`, `workspace_name`, and `workspace_id`. If the ID is unknown, continue in `draft` mode.\n\nDo not paste real workspace IDs, private channel IDs, or message links into shared issue trackers or reusable examples. Keep them in the runtime configuration surface with the same access controls as other operational metadata.\n\n### 3. Choose surfaces\n\nAsk:\n\n```text\nQuais tópicos\u002Fcanais\u002Fthreads devem ter agentes especialistas?\nPode mandar uma lista com nomes humanos primeiro; depois descobrimos os IDs estáveis.\n```\n\nCapture `surface_name`, `surface_type`, and `surface_id` when available.\n\n### 4. Choose or create soul\u002Fprofile\n\nFor each surface, ask:\n\n```text\nPara \"\u003CSURFACE_NAME>\", você quer usar uma soul\u002Fprofile existente ou criar uma nova especialista?\n```\n\nIf creating a new specialist, collect:\n\n- short name;\n- role\u002Ffunction;\n- domain expertise;\n- tone;\n- response style;\n- boundaries\u002Fapproval rules;\n- default board\u002Fprofile for explicit tasks;\n- examples of conversation vs task messages.\n\n### 5. Generate a SOUL\u002Fprofile prompt\n\nUse this structure:\n\n```markdown\n# \u003CSOUL_ID> SOUL\n\n## Role\n\u003CWhat this specialist agent is responsible for.>\n\n## Domain\n\u003CSubjects and decisions this agent is strong at.>\n\n## Tone\n\u003CDirect, consultative, technical, didactic, critical, creative, etc.>\n\n## Response Style\n\u003CHow normal replies should be shaped.>\n\n## Boundaries\n\u003CWhat requires explicit approval.>\n\n## Task Intake\nNormal messages are conversation. Create tasks only for explicit prefixes:\n`task:`, `tarefa:`, `kanban:`, `\u002Ftask`, `\u002Fkanban`, `criar task:`, `criar tarefa:`.\n\nDefault board: `\u003CBOARD_KEY>`.\nDefault profile\u002Fassignee: `\u003CPROFILE_NAME>`.\n```\n\n### 6. Confirm behavior per surface\n\nRecommended default:\n\n```yaml\nmode: direct\nkanban_intake: explicit\n```\n\nOnly use automatic task creation if the operator explicitly says the surface is intake-only.\n\n### 7. Review before writing\n\nShow a summary and ask for confirmation:\n\n```text\nResumo antes de aplicar:\n- Platform: \u003CPLATFORM>\n- Workspace: \u003CWORKSPACE_NAME> (\u003CWORKSPACE_ID>)\n- Surface: \u003CSURFACE_NAME> \u002F \u003CSURFACE_ID>\n- Soul: \u003CSOUL_ID>\n- Profile: \u003CPROFILE_NAME>\n- Board: \u003CBOARD_KEY>\n- Behavior: conversa por padrão; Kanban só explícito\n\nConfirmar aplicação?\n```\n\nDo not write config until the operator confirms.\n\n### 8. Apply safely\n\nAfter confirmation:\n\n1. Back up the current routing\u002Fprofile config.\n2. Write or patch registry entries.\n3. Create or update SOUL\u002Fprofile files if the runtime supports them.\n4. Validate every `active` entry has platform, workspace id, surface id, profile\u002Fsoul, and board if task intake is enabled.\n5. Run a dry-run resolver or unit tests if available.\n6. Ask before restarting gateway\u002Fbot\u002Fruntime.\n\n## Runtime Behavior Contract\n\n### Route resolution and policy order\n\nThe runtime should resolve policy before loading surface-specific prompt text:\n\n```text\nNormalize tenant\u002Faccount scope\n+ normalize platform\u002Fworkspace\u002Fsurface IDs\n+ match active route by tenant scope + platform + workspace_id + surface_id\n+ verify route activation and operator approval\n+ detect explicit task prefix\n+ load base runtime prompt\n+ load configured SOUL\u002Fprofile as behavior context only\n```\n\nThe SOUL\u002Fprofile can shape tone, domain focus, and response format. It must not expand route scope, grant tools, bypass tenant isolation, read secrets, switch task intake to automatic, or override human-approval requirements.\n\n### Normal message\n\nInput:\n\n```text\nQual é sua função neste tópico\u002Fcanal?