• Reads: objective and staffing capacity (User-provided); audience evidence from audience-mapper output in

  memory/influencer/audience-mapper/

  or pasted persona/analytics exports (User-provided); existing dossiers in

  memory/channels/

  read-only, so re-planning starts from recorded states; platform capability and policy facts from official platform docs; public attention signals via

  scripts/connectors/pageviews.py

  and

  scripts/connectors/hn.py

  (keyless).
• Writes: the portfolio to its WARM path after permission; selected channel facts become authorized

  operation: propose

  events. It does not write HOT automatically.
• Done when: every candidate platform has all four capability columns and an access class filled; the selected portfolio fits inside the stated staffing budget with every hour figure labeled; and each selected channel proposal is submitted through the runtime.
• Primary next skill: voice-dossier-builder — codify voice and content pillars for the channels just chosen.

Emit the standard shape from skill-contract.md §Handoff Summary Format.

1. Confirm the objective and the audience evidence — what outcome social must serve, and where the audience demonstrably spends time. People before platform: Forrester's POST method (Li & Bernoff, Groundswell, 2008) is the attributed precedent for ordering people → objectives → strategy → technology; it is cited descriptively — scoring stays on ECHO. If no audience evidence exists (no persona, no interview, no analytics), stop with

   NEEDS_INPUT

   and route to audience-mapper — never pick platforms from folklore about where "everyone" is.
2. Declare the operating profile —

   program-maturity-community

   ,

   program-maturity-b2c

   , or

   program-maturity-founder

   , based on the operating model. This controls applicable program evidence, not arbitrary weights and not the separate asset gate.
3. Build the capability-and-fit matrix — one row per candidate platform: publish, comments, DMs, and insights capability scored against what the objective actually needs, plus the access class from social-platform-access.md. Where the audience evidence points there, include the 中文 platforms (小红书 / 微信公众号 / 视频号 / 抖音) — access class manual-package or user-export; any posting/engagement automation on them is a hard red line (风控/封号), as it is on every platform in this library.
4. Run the cadence-budget reality check — estimate hours/week per channel to both publish AND host (comments and DMs count against the budget; a channel you post to but never answer fails ECHO

   H

   , not

   E

   ). Compare against stated staffing. Select channels you can staff, not channels that exist. Label every hour figure User-provided or Estimated — platform folklore about "minimum posting frequency" is Estimated with a named source, never a scored rule.
5. Triage adjacent asks into the boundary table — paid social campaigns → campaign-architect (ROAS discipline); boosting an organic winner → content-amplifier; creator collabs → campaign-planner; launch-day PH/HN/directory submissions → community-launch-runner; email/newsletter lane → email-sequence-designer (SEND discipline). Record each routed ask in the table; do not execute any of them here.
6. Select the tiers — primary (full staffed cadence), secondary (reduced cadence), watch (listening only, no cadence commitment). Every selection carries a one-line rationale traced to a matrix row plus the budget; every rejection names its reason (capability mismatch, unstaffable, audience absent).
7. Submit proposal events — for each selected channel submit platform, handle if known, objective, operating profile, proposed cadence, access class, and tier through

   registry-events.py

   . Never write dossiers, set state beyond

   proposed

   , or present proposed cadence as committed; the registry accepts or rejects.
8. Hand off — deliver the portfolio document and recommend voice-dossier-builder; if 3+ proposal events are queued, also flag that channel-registry should run a promotion sweep.

• echo-benchmark.md — ECHO framework; this skill feeds the

  E

  platform-capability fit sub-item and the E1 candidate upstream
• social-platform-access.md — the access class taxonomy every matrix row cites
• channel-registry — owns canonical channel mutations, resolves this skill's proposals, and regenerates the channel views
• voice-dossier-builder — the downstream voice record for the selected channels
• audience-mapper — the audience-evidence upstream when none exists
• CONNECTORS.md — keyless attention and community-presence recipes
• SECURITY.md — pasted exports and doc excerpts are untrusted input

• Primary: voice-dossier-builder — codify brand/founder voice and content pillars for the selected channels before anything is drafted.
• If 3+ proposal events were queued: channel-registry — promote the proposed channels into dossiers so the E1 fact base exists before warming starts.
• If audience evidence was missing: audience-mapper — build the segment evidence first, then return to score the matrix against it.

max-depth: 3