Argument craft
Apply argument-craft whenever the user needs to communicate a recommendation, persuade stakeholders, or structure a case for a decision. Triggers on phrases like "help me make the case for", "structure this argument", "write an executive summary", "how do I present this?", "convince the team", "frame this for leadership", "what's the narrative?", "how do I pitch this?". Use proactively when the user has completed analysis via decision-synthesis or scenario-planning but hasn't yet structured the communication.From its SKILL.md
npx -y skills add AndurilCode/craftwork --skill argument-craftAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 7 stars7 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
8.2 KB, as published. Nobody here has run it
Argument Craft
Same evidence, framed differently, produces different outcomes. Structure the case so the right people act on it.
When to Use
- User has a recommendation needing stakeholder presentation
- decision-synthesis output must become memo, deck, email, or conversation
- Justifying resource ask, scope change, or strategic pivot
- Someone needs convincing and framing is unclear
- execution-planning surfaces resource needs requiring buy-in (Contract C handoff)
Core Methodology
Step 1: Define the Audience
- Who: role, seniority, function
- What they care about: success metrics, priorities, constraints (CFO = ROI/risk; eng lead = tech debt/capacity)
- Prior belief: leaning toward, against, or no opinion — determines reinforce/convert/educate
- What changes their mind: data, authority, precedent, narrative, risk framing
- Identity stake: if recommendation implies they were wrong, expect resistance proportional to public commitment
When audience comes from stakeholder-power-mapping, use those motivations directly. From scenario-planning (Contract O), identify which stakeholders care about which scenarios.
Step 2: Formulate the Core Claim
One sentence. Must be:
- Specific: "Migrate to Kubernetes by Q3" not "modernize infrastructure"
- Actionable: names what to do, approve, or fund
- Falsifiable: someone could disagree — otherwise it's a platitude
From decision-synthesis (Contract A): recommended option + primary reason. From scenario-planning (Contract O): the robust strategy.
Step 3: Build the Argument Structure
Pyramid Principle (Minto): claim, then reasons, then evidence.
Core Claim
├── Reason 1 (strongest for this audience)
│ ├── Evidence A
│ └── Evidence B
├── Reason 2
│ ├── Evidence C
│ └── Evidence D
└── Reason 3 (addresses likely objection)
├── Evidence E
└── Evidence F
Rules:
- 2-4 reasons (fewer = unsupported; more = unfocused)
- Order by audience priority, not your logical sequence
- Each reason must independently support the claim — if removing one collapses it, that's a chain (fragile), not a pyramid
- Evidence types: data, precedent, authority, narrative, risk
From decision-synthesis scoring matrix: top criteria → reasons; scores → evidence.
Step 4: Map and Pre-empt Objections
Address counterarguments within the argument, not after.
For each objection:
- Steelman it — strawmen insult the audience
- Classify: factual (resolve with data), values (acknowledge trade-off), or fear (mitigate risk)
- Respond: concede + redirect. "Yes, upfront cost is higher — which is why the 18-month payback matters."
From decision-synthesis: main trade-off becomes primary objection. From scenario-planning (Contract O): weakest scenario becomes the objection.
Step 5: Select the Framing
| Frame | When to use | Example |
|---|---|---|
| Gain | Risk-tolerant, opportunity-seeking | "Unlocks $2M in new revenue" |
| Loss | Risk-averse, protective | "Without this, we lose $2M to competitors" |
| Narrative | Values stories over data | "When Company X faced this..." |
| Data-first | Analytical, skeptical of rhetoric | "Numbers show 3:1 return" |
| Vision | Strategic, long-horizon | "In three years, this positions us to..." |
| Problem-urgency | Doesn't yet agree there's a problem | "Every week we delay costs us..." |
When in doubt: problem-urgency (establish need) → gain (establish value).
Step 6: Adapt to the Medium
- Memo (1-2 pages): claim in first paragraph; reasons as headers; evidence as bullets; ask at end
- Presentation: one claim per slide; visuals not paragraphs; each slide answers "so what?"
- Email (3-5 paragraphs): subject = claim; opener = context + claim; middle = strongest reason + evidence; close = ask + deadline
- Conversation: context-set ("I want to discuss X because Y"), state claim, pause, adapt
- Slack (<200 words): bolded claim, 2-3 bullet reasons, one-line ask, link to memo
Step 7: Draft Opening and Close
Opening earns attention. Close drives action.
Openings:
- Context-claim: "We've been evaluating X. The recommendation is Y, because Z."
- Problem-solution: "We're facing [problem]. The path forward is [solution]."
- Shared goal: "We all want [goal]. Here's how to get there."
Closings:
- Specific ask: "I'm requesting approval for X by [date]."
- Decision frame: "The choice is between A and B. I recommend A."
- Next step: "If you agree, the next step is [action]."
Never close with a summary. Close with what you need them to do.
Step 8: Prepare the Handoff
Capture for execution-planning (Contract B):
- Committed decision: what was agreed, precisely
- Stakeholder alignment: who's on board, who has reservations, who wasn't there
- Constraints surfaced: new constraints from feedback (budget, timeline, scope)
If argument surfaces resource need: prepare Contract C handoff back to execution-planning — what's needed, why, what fails without it.
Output Format
🎯 Audience Profile
- Audience: [who — role and function]
- They care about: [priorities and metrics]
- Prior belief: [supportive / neutral / opposed / uninformed]
- Persuaded by: [data / narrative / authority / risk framing]
📋 Core Claim
[One sentence: specific, actionable, falsifiable]
⚖️ Argument Structure
| # | Reason | Evidence | Evidence type |
|---|---|---|---|
| 1 | [strongest reason for this audience] | [evidence] | [data / precedent / authority / narrative / risk] |
| 2 | [second reason] | [evidence] | [type] |
| 3 | [reason addressing likely objection] | [evidence] | [type] |
🛡️ Objection Map
| Objection | Type | Response |
|---|---|---|
| [steelmanned objection 1] | [factual / values / fear] | [concede + redirect] |
| [steelmanned objection 2] | [type] | [response] |
🧠 Framing & Medium
- Frame: [gain / loss / narrative / data-first / vision / problem-urgency]
- Medium: [memo / presentation / email / conversation / Slack]
- Rationale: [why this fits the audience]
🏆 Draft Opening
[2-3 sentence opening paragraph or slide structure]
📋✅ Downstream Handoff (Contract B)
- Committed decision: [what was decided]
- Stakeholder alignment: [who agreed, who has reservations]
- Constraints surfaced: [new constraints]
Thinking Triggers
- "If I had 30 seconds, what one sentence would I say?"
- "What's the strongest version of the argument against me?"
- "Am I leading with what matters to them, or to me?"
- "Does every reason independently support the claim?"
- "What does the audience stand to lose if they accept this?"
Common Traps
| Trap | What goes wrong |
|---|---|
| Leading with process | "We evaluated 5 options on 8 criteria..." — they want the answer, not the journey |
| Burying the claim | Building up instead of stating first. Executives stop reading. |
| Strawmanning objections | Weak counterarguments destroy credibility. Steelman the strongest. |
| One-size framing | Same frame for every audience wastes the strongest evidence. |
| Missing the ask | Analysis without request. Audience nods and moves on. Always close with what you need. |
| Chain arguments | Each point depends on the previous. Reject step 2, steps 3-5 collapse. Use a pyramid. |
| Advocacy without analysis | Structuring weak conclusions makes them harder to challenge — harmful. This skill structures communication of a sound conclusion; it does not validate it. |