\n```\n\nExpected behavior:\n\n```text\nLoad base runtime prompt\n+ load platform\u002Fchat context\n+ load surface-specific route\n+ load configured soul\u002Fprofile prompt as behavior context only\n+ answer conversationally\n+ do not create a Kanban task\n```\n\n### Explicit task message\n\nInput:\n\n```text\ntask: pesquisar benchmarks para agentes autônomos e documentar recomendações\n```\n\nExpected behavior:\n\n```text\nCreate task on configured board\nAssign configured profile\u002Fagent\nInclude minimal surface context in title\u002Fbody\nReturn confirmation with board, task id, profile, and dashboard link when available\n```\n\nStore a minimal source reference with the task:\n\n```yaml\nsource_ref:\n  platform: \"\u003CPLATFORM>\"\n  tenant_key: \"\u003CTENANT_KEY>\"\n  workspace_id: \"\u003CWORKSPACE_ID>\"\n  surface_id: \"\u003CSURFACE_ID>\"\n  message_id: \"\u003CMESSAGE_ID>\"\n  actor_ref: \"\u003CHASHED_OR_INTERNAL_ACTOR_REF>\"\n```\n\nDo not include raw message metadata, user profile data, transcripts, or private permalinks unless the deployment has an explicit retention and access-control policy for that data.\n\n### Supported explicit prefixes\n\nSupport at least:\n\n```text\n\u002Ftask ...\n\u002Fkanban ...\ntask: ...\nkanban: ...\ntarefa: ...\ncriar task: ...\ncriar tarefa: ...\n```\n\nDo not trigger tasks from vague phrases such as “precisamos”, “seria bom”, “vamos fazer”, or “alguém deveria”.\n\n## Validation Checklist\n\n- [ ] Existing config, profiles\u002Fsouls, boards, and platform metadata inspected before mutation.\n- [ ] Multi-tenant runtimes include tenant\u002Faccount\u002Fproject scope in route matching.\n- [ ] Every active route has stable workspace and surface IDs.\n- [ ] Active routes require operator approval and cannot be activated by prompt text alone.\n- [ ] Normal message test answers conversationally and creates no task.\n- [ ] Explicit `task:` test creates one task on the configured board\u002Fprofile.\n- [ ] Disabled\u002Fdraft entries are ignored by runtime.\n- [ ] Wrong workspace\u002Fsurface does not match a route.\n- [ ] Wrong tenant\u002Faccount\u002Fproject scope does not match a route.\n- [ ] Task source references are minimized and do not store raw private metadata by default.\n- [ ] SOUL\u002Fprofile prompt cannot grant tools, secrets, route scope, or automatic intake.\n- [ ] SOUL\u002Fprofile files contain no secrets.\n- [ ] Runtime restart, if needed, was explicitly approved.\n- [ ] Operator-facing README or handoff has been provided for non-technical users.\n\n## Common Pitfalls\n\n1. **Routing every message into Kanban.** Specialist surfaces are conversation-first unless explicitly configured as intake-only.\n2. **Using human names as the only routing key.** Always discover stable platform IDs before activating routes.\n3. **Skipping tenant scope.** Workspace IDs can be stable platform identifiers, but they are not tenant authorization boundaries.\n4. **Copying private metadata into tasks.** Store compact source references by default; avoid raw permalinks, transcripts, and user profile data.\n5. **Letting prompt text change policy.** SOUL\u002Fprofile text cannot grant tools, secrets, route scope, or automatic task creation.\n6. **Mixing secrets into SOUL files.** Store credentials in the runtime’s secret manager\u002Fenv\u002FKeePass, never in prompts.\n7. **Skipping confirmation.** A wizard can draft config, but it must not apply active routes without operator confirmation.\n8. **Restarting the bot\u002Fgateway too early.** Validate config and ask first; many runtimes can dry-run route resolution without restart.\n9. **Duplicating every global skill into every profile.** Prefer profile overlays only for exceptions; use global skill discovery\u002Ffallback when available.