Consultation Summary

v1.0.0

consultation-summary · released 2026-09-05

Synthesizes raw public submissions, survey responses, or consultation feedback into a themed "what we heard" summary — recurring themes, representative quotes, and where views diverged — without silently dropping a dissenting view into the majority…

Consultation Summary

name:
consultation-summary
description:
Synthesizes raw public submissions, survey responses, or consultation feedback into a themed "what we heard" summary — recurring themes, representative quotes, and where views diverged — without silently dropping a dissenting view into the majority theme it disagrees with. Trigger when a policy or engagement officer needs a consultation report, "what we heard" document, or submissions analysis.
version:
1.0.0

Consultation "What We Heard" Summary

A "what we heard" report is a distinct deliverable from "what we'll do" — it reports faithfully, it doesn't yet decide. This skill's discipline is keeping a strongly-held minority view visible as its own theme, rather than folding it into the majority position it disagrees with.


What this skill needs

  • The submissions or feedback — pasted, summarized, or described.
  • The question(s) the consultation asked.

Step 1: Group into themes

Cluster submissions into recurring themes, noting approximately how many or what proportion raised each one.

Completion criterion: every theme has an approximate weight attached — not just a list of topics with no sense of how common each was.


Step 2: Pull representative quotes

Select 1–2 representative quotes per theme, anonymised per your consultation's privacy commitment.

Completion criterion: every major theme has at least one quote that a reader could recognise as representative, not a paraphrase presented as a quote.


Step 3: Name the divergence

Where views split or a minority position was strongly held, give it its own theme rather than merging it into the majority theme it disagrees with.

Completion criterion: any submission that disagrees with the majority view on a point appears in a distinct, named theme — never silently absorbed.


Output format

Per theme: theme name, approximate weight, 1–2 quotes, one-paragraph summary. Divergent themes clearly labelled as such.


Gotchas

  • This reports what was said — keep it separate from any recommendation on what to do about it.
  • Never invents a theme that isn't present in the source submissions.

Evidence base

  • IAP2, Spectrum of Public Participation — backs treating "what we heard" as a distinct deliverable from "what we'll do," matching the consult/ involve stages of the spectrum rather than jumping ahead to a decision.
  • Australian Government Consultation Principles — backs proportionate, transparent reporting of feedback, including minority views.
  • OECD, Recommendation on Open Government (2017) — backs publishing consultation outcomes as an accountability practice this summary supports.

FOI Response Drafter

v1.0.0

foi-response-drafter · released 2026-09-05

Drafts the structure of a freedom-of-information or public-records request response — decision, reasons, exemptions cited, and review rights — from case notes and a decision already made, leaving the actual release, redaction, and exemption decision…

FOI Response Drafter

name:
foi-response-drafter
description:
Drafts the structure of a freedom-of-information or public-records request response — decision, reasons, exemptions cited, and review rights — from case notes and a decision already made, leaving the actual release, redaction, and exemption decision to the authorised officer. Trigger when an FOI, GIPA, or public-records officer needs to draft a decision letter or response to an information request.
version:
1.0.0

FOI/Public-Records Response Drafter

Name the source, never the decision: this skill needs the officer's actual release decision — what's released, what's withheld, and which exemption applies to each withheld item — already made. It never chooses an exemption or decides what to release; that judgement, including its legal risk, stays with the authorised decision-maker.


What this skill needs

  • The request itself and the items identified as responsive to it.
  • For each item: the release decision, and the exemption or ground cited for anything withheld.

If the decision inputs aren't present:

"For each item, what's the release decision, and which exemption or ground applies to anything withheld?"

Do not proceed until this is supplied — drafting around a guessed decision is the failure mode this skill exists to avoid.


Step 1: Confirm the decision is complete

Check that every item identified as responsive has a stated decision and, where withheld, a named exemption.

Completion criterion: no item is missing a release/withhold decision.


Step 2: Draft the letter

Structure: decision summary, item-by-item schedule (with the exemption cited per withheld item), reasons, and review/appeal rights with the applicable timeframe.

Completion criterion: every withheld item in the draft cites the exemption the officer specified — never one inferred by this skill.


Step 3: Check completeness

Verify every item in the original request appears in the schedule — either released or withheld with reasons — and that review rights and their timeframe are stated.

Completion criterion: every item in the request has a decision reflected in the draft, and review rights appear once, correctly scoped to the jurisdiction's requirements.


Output format

Decision summary paragraph, item-by-item schedule table (Item | Decision | Exemption/Reason), review rights paragraph.


Gotchas

  • Never selects or infers an exemption on its own — that call belongs entirely to the authorised decision-maker.
  • This drafts the letter around a decision already made, not the decision itself.

Evidence base

  • OAIC FOI Guidelines — backs the decision/reasons/exemption/review-rights structure required for a valid Australian FOI decision letter.
  • UK ICO guidance and US DOJ FOIA Guide — the same structural requirement appears in UK and US regimes, backing this as a common shape across jurisdictions rather than one-off local practice.

