{
  "id": "page:library-communication",
  "_kind": "Page",
  "_file": "wiki/library/communication.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "title": "Communication specialization (Library)",
    "displayName": "Communication specialization (Library)",
    "slug": "library/communication",
    "articlePath": "wiki/library/communication.md",
    "article": "\n# Communication specialization\n\nProcesses for communication work: channel management (Slack/Discord), content\nproduction and validation, and governed multi-audience announcements. The\nspecialization layers a flagship orchestrated pipeline over four Style-B\npersona point files.\n\n## Flagship\n\n`multi-audience-announcement-pipeline.js` — end-to-end governed announcement\nflow: source-material intake -> single canonical fact sheet -> `ctx.parallel`\nper-audience variants (internal, customers, press, partners) -> adversarial\nclaim-trace/tone/cross-audience-drift gate -> per-audience routed approvals\n(frozen audience->expert map) -> policy-gated sends -> delivery confirmation;\nkip recall at intake and assert at close (kind `communication`).\n\nPolicy-gated actions:\n\n| actionId | Audiences covered | Expert |\n| --- | --- | --- |\n| `internal-broadcast` | internal | `exec-sponsor` |\n| `external-comms-send` | customers, partners (single gate covering both variants) | `comms-lead` |\n| `press-release-publish` | press | `comms-lead` |\n\n## Point files\n\n- `slack-manager.js` — Slack-manager persona: scan channels/unanswered mentions,\n  classify each (respond / moderate / escalate / notify), draft per-mention\n  responses in parallel, send via `@slack/web-api` ad-hoc code.\n- `discord-manager.js` — Discord counterpart: scan server/unanswered mentions,\n  classify, draft and execute per-mention actions in parallel via `discord.js`.\n- `content-writer.js` — brief -> plan -> draft -> self-edit persona; drives the\n  flagship's per-audience drafting stage.\n- `content-validator.js` — 5-axis devil's-advocate review (clarity, audience\n  perspectives, ambiguity, metaphors, consistency); drives the flagship's\n  tone-audience critic.\n\nNote: the flagship composes these point files BY NAME as draft/send-stage\nexecutor personas in task prompts — there are no ESM imports of point files.\n\n## Shared combinators\n\nThe flagship depends on `../common-utilities/routed-gate-combinators.js`\n(`routedBreakpoint`, `adversarialGate`, `kipRecall`, `kipAssert`,\n`gateFixerTask`).\n\n## Governance model\n\nFail-closed policy gates: `breakpointId` equals the `actionId`, send executors\nrun ONLY inside `approved === true` guards, and every outcome — including\nskipped gates on a failed accuracy gate and non-interactive auto-approvals —\nis recorded in `outputs.gatedActions`. One audience's rejection records the\ndecision and never blocks the other audiences.\n",
    "documents": [
      "specialization:communication"
    ]
  },
  "outgoingEdges": [
    {
      "from": "page:library-communication",
      "to": "specialization:communication",
      "kind": "documents"
    }
  ],
  "incomingEdges": [
    {
      "from": "page:index",
      "to": "page:library-communication",
      "kind": "contains_page"
    }
  ]
}