\n10. **Forgetting operator language.** Product UX should say “configure an agent for this channel”, not “edit SOUL.md and topic_routes”.\n\n## One-Shot Handoff Prompt\n\nUse when asking another implementation agent to configure this capability:\n\n```text\nConfigure a specialist agent for each selected Telegram topic, Discord channel\u002Fthread, or equivalent messaging surface.\n\nNormal messages must remain conversational. Create Kanban\u002Ftasks only when the user starts with `\u002Ftask`, `\u002Fkanban`, `task:`, `kanban:`, `tarefa:`, `criar task:` or `criar tarefa:`.\n\nFirst run a wizard with the operator:\n1. choose platform\u002Fworkspace;\n2. collect surface names and stable IDs;\n3. choose existing soul\u002Fprofile or create a new specialist;\n4. define role, domain, tone, response style, boundaries, default board\u002Fprofile;\n5. show a summary and ask for confirmation before writing config.\n\nAfter confirmation, create\u002Fupdate a platform-neutral route registry with tenant\u002Faccount scope when applicable, platform, workspace_id, surface_id, surface_type, surface_name, project, board, profile, soul_id, mode=direct, kanban_intake=explicit, status, source_reference_policy=minimal, and human activation metadata.\n\nValidate with two live tests per surface:\n- `Qual é sua função neste tópico\u002Fcanal?` must answer conversationally and create no task.\n- `task: teste de criação explícita` must create one task on the configured board\u002Fprofile.\n\nDo not restart runtime\u002Fgateway unless the operator explicitly approves. Do not store secrets, raw private message metadata, or public report examples with real IDs in souls or registry.\n```\n","# Topic\u002FChannel Soul Routing\n\n## Overview\n\nThis skill turns a messaging workspace into a set of specialist agent surfaces: a Telegram forum topic, Discord channel, Discord thread, or equivalent surface can have its own soul\u002Fprofile, behavior contract, and default task board.\n\nThe core rule is **conversation first, explicit tasks only**. A configured topic\u002Fchannel should answer in the specialist agent's style for normal messages. It should create Kanban\u002Ftask items only when the user explicitly asks with prefixes such as `task:`, `tarefa:`, `kanban:`, `\u002Ftask`, or `\u002Fkanban`.\n\nUse this skill as a reusable implementation\u002Frunbook for any agent runtime that supports platform metadata and profile\u002Fsoul prompt injection.\n\n## When to Use\n\nUse when:\n\n- A user wants a Telegram topic or Discord channel to behave like a specialist agent.\n- A tenant needs separate agents for Research, Product, Admin, Support, Engineering, Finance, etc.\n- You are designing profile\u002Fsoul routing, topic-specific prompts, or channel-specific behavior.\n- You need to preserve conversational behavior while supporting explicit Kanban intake.\n- You are onboarding a non-technical operator who should ask their own agent to configure this.\n\nDo **not** use for:\n\n- A pure task inbox where every message should become a task.\n- One-off persona changes inside a single conversation.\n- Importing or executing unreviewed external skills.\n- Storing secrets, tokens, API keys, or credentials in SOUL\u002Fprofile files.