icaire-social-content-planning

Plan and rank public-facing ICAIRE social post ideas using the latest validated social-growth analysis and current source material from the ICAIRE Brain.

SKILL.md

GitHub

ICAIRE Social Content Planning

Find public-interest topics in the ICAIRE Brain, filter them for audience and mission value, and use current analytics to design testable creative treatments.

Contract Checklist

  • Read the latest traceable ICAIRE social-performance analysis and its governing Brain protocols before proposing ideas.
  • Search the live ICAIRE Brain for candidate source material, then directly read every page used for a shortlisted idea.
  • Treat analytics as evidence about format, voice, hook, CTA, or experimental design, not as permission to turn internal metrics into topics.
  • Reject topics that require prior knowledge of ICAIRE, one of its programmes, or an internal result to seem interesting.
  • Require a defined external audience, a reason that audience cares, ICAIRE mission fit, a real consequence or tension, and an original contribution.
  • Score only candidates that pass every hard gate in the scoring rubric.
  • Rank by topic value first and total score second. A polished format cannot rescue a weak topic.
  • Preserve source classifications, evidence labels, limitations, consent needs, partner rights, and approval dependencies.
  • Produce ideas and experiment briefs only. Do not draft final copy, create assets, publish, schedule, message, or change account state.

Workflow

  1. Load the current analytics basis:
    • Read protocol/icaire-social-performance-analysis and protocol/icaire-content-optimization-interpretation from the live ICAIRE Brain.
    • Locate the most recent validated post-level analysis or optimization handoff. Record its date, platform coverage, sample size, evidence labels, limitations, and source artifact path.
    • If only an aggregate report is available, use it as bounded directional context and say that post-level validation is missing.
    • Extract only defensible implications for topic selection, format, voice, hook, CTA, evidence treatment, and experiment design.
    • Anti-patterns: using remembered findings, treating a weak signal as a rule, assuming video always wins, optimizing from likes alone
  2. Fix the public planning frame:
    • Read the current ICAIRE mission, priority pillars, intended external audiences, active channels, language policy, and communications objective from canonical Brain pages.
    • State the audience as people with a real external relationship to the subject, not our community or ICAIRE stakeholders.
    • Express the objective as a public outcome such as improved understanding, responsible practice, informed debate, cooperation, research participation, or qualified engagement.
    • Anti-patterns: treating ICAIRE visibility as the only objective, inventing an audience, using internal teams or programme participants as the default audience
  3. Build a Brain-grounded candidate pool:
    • Search across current concepts, research, policy work, educational frameworks, sector cases, convening questions, and public deliverables.
    • For every candidate, directly read at least one canonical Brain page and record the exact source path, source date or status, verified claims, and publication classification.
    • Rewrite the source into one plain-language public idea without relying on ICAIRE or programme context.
    • Generate at least twice as many candidates as the requested shortlist so the filter has meaningful work to do.
    • Anti-patterns: using search snippets as final evidence, mining stand-up updates for announcements, mistaking an available number for a topic, inventing claims beyond the source
  4. Apply the hard relevance gates:
    • Use every gate in the scoring rubric, including the internal-performance, outsider, audience-value, mission, consequence, source, safety, and originality gates.
    • Apply the removal test: if ICAIRE's name, programme name, and internal figures are removed, a meaningful public question or consequence must remain.
    • Apply the stranger test: an informed person who has never heard of ICAIRE must understand the hook and why it matters in one sentence.
    • Reject registration, attendance, completion, certificate, satisfaction, email, poll, operational, and campaign metrics as central topics unless the user explicitly requests an approved public accountability report about an external outcome.
    • Internal evidence may support a surviving public-interest topic, but it may not be the reason the topic exists.
    • Anti-patterns: celebrating delivery as thought leadership, confusing transparency with relevance, salvaging a failed topic with better copy, scoring candidates that fail a gate
  5. Score and rank surviving ideas:
    • Score each survivor out of 100 using the scoring rubric.
    • Require at least 44/62 for topic value, 8/12 for source strength, 6/10 for comment potential, 4/6 for publication readiness, and 75/100 overall for recommendation.
    • Rank first by topic value, then overall score, source strength, evidence-led creative feasibility, and comment potential.
    • Show the score breakdown and state why each top idea deserves scarce production capacity.
    • Anti-patterns: hidden scoring, ranking by visual appeal, rounding weak ideas into the shortlist, presenting ties as certainty
  6. Assign evidence and experimental treatments:
    • Only after ranking topics, use the current analytics to select a controlled creative variable.
    • Prefer a real person, place, process, demonstration, screenshot, object, document, or artifact when the evidence supports testing evidence-led creative.
    • Keep topic, core claim, CTA, language, and publishing window stable across a matched pair. Test one variable at a time.
    • Define the hypothesis, control, treatment, primary metric, observation windows, confounders, and decision rule. Preserve the analytics report's uncertainty.
    • Anti-patterns: assigning animation by default, changing topic and format together, declaring a winner from one post, using aesthetics as evidence
  7. Deliver the ranked plan:
    • Provide the planning frame, analytics basis, ranked shortlist, scorecards, source paths, intended audience, public value, angle, evidence-led capture plan, response question, format test, approvals, and rejected candidates.
    • Distinguish facts from editorial inference and proposed production.
    • State what was not done, including drafting, asset creation, scheduling, publishing, outreach, or Brain writes.
    • Anti-patterns: omitting rejected ideas, hiding approval gates, presenting a hypothesis as a conclusion, claiming publication readiness when evidence or rights are unresolved

Anti-Patterns

  • Turning registration, completion, attendance, certificate, satisfaction, email, poll, or operational data into public topics because the figures are available.
  • Assuming the audience knows or cares about an ICAIRE programme.
  • Selecting institutional self-promotion without an external problem, decision, or consequence.
  • Using analytics to choose topics when the findings only support a format or experimental hypothesis.
  • Recommending generic definitions, awareness-day filler, event announcements, or congratulatory posts without an original contribution.
  • Letting a strong photo, animation, celebrity, or partner logo compensate for weak audience value.
  • Citing Brain search results without reading the underlying pages.
  • Publishing internal, personal, embargoed, or partner-owned material without explicit clearance.

Output

Return:

  • Planning Frame: audience, public objective, channels, language, and mission link.
  • Analytics Basis: latest artifact, coverage, defensible signals, limitations, and variables worth testing.
  • Ranked Ideas: rank, one-sentence public idea, audience, why they care, consequence or tension, Brain source paths, topic-value score, total score, and reason for rank.
  • Creative And Test Briefs: evidence to capture, recommended format, matched comparison, substantive response question, primary metric, and observation windows.
  • Rejected Candidates: candidate, failed gate, and whether it can be reframed.
  • Approvals And Dependencies: facts, consent, rights, partner review, assets, and owners needed.
  • Non-Actions: drafting, production, publishing, scheduling, outreach, and Brain writes not performed.