Learning Agenda

Learning Agenda

Three output types. v1 = canonical federation agenda (the annual standing document). v2 = quarterly 1-page RAG dashboard. v3 = MA adaptation template each Member Association uses to anchor its own agenda to the federation. The skill enforces the A+C hybrid discipline locked 2026-05-10: USAID document structure plus Patton use-anchoring per question, with Britton-style after-action review posture in the annual refresh.

When to use

Trigger on any of:

• "learning agenda", "federation learning agenda", "annual learning questions"
• "RAG dashboard" in the context of the agenda; "quarterly status of learning agenda"
• "MA learning agenda", "MA adaptation template", "MA-level learning agenda"
• "refresh the agenda", "draft the 2027 cycle agenda", "federation MEL annual refresh"
• explicit naming of an output type (v1, v2, v3, canonical, rag-dashboard, ma-template)

Do not trigger for:

• learning products derived from a Researcher Evidence Brief — that is the learning-product skill
• general decision memos — that is the decision-memo skill
• general organisational learning frameworks — that is the evidence-synthesis skill running against the organisational-learning-senge-argyris wiki page
• programme M&E plans, results frameworks, indicator sets — those are separate skills
• the Researcher Evidence Brief that answers a learning-agenda question — that is the researcher skill

Required inputs

Ask in one batch. Inputs marked required must be present at invocation; inputs marked optional carry defaults.

1. Output type (required): one of canonical (v1, full federation agenda), fragment (single-question entry — a sub-mode of v1 producing one Tier 1 or Tier 2 question entry without full-document scaffolding; used for validation, peer-review of a candidate question, or iterative agenda construction), rag-dashboard (v2), ma-template (v3).
2. Cycle name (required): e.g. 2027, FY2027-28, 2026-2028 biennium. Used in the title block and refresh cadence.
3. Federation strategic context (required for v1; optional for v2/v3): one paragraph or pointer to a strategic-plan-refresh document naming the strategic pressures that shape the cycle.
4. Question list (optional for v1): if Ane provides a list of learning questions with proposed Tier 1/Tier 2 classification, the skill structures them into the agenda format. If absent, the skill drafts illustrative questions from the strategic context and flags every question with ⚠️ Reasoned default — confirm with federation MEL function.
5. Anchored v1 path (required for v2 + v3): file path to the canonical agenda the v2 or v3 references. Default: agent-improvements/learning-agenda-exemplar-v1.md (the exemplar) for skill validation; in real use, the path to the current cycle's canonical agenda.
6. Period (required for v2 only): e.g. Q2 2027. Used in the dashboard title block.
7. Status updates (optional for v2): per-question RAG status + one-line note. If absent, the skill produces an illustrative status flagged as such.
8. MA name (required for v3): the IPPF EN Member Association name (e.g. IPPF Romania (SECS), IPPF Albania).
9. MA strategic context (required for v3): one paragraph naming the MA's operating context that shapes its agenda.
10. MA question list (optional for v3): MA-specific questions; if absent, skill drafts from the strategic context and flags as reasoned default.
11. Cross-MA naming convention (optional, all output types): name (default per locked decision 11) or sub-region (the redaction-fallback). Country-level naming triggers a redaction-candidate flag in the provenance footer.

Session start

1. Read this SKILL.md fully.
2. Read the matching exemplar at agent-improvements/learning-agenda-exemplar-v[1|2|3].md for the requested output type. The exemplar is the structural target.
3. Read the framework foundation at mel_wiki/wiki/frameworks/learning-agenda-usaid-cla-2022.md and mel_wiki/wiki/frameworks/utilisation-focused-knowledge-management.md once per session if not already in context.
4. Read ~/.claude/CLAUDE.md audience-tier register section once per session if not already in context.

Workflow

Step 1 — parse inputs and route

Identify output type. If missing or ambiguous, ask Ane in one round:

"Which output type? canonical (v1, the annual federation agenda), rag-dashboard (v2, the 1-page quarterly status), or ma-template (v3, the MA adaptation template)?"

If ma-template is selected without an ma-name parameter, ask for it in the same round.

If fragment is selected, additionally ask: which Tier (1 or 2) and what question text. If a Tier 1 fragment is requested without enough context to populate the three Patton anchors (decision-maker + commitment to use + use champion), the skill drafts reasoned defaults from the federation strategic context and flags each anchor ⚠️ Reasoned default — confirm before publication.

Step 2 — load source material

Read in order:

1. agent-improvements/learning-agenda-exemplar-v[N].md — structural target for the requested output.
2. mel_wiki/wiki/frameworks/learning-agenda-usaid-cla-2022.md — framework foundation; the four-element USAID template + three-anchor Patton extension definitions.
3. mel_wiki/wiki/frameworks/utilisation-focused-knowledge-management.md — use-anchoring operational principles.
4. For v2 and v3: the v1 path provided.
5. agent-improvements/learning-agenda-design-brief.md "Decisions locked" section — the 12 spec-session decisions that shape the output.

Step 3 — apply the output-type sub-prompt

Route to one of the three sub-prompts below. Each sub-prompt instantiates the corresponding exemplar's structure against the inputs.

Step 4 — apply quality gates (all output types)

Before returning the draft:

• A+C hybrid discipline. USAID document structure + Patton use-anchoring per Tier 1 question + Britton-style AAR posture in the annual-refresh section (v1 only). Mixing partial discipline produces partial output.
• Tier 1 discipline. Every Tier 1 question carries: question + method + decision + use plan (USAID four elements) PLUS decision-maker + commitment to use + use champion (Patton three anchors). Missing any anchor flags ⚠️ Tier 1 anchor incomplete — confirm or downgrade to Tier 2.
• Tier 2 discipline. Tier 2 questions carry method + indicative use case. They do NOT require a named decision-maker, by design (decision 7).
• Tier 1 / Tier 2 split. ~60/40 ratio with Tier 1 dominant. v1 with all Tier 2 fails the A+C hybrid; v1 with no Tier 2 collapses to pure Patton (rejected at the spec session).
• Voice register. v1 uses junior-MEL register: framework names visible, callout boxes (common pitfall, why this matters, worked example), glossary footer, annotated evidence base. v2 and v3 use Tier 1 working brief default: no callouts, no glossary, compact.
• Cross-MA naming. Default name per locked decision 11 (different from learning-product which defaults to sub-region). Country-level naming triggers redaction-candidate flag in provenance footer.
• Citation placement. Inline citations only when the framework name is itself the teaching point in v1 junior-MEL register; otherwise **Evidence base:** at end of section. URLs reused from the wiki; not fabricated.
• Citation rigour. Author surname + year + venue + section where applicable. Verified URLs from the framework wiki pages only. No new sources introduced by the skill.
• Sentence length. Under 25 words. Active voice. No em-dashes in body prose. No nominalisations where a verb works.
• Plain English. Anglo-Saxon over Latinate. Acronyms spelled on first use, every document. Translatability test for ECA / SSA / MENA readers.