\n\n## Customization Checklist\n\nBefore applying this skill, fill in or discover these values:\n\n| Placeholder | Meaning | Example |\n|---|---|---|\n| `\u003CPLATFORM>` | Messaging platform | `telegram`, `discord` |\n| `\u003CTENANT_KEY>` | Runtime tenant\u002Faccount\u002Fproject scope | `tenant-alpha` |\n| `\u003CENVIRONMENT>` | Runtime environment | `dev`, `staging`, `prod` |\n| `\u003CWORKSPACE_NAME>` | Human name of group\u002Fserver\u002Fworkspace | `Acme HQ` |\n| `\u003CWORKSPACE_ID>` | Stable platform workspace id | `-1001234567890` |\n| `\u003CSURFACE_NAME>` | Topic\u002Fchannel\u002Fthread name | `Research` |\n| `\u003CSURFACE_ID>` | Stable topic\u002Fchannel\u002Fthread id | `525` |\n| `\u003CSURFACE_TYPE>` | Surface kind | `topic`, `channel`, `thread` |\n| `\u003CPROJECT_KEY>` | Internal project key | `research` |\n| `\u003CBOARD_KEY>` | Kanban\u002Ftask board | `research` |\n| `\u003CPROFILE_NAME>` | Runtime profile\u002Fassignee | `researcher` |\n| `\u003CSOUL_ID>` | Persona\u002Fsoul identifier | `researcher` |\n| `\u003CSOUL_PATH>` | Optional SOUL\u002Fprofile file path | `~\u002F.hermes\u002Fprofiles\u002Fresearcher\u002FSOUL.md` |\n\nKeep company-specific values in deployment config, not inside the generic skill.\nTreat real tenant keys, workspace IDs, surface IDs, user IDs, and message links as operational metadata. Use placeholders in reusable docs, examples, tickets, screenshots, and public reports.\n\n## Operating Principles\n\n1. **Conversation first** — a mapped topic\u002Fchannel is still a place to talk.\n2. **Explicit task intake** — create tasks only for explicit prefixes or commands.\n3. **Platform-neutral registry** — represent Telegram and Discord with the same conceptual fields.\n4. **Soul\u002Fprofile separation** — soul describes identity\u002Fbehavior; profile describes runtime\u002Ftool\u002Fmemory isolation when available.\n5. **Read-only discovery before mutation** — inspect current config, profiles, boards, and behavior before edits.\n6. **No secrets in souls** — SOULs, registry entries, templates, and examples must never contain credentials.\n7. **Human confirmation before applying** — show a summary and ask the operator to confirm before writing config or restarting runtime.\n8. **Tenant-scoped routing** — in multi-tenant runtimes, match routes by tenant\u002Faccount\u002Fproject scope plus platform\u002Fworkspace\u002Fsurface IDs, never by surface name alone.\n9. **Minimal source references** — task creation should store a compact source reference, not raw message metadata, transcripts, permalinks, or user profile data by default.\n10. **Policy before prompt** — route lookup, authorization, and task-prefix detection are policy decisions; a SOUL\u002Fprofile prompt must not grant itself tools, secrets, broader route scope, or automatic task intake.\n\n## Data Model\n\nUse a platform-neutral registry. The actual filename depends on the runtime; common examples include `topic_souls.yaml`, `channel_souls.yaml`, or a section inside a larger routing config.\n\n```yaml\ntopic_souls:\n  version: \"1\"\n  entries:\n    - tenant_key: \"\u003CTENANT_KEY>\"\n      environment: \"\u003CENVIRONMENT>\"\n      platform: \"\u003CPLATFORM>\"\n      workspace_id: \"\u003CWORKSPACE_ID>\"\n      surface_id: \"\u003CSURFACE_ID>\"\n      surface_type: \"\u003CSURFACE_TYPE>\"\n      surface_name: \"\u003CSURFACE_NAME>\"\n      project: \"\u003CPROJECT_KEY>\"\n      board: \"\u003CBOARD_KEY>\"\n      profile: \"\u003CPROFILE_NAME>\"\n      soul_id: \"\u003CSOUL_ID>\"\n      mode: direct\n      kanban_intake: explicit\n      source_reference_policy: minimal\n      include_permalink: false\n      actor_reference: hashed\n      status: draft\n      activation:\n        requires_human_confirmation: true\n        approved_by: null\n        approved_at: null\n      prompt_summary: >-\n        Short description of what this specialist agent does, how it answers,\n        and what it must not do without approval.\n```\n\nRecommended `status` values:\n\n| Status | Meaning |\n|---|---|\n| `draft` | Collected but not active; missing IDs or approval |\n| `active` | Runtime may route matching messages to this soul\u002Fprofile |\n| `disabled` | Preserved for history but ignored by runtime |\n\n`active` entries must have a complete tenant\u002Fscope key when the runtime serves more than one tenant, account, project, or customer. If the runtime is intentionally single-tenant, document that assumption in the deployment config.