Governance Meeting Minutes

v1.0.0

governance-meeting-minutes · released 2026-09-05

Structures minutes and an action register from a board, committee, or governance meeting transcript or notes — decisions and any dissent kept separate from general discussion, actions tied to an owner and due date.

Governance Meeting Minutes

name:
governance-meeting-minutes
description:
Structures minutes and an action register from a board, committee, or governance meeting transcript or notes — decisions and any dissent kept separate from general discussion, actions tied to an owner and due date. Trigger when an officer needs minutes for a board, committee, steering group, or governance meeting.
version:
1.0.0

Governance Meeting Minutes

Governance minutes are a compliance record, not a discussion transcript. This skill's job is separating what was decided from what was merely discussed, and recording dissent against a decision rather than smoothing it into consensus that wasn't actually reached.


What this skill needs

  • The meeting transcript, notes, or dictation.
  • The agenda, if the meeting followed one.

Step 1: Separate discussion from decision

For every agenda item, write a one-line discussion summary and a clearly marked decision — or "no decision — deferred" if none was reached.

Completion criterion: every agenda item has an explicit decision or deferral marked; none are left implicit in the discussion summary.


Step 2: Build the action register

Pull actions only from what was actually agreed — not from suggestions raised but not adopted. Each action needs an owner and a due date.

Completion criterion: every action has a named owner and due date, or is flagged as missing one rather than left blank.


Step 3: Record dissent

Where a decision wasn't unanimous, note who dissented and, briefly, why — governance minutes exist partly to protect that record.

Completion criterion: any recorded disagreement with a decision appears in the minutes, not smoothed over into apparent consensus.


Output format

Per agenda item: Discussion (1–2 lines) / Decision (or deferral) / Dissent (if any). Followed by a consolidated Action Register table (Action | Owner | Due Date).


Gotchas

  • Minutes are a record, not a decision-making tool — an unclear or disputed decision gets flagged for the chair to clarify before minutes are confirmed, never resolved by this skill on its own judgement.

Evidence base

  • AICD, Good Governance Guides — backs the decision-and-dissent record as the core function of governance minutes, distinct from a discussion transcript.
  • UK Corporate Governance Code for central government departments — backs the action-with-owner-and-date discipline as an accountability control, not just good practice.

Ministerial Response

v1.0.0

ministerial-response · released 2026-09-05

Drafts a formal response to ministerial correspondence, an executive briefing request, or a constituent letter escalated to a minister's office — from talking points or case notes, in the issue/background/ response structure most correspondence…

Ministerial Response

name:
ministerial-response
description:
Drafts a formal response to ministerial correspondence, an executive briefing request, or a constituent letter escalated to a minister's office — from talking points or case notes, in the issue/background/ response structure most correspondence units require. Trigger when a policy or correspondence officer needs to draft a ministerial response, MC (ministerial correspondence) reply, or briefing note for sign-off.
version:
1.0.0

Ministerial Correspondence Drafter

A correspondence unit's job is "no surprises": every response is accurate, answers what was actually asked, and never states a position the department hasn't settled on. This skill drafts to that discipline — it flags where no settled position exists rather than inventing one to sound complete.


What this skill needs

  • The original letter or query.
  • Any existing talking points or department position.
  • The office's template or format, if one exists.

Step 1: Extract the actual question

Identify exactly what was asked or raised — no more, no less. A response that answers a broader question than the one asked reads as evasive; one that answers a narrower question misses the point.

Completion criterion: every question or issue raised in the original correspondence is listed before drafting begins.


Step 2: Draft the response

Structure: Issue / Background / Response (/ Recommendation, if this is briefing up to a decision-maker rather than closing out a reply directly). Cite the department's position where one has been supplied. Where none exists, write [Position to be confirmed with policy area] rather than inferring one.

Completion criterion: every question from Step 1 is either answered, or explicitly marked out of scope with a referral to the right area.


Step 3: Match the register

Adjust formality to the recipient — a minister's office expects a different register from a direct constituent reply. Keep both factual and courteous; neither becomes casual.

Completion criterion: the draft's tone is consistent throughout and matches the stated recipient.


Output format

Issue / Background / Response headings (plus Recommendation where applicable), each 2–5 sentences, in the office's template if supplied.


Gotchas

  • Never states a policy position that wasn't supplied — flag it for confirmation instead of inventing one that sounds plausible.
  • This drafts for sign-off; the actual decision and release stay with the officer and delegate.

Evidence base

  • Australian Government Style Manual — backs plain, direct correspondence structure and register-matching to the recipient.
  • APSC Values and Code of Conduct — backs the no-invented-position rule as an integrity requirement: impartial, accurate advice over a confident- sounding guess.
  • UK Cabinet Office, Guide to Handling Correspondence — backs the issue/background/response shape as the standard ministerial- correspondence structure across Westminster-system public services.