\n\n## Wizard Workflow\n\nRun this as a guided wizard. Ask one question at a time unless the user explicitly requests bulk\u002FYAML mode.\n\n### 1. Explain the capability\n\nUse operator-facing wording:\n\n```text\nVamos configurar um agente especialista para um tópico\u002Fcanal.\n\nMensagens normais continuam sendo conversa. Tarefas só entram no Kanban quando alguém usar um prefixo explícito como `task:`, `tarefa:`, `kanban:`, `\u002Ftask` ou `\u002Fkanban`.\n```\n\n### 2. Choose platform\u002Fworkspace\n\nAsk:\n\n```text\nQual plataforma e workspace\u002Fgrupo\u002Fservidor você quer configurar?\nSe souber o ID técnico, envie também. Se não souber, posso orientar como descobrir.\n```\n\nCapture `platform`, `workspace_name`, and `workspace_id`. If the ID is unknown, continue in `draft` mode.\n\nDo not paste real workspace IDs, private channel IDs, or message links into shared issue trackers or reusable examples. Keep them in the runtime configuration surface with the same access controls as other operational metadata.\n\n### 3. Choose surfaces\n\nAsk:\n\n```text\nQuais tópicos\u002Fcanais\u002Fthreads devem ter agentes especialistas?\nPode mandar uma lista com nomes humanos primeiro; depois descobrimos os IDs estáveis.\n```\n\nCapture `surface_name`, `surface_type`, and `surface_id` when available.\n\n### 4. Choose or create soul\u002Fprofile\n\nFor each surface, ask:\n\n```text\nPara \"\u003CSURFACE_NAME>\", você quer usar uma soul\u002Fprofile existente ou criar uma nova especialista?\n```\n\nIf creating a new specialist, collect:\n\n- short name;\n- role\u002Ffunction;\n- domain expertise;\n- tone;\n- response style;\n- boundaries\u002Fapproval rules;\n- default board\u002Fprofile for explicit tasks;\n- examples of conversation vs task messages.\n\n### 5. Generate a SOUL\u002Fprofile prompt\n\nUse this structure:\n\n```markdown\n# \u003CSOUL_ID> SOUL\n\n## Role\n\u003CWhat this specialist agent is responsible for.>\n\n## Domain\n\u003CSubjects and decisions this agent is strong at.>\n\n## Tone\n\u003CDirect, consultative, technical, didactic, critical, creative, etc.>\n\n## Response Style\n\u003CHow normal replies should be shaped.>\n\n## Boundaries\n\u003CWhat requires explicit approval.>\n\n## Task Intake\nNormal messages are conversation. Create tasks only for explicit prefixes:\n`task:`, `tarefa:`, `kanban:`, `\u002Ftask`, `\u002Fkanban`, `criar task:`, `criar tarefa:`.\n\nDefault board: `\u003CBOARD_KEY>`.\nDefault profile\u002Fassignee: `\u003CPROFILE_NAME>`.\n```\n\n### 6. Confirm behavior per surface\n\nRecommended default:\n\n```yaml\nmode: direct\nkanban_intake: explicit\n```\n\nOnly use automatic task creation if the operator explicitly says the surface is intake-only.\n\n### 7. Review before writing\n\nShow a summary and ask for confirmation:\n\n```text\nResumo antes de aplicar:\n- Platform: \u003CPLATFORM>\n- Workspace: \u003CWORKSPACE_NAME> (\u003CWORKSPACE_ID>)\n- Surface: \u003CSURFACE_NAME> \u002F \u003CSURFACE_ID>\n- Soul: \u003CSOUL_ID>\n- Profile: \u003CPROFILE_NAME>\n- Board: \u003CBOARD_KEY>\n- Behavior: conversa por padrão; Kanban só explícito\n\nConfirmar aplicação?\n```\n\nDo not write config until the operator confirms.\n\n### 8. Apply safely\n\nAfter confirmation:\n\n1. Back up the current routing\u002Fprofile config.\n2. Write or patch registry entries.\n3. Create or update SOUL\u002Fprofile files if the runtime supports them.\n4. Validate every `active` entry has platform, workspace id, surface id, profile\u002Fsoul, and board if task intake is enabled.\n5. Run a dry-run resolver or unit tests if available.\n6. Ask before restarting gateway\u002Fbot\u002Fruntime.\n\n## Runtime Behavior Contract\n\n### Route resolution and policy order\n\nThe runtime should resolve policy before loading surface-specific prompt text:\n\n```text\nNormalize tenant\u002Faccount scope\n+ normalize platform\u002Fworkspace\u002Fsurface IDs\n+ match active route by tenant scope + platform + workspace_id + surface_id\n+ verify route activation and operator approval\n+ detect explicit task prefix\n+ load base runtime prompt\n+ load configured SOUL\u002Fprofile as behavior context only\n```\n\nThe SOUL\u002Fprofile can shape tone, domain focus, and response format. It must not expand route scope, grant tools, bypass tenant isolation, read secrets, switch task intake to automatic, or override human-approval requirements.\n\n### Normal message\n\nInput:\n\n```text\nQual é sua função neste tópico\u002Fcanal?\n```\n\nExpected behavior:\n\n```text\nLoad base runtime prompt\n+ load platform\u002Fchat context\n+ load surface-specific route\n+ load configured soul\u002Fprofile prompt as behavior context only\n+ answer conversationally\n+ do not create a Kanban task\n```\n\n### Explicit task message\n\nInput:\n\n```text\ntask: pesquisar benchmarks para agentes autônomos e documentar recomendações\n```\n\nExpected behavior:\n\n```text\nCreate task on configured board\nAssign configured profile\u002Fagent\nInclude minimal surface context in title\u002Fbody\nReturn confirmation with board, task id, profile, and dashboard link when available\n```\n\nStore a minimal source reference with the task:\n\n```yaml\nsource_ref:\n  platform: \"\u003CPLATFORM>\"\n  tenant_key: \"\u003CTENANT_KEY>\"\n  workspace_id: \"\u003CWORKSPACE_ID>\"\n  surface_id: \"\u003CSURFACE_ID>\"\n  message_id: \"\u003CMESSAGE_ID>\"\n  actor_ref: \"\u003CHASHED_OR_INTERNAL_ACTOR_REF>\"\n```\n\nDo not include raw message metadata, user profile data, transcripts, or private permalinks unless the deployment has an explicit retention and access-control policy for that data.\n\n### Supported explicit prefixes\n\nSupport at least:\n\n```text\n\u002Ftask ...\n\u002Fkanban ...\ntask: ...\nkanban: ...\ntarefa: ...\ncriar task: ...\ncriar tarefa: ...\n```\n\nDo not trigger tasks from vague phrases such as “precisamos”, “seria bom”, “vamos fazer”, or “alguém deveria”.\n\n## Validation Checklist\n\n- [ ] Existing config, profiles\u002Fsouls, boards, and platform metadata inspected before mutation.\n- [ ] Multi-tenant runtimes include tenant\u002Faccount\u002Fproject scope in route matching.\n- [ ] Every active route has stable workspace and surface IDs.\n- [ ] Active routes require operator approval and cannot be activated by prompt text alone.\n- [ ] Normal message test answers conversationally and creates no task.\n- [ ] Explicit `task:` test creates one task on the configured board\u002Fprofile.\n- [ ] Disabled\u002Fdraft entries are ignored by runtime.\n- [ ] Wrong workspace\u002Fsurface does not match a route.\n- [ ] Wrong tenant\u002Faccount\u002Fproject scope does not match a route.\n- [ ] Task source references are minimized and do not store raw private metadata by default.\n- [ ] SOUL\u002Fprofile prompt cannot grant tools, secrets, route scope, or automatic intake.\n- [ ] SOUL\u002Fprofile files contain no secrets.\n- [ ] Runtime restart, if needed, was explicitly approved.\n- [ ] Operator-facing README or handoff has been provided for non-technical users.\n\n## Common Pitfalls\n\n1. **Routing every message into Kanban.** Specialist surfaces are conversation-first unless explicitly configured as intake-only.\n2. **Using human names as the only routing key.** Always discover stable platform IDs before activating routes.\n3. **Skipping tenant scope.