Plain Language Rewrite

v1.0.0

plain-language-rewrite · released 2026-09-05

Rewrites dense legislative, regulatory, or technical text into plain language for public-facing content — web copy, fact sheets, application guidance — without dropping a condition, exemption, or deadline the rewrite needs to keep.

Plain Language Rewrite

name:
plain-language-rewrite
description:
Rewrites dense legislative, regulatory, or technical text into plain language for public-facing content — web copy, fact sheets, application guidance — without dropping a condition, exemption, or deadline the rewrite needs to keep. Trigger when a communications or policy officer needs to simplify legislation, a regulation, technical guidance, or an internal document for public consumption.
version:
1.0.0

Plain Language Rewriter

Simplifying legal or technical text is easy to get wrong in one specific way: a condition, exemption, or deadline quietly disappears in the effort to make the sentence shorter. This skill audits for exactly that failure mode before it drafts a single simplified sentence.


What this skill needs

  • The source text.
  • The audience and channel — web page, fact sheet, letter, application guidance.

Step 1: Inventory every condition

Before rewriting, list every condition, exemption, deadline, or qualifier in the source text. This list is the checklist Step 3 verifies against.

Completion criterion: every "if," "unless," "must," "by [date]," and "except" in the source appears in this inventory.


Step 2: Rewrite in plain language

Short sentences, common words, active voice, second person where the channel allows it ("you must apply by," not "applicants are required to submit").

Completion criterion: no sentence exceeds roughly 20–25 words, and no undefined technical or legal term appears unexplained.


Step 3: Verify nothing was dropped

Check every item from Step 1 still appears in the rewrite. Where a condition had to be simplified enough that its legal meaning might have shifted, flag it for legal or policy review rather than resolving the tension yourself.

Completion criterion: every condition from Step 1 is traceable in the final text, or explicitly flagged as needing review before publication.


Output format

The rewritten text, followed by a short "Flagged for review" list of any condition whose simplification needs sign-off.


Gotchas

  • Simplifying legal text risks silently loosening or tightening a condition. This skill flags that risk rather than resolving it — a rewrite touching legal obligations should go back through legal or policy sign-off before it's published.

Evidence base

  • plainlanguage.gov (Plain Writing Act of 2010) — backs the sentence- length, active-voice, and common-word criteria applied in Step 2.
  • Australian Government Style Manual — the same plain-language standard for Australian public-facing content.
  • UK Government Digital Service, Content Design guidance — backs auditing for dropped conditions as the specific, named failure mode of legal-to- plain rewrites, which is what Step 1 and Step 3 exist to catch.

Policy Brief

v1.0.0

policy-brief · released 2026-09-05

Turns research notes, consultation findings, or a rough problem statement into a structured policy brief — problem, evidence, options with genuine trade-offs, and a recommendation — in the format policy and program areas use to brief decision-makers.

Policy Brief

name:
policy-brief
description:
Turns research notes, consultation findings, or a rough problem statement into a structured policy brief — problem, evidence, options with genuine trade-offs, and a recommendation — in the format policy and program areas use to brief decision-makers. Trigger when a policy officer needs to write a policy brief, options paper, or decision-ready advice note.
version:
1.0.0

Policy Brief Writer

A policy brief that presents one obviously-right option and two strawmen hasn't actually helped the decision-maker — it's dressed up advocacy. This skill requires a genuine trade-off behind every option, and a recommendation that names what it gives up, not just what it gains.


What this skill needs

  • The problem or issue.
  • Whatever evidence or consultation input already exists.
  • The decision the brief needs to support, if known.

Step 1: State the problem

One paragraph: who's affected, why it matters now, what happens if nothing changes.

Completion criterion: the problem statement names who's affected and why now, not just a general description of the issue area.


Step 2: Lay out the options

Present 2–4 options, each with the evidence behind it and a real trade-off — cost, risk, timeframe, equity, or political feasibility. Where evidence for an option is thin, say so rather than smoothing over the gap.

Completion criterion: every option has a stated trade-off, not just a list of advantages; thin evidence is flagged, not disguised.


Step 3: Make one recommendation

State a clear recommendation and name the trade-off it accepts — what you're giving up by choosing it over the alternatives.

Completion criterion: the recommendation is singular and explicitly names its accepted trade-off.


Output format

Problem / Options (with evidence and trade-off per option) / Recommendation, each section scannable in under a minute.


Gotchas

  • Never fabricates evidence or data to fill a gap — a brief with thin evidence says so.
  • Recommends; doesn't decide. The accountable officer makes the call.

Evidence base

  • ODI, Preparing Effective Policy Briefs — backs the problem/evidence/options/recommendation structure and the one-clear-recommendation discipline.
  • APS Policy Profession model — backs trade-off transparency as a core policy-craft skill, distinct from advocacy that hides the cost of its preferred option.