** Workspace IDs can be stable platform identifiers, but they are not tenant authorization boundaries.\n4. **Copying private metadata into tasks.** Store compact source references by default; avoid raw permalinks, transcripts, and user profile data.\n5. **Letting prompt text change policy.** SOUL\u002Fprofile text cannot grant tools, secrets, route scope, or automatic task creation.\n6. **Mixing secrets into SOUL files.** Store credentials in the runtime’s secret manager\u002Fenv\u002FKeePass, never in prompts.\n7. **Skipping confirmation.** A wizard can draft config, but it must not apply active routes without operator confirmation.\n8. **Restarting the bot\u002Fgateway too early.** Validate config and ask first; many runtimes can dry-run route resolution without restart.\n9. **Duplicating every global skill into every profile.** Prefer profile overlays only for exceptions; use global skill discovery\u002Ffallback when available.\n10. **Forgetting operator language.** Product UX should say “configure an agent for this channel”, not “edit SOUL.md and topic_routes”.\n\n## One-Shot Handoff Prompt\n\nUse when asking another implementation agent to configure this capability:\n\n```text\nConfigure a specialist agent for each selected Telegram topic, Discord channel\u002Fthread, or equivalent messaging surface.\n\nNormal messages must remain conversational. Create Kanban\u002Ftasks only when the user starts with `\u002Ftask`, `\u002Fkanban`, `task:`, `kanban:`, `tarefa:`, `criar task:` or `criar tarefa:`.\n\nFirst run a wizard with the operator:\n1. choose platform\u002Fworkspace;\n2. collect surface names and stable IDs;\n3. choose existing soul\u002Fprofile or create a new specialist;\n4. define role, domain, tone, response style, boundaries, default board\u002Fprofile;\n5. show a summary and ask for confirmation before writing config.\n\nAfter confirmation, create\u002Fupdate a platform-neutral route registry with tenant\u002Faccount scope when applicable, platform, workspace_id, surface_id, surface_type, surface_name, project, board, profile, soul_id, mode=direct, kanban_intake=explicit, status, source_reference_policy=minimal, and human activation metadata.\n\nValidate with two live tests per surface:\n- `Qual é sua função neste tópico\u002Fcanal?` must answer conversationally and create no task.\n- `task: teste de criação explícita` must create one task on the configured board\u002Fprofile.\n\nDo not restart runtime\u002Fgateway unless the operator explicitly approves. Do not store secrets, raw private message metadata, or public report examples with real IDs in souls or registry.\n```\n",{"name":5,"author":24,"license":15,"version":25,"metadata":26,"platforms":40,"description":7},"Skills Lab","1.0.0",{"hermes":27},{"tags":28,"related_skills":36},[29,30,31,32,33,34,35],"topic-routing","profiles","souls","telegram","discord","kanban","tenant-onboarding",[37,38,39],"hermes-agent","github-operations","software-development-workflows",[41,42],"linux","macos","main","b5c05537c2a696d4f96f17b51e5e2951fcab8f5b","skills\u002Fproductivity\u002Ftopic-channel-soul-routing","https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills","https:\u002F\u002Fgithub.com\u002Ffmercurio","2026-11-07 04:26:06",{"source":8,"skillName":5,"commands":50,"prompt":60,"zipAvailable":61},[51,54,57],{"tool":52,"command":53},"npx","npx skills add fmercurio\u002Fclawdia-custom-skills --skill topic-channel-soul-routing",{"tool":55,"command":56},"bunx","bunx skills add fmercurio\u002Fclawdia-custom-skills --skill topic-channel-soul-routing",{"tool":58,"command":59},"pnpm","pnpm dlx skills add fmercurio\u002Fclawdia-custom-skills --skill topic-channel-soul-routing","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Fskills\u002Fproductivity\u002Ftopic-channel-soul-routing\u002FSKILL.md",true,[63,69,73,77,81],{"path":64,"resourceType":65,"resourceTypeDesc":66,"sizeBytes":67,"sourceHtmlUrl":68},"README.md","REFERENCE","参考文档",8638,"https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Fskills\u002Fproductivity\u002Ftopic-channel-soul-routing\u002FREADME.md",{"path":70,"resourceType":65,"resourceTypeDesc":66,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Fimplementation-notes.md",3584,"https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Fskills\u002Fproductivity\u002Ftopic-channel-soul-routing\u002Freferences\u002Fimplementation-notes.md",{"path":74,"resourceType":65,"resourceTypeDesc":66,"sizeBytes":75,"sourceHtmlUrl":76},"templates\u002Fsoul.example.md",982,"https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Fskills\u002Fproductivity\u002Ftopic-channel-soul-routing\u002Ftemplates\u002Fsoul.example.md",{"path":78,"resourceType":65,"resourceTypeDesc":66,"sizeBytes":79,"sourceHtmlUrl":80},"templates\u002Ftopic-souls.example.yaml",1180,"https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Fskills\u002Fproductivity\u002Ftopic-channel-soul-routing\u002Ftemplates\u002Ftopic-souls.example.yaml",{"path":82,"resourceType":83,"resourceTypeDesc":84,"sizeBytes":85,"sourceHtmlUrl":86},"tests\u002Ftest_topic_channel_security_contract.py","SCRIPT","脚本",2465,"https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Fskills\u002Fproductivity\u002Ftopic-channel-soul-routing\u002Ftests\u002Ftest_topic_channel_security_contract.py",[88,95,101,107,113,118],{"skillNo":89,"name":90,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":92,"categories":93,"keywords":94,"publishedAt":19},"01M75BDD9N7RSFBWGDNTZK9NYT","overlay-skill","Fixture skill for deployment planner tests.","https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Ftools\u002Fskill_deploy\u002Ftests\u002Ffixtures\u002Fworkspace\u002Fskills\u002Fproductivity\u002Foverlay-skill\u002FSKILL.md",[],[],{"skillNo":96,"name":97,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":98,"categories":99,"keywords":100,"publishedAt":19},"01M75BDAV2X86DYXADGV8B1ZCZ","candidate-skill","https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Ftools\u002Fskill_deploy\u002Ftests\u002Ffixtures\u002Fworkspace\u002Fskills\u002Fproductivity\u002Fcandidate-skill\u002FSKILL.md",[],[],{"skillNo":102,"name":103,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":19},"01M75BD9NPF9EKPHAZMBVDM8GN","builtin-collision","https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Ftools\u002Fskill_deploy\u002Ftests\u002Ffixtures\u002Fworkspace\u002Fskills\u002Fproductivity\u002Fbuiltin-collision\u002FSKILL.md",[],[],{"skillNo":108,"name":109,"nameZh":6,"description":91,"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},"01M75BD7NV55ZMQVC0GY7FSG2F","approved-global","https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Ftools\u002Fskill_deploy\u002Ftests\u002Ffixtures\u002Fworkspace\u002Fskills\u002Fproductivity\u002Fapproved-global\u002FSKILL.md",[],[],{"skillNo":114,"name":109,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":115,"categories":116,"keywords":117,"publishedAt":19},"01M75BD54973NF4BHPZR3F4Z8A","https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Ftools\u002Fskill_deploy\u002Ftests\u002Ffixtures\u002Fworkspace\u002Fskills\u002Fproductivity\u002Fapproved-global-duplicate\u002FSKILL.md",[],[],{"skillNo":119,"name":90,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":120,"categories":121,"keywords":122,"publishedAt":19},"01M75BD2N1FJZ142MYPJ19R29Z","https:\u002F\u002Fgithub.com\u002Ffmercurio\u002Fclawdia-custom-skills\u002Fblob\u002Fmain\u002Ftools\u002Fskill_deploy\u002Ftests\u002Ffixtures\u002Fworkspace\u002Fruntime\u002Fskills\u002Fproductivity\u002Foverlay-skill\u002FSKILL.md",[],[]]