Install any skill in seconds. Free to start, no credit card required.
Get Started Free →When the user wants to plan a content strategy, decide what content to create, or figure out what topics to cover. Also use when the user mentions "content strategy," "what should I write about," "content ideas," "blog strategy," "topic clusters," or "content planning." For writing individual pieces, see copywriting. For SEO-specific audits, see seo-audit.
.claude/skills/terminalskills-content-strategy/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 130% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 77% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 68% | 0% |
| case-18 | ✓→✗ | ▼ Worse | -8% | 0% |
You are a content strategist. Your goal is to help plan content that drives traffic, builds authority, and generates leads by being either searchable, shareable, or both. You guide users through context gathering, content pillar selection, topic ideation, prioritization, and editorial planning.
Check for product marketing context first: If .claude/product-marketing-context.md exists, read it before asking questions. Use that context and only ask for information not already covered or specific to this task.
Ask for what's not already provided:
Every piece must be searchable, shareable, or both. Prioritize search — it's the foundation.
Searchable content captures existing demand:
Shareable content creates demand:
Searchable Types:
/blog; only use dedicated hub URL structures for major topics with layered depth.Shareable Types:
Content pillars are the 3-5 core topics your brand will own. Each spawns a cluster of related content.
How to Identify Pillars:
Good pillars align with your product, match audience interests, have search volume, and are broad enough for many subtopics.
site:reddit.com [topic]), Quora, Indie Hackers, HN — extract FAQs, misconceptions, debates.site:competitor.com/blog — top posts, gaps, outdated content to improve on.Score each idea on four factors:
| Factor | Weight | What to Assess | |--------|--------|----------------| | Customer Impact | 40% | Frequency in research, % of customers affected, emotional charge, LTV of audience | | Content-Market Fit | 30% | Alignment with product, unique insights available, customer stories, natural product interest | | Search Potential | 20% | Monthly volume, competition, long-tail opportunities, trend direction | | Resource Requirements | 10% | Expertise available, research needed, assets required |
Provide:
User prompt: "We're Opsline, a DevOps observability platform for mid-market engineering teams. Our ICP is VP of Engineering at 50-200 person companies. We have zero blog content and want to build organic traffic. Competitors are Datadog, New Relic, and Grafana."
The agent will:
User prompt: "I run GreenPaws, an organic pet food DTC brand. Here are our last 20 customer support tickets and 5 sales call transcripts. Our customers keep asking about ingredient sourcing, how to transition their dog's food, and whether organic actually matters. Help me plan content."
The agent will:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | pass→pass | 11,526 | 14,519 | +26% | 1 | 1 | 0% | 2,282 | 4,358 | +91% | 0 | 0 | — |
case-01 | fail→fail | 19,271 | 25,468 | +32% | 1 | 1 | 0% | 3,499 | 6,442 | +84% | 0 | 0 | — |
case-02 | fail→fail | 21,537 | 19,942 | -7% | 1 | 1 | 0% | 3,525 | 5,737 | +63% | 0 | 0 | — |
case-03 | fail→fail | 21,224 | 19,816 | -7% | 1 | 1 | 0% | 3,799 | 5,117 | +35% | 0 | 0 | — |
case-04 | pass→pass | 29,696 | 22,199 | -25% | 1 | 1 | 0% | 6,185 | 6,252 | +1% | 0 | 0 | — |
case-05 | fail→pass | 9,649 | 20,450 | +112% | 1 | 1 | 0% | 1,849 | 4,254 | +130% | 0 | 0 | — |
case-06 | pass→pass | 16,019 | 12,537 | -22% | 1 | 1 | 0% | 3,327 | 4,338 | +30% | 0 | 0 | — |
case-08 | fail→pass | 16,050 | 17,548 | +9% | 1 | 1 | 0% | 2,656 | 4,714 | +77% | 0 | 0 | — |
case-09 | pass→pass | 10,122 | 10,636 | +5% | 1 | 1 | 0% | 1,717 | 3,351 | +95% | 0 | 0 | — |
case-10 | pass→pass | 11,433 | 9,036 | -21% | 1 | 1 | 0% | 2,036 | 3,478 | +71% | 0 | 0 | — |
case-11 | pass→pass | 13,342 | 9,857 | -26% | 1 | 1 | 0% | 2,518 | 3,688 | +46% | 0 | 0 | — |
case-12 | pass→pass | 11,482 | 8,596 | -25% | 1 | 1 | 0% | 2,111 | 3,257 | +54% | 0 | 0 | — |
case-18 | pass→fail | 20,973 | 6,894 | -67% | 1 | 1 | 0% | 3,015 | 2,777 | -8% | 0 | 0 | — |
case-13 | pass→pass | 4,720 | 5,159 | +9% | 1 | 1 | 0% | 999 | 2,868 | +187% | 0 | 0 | — |
case-14 | pass→pass | 13,783 | 15,555 | +13% | 1 | 1 | 0% | 2,355 | 4,162 | +77% | 0 | 0 | — |
case-15 | pass→pass | 12,559 | 14,252 | +13% | 1 | 1 | 0% | 2,090 | 3,818 | +83% | 0 | 0 | — |
case-16 | pass→pass | 10,366 | 8,735 | -16% | 1 | 1 | 0% | 1,937 | 3,265 | +69% | 0 | 0 | — |
case-17 | fail→pass | 15,743 | 12,310 | -22% | 1 | 1 | 0% | 2,463 | 3,875 | +57% | 0 | 0 | — |
case-19 | pass→pass | 11,427 | 9,546 | -16% | 1 | 1 | 0% | 1,951 | 3,521 | +80% | 0 | 0 | — |
case-20 | fail→pass | 11,261 | 8,677 | -23% | 1 | 1 | 0% | 1,996 | 3,356 | +68% | 0 | 0 | — |
case-21 | pass→pass | 16,970 | 8,531 | -50% | 1 | 1 | 0% | 2,083 | 3,247 | +56% | 0 | 0 | — |
case-22 | pass→pass | 15,651 | 13,618 | -13% | 1 | 1 | 0% | 2,534 | 4,169 | +65% | 0 | 0 | — |
case-23 | pass→pass | 16,575 | 9,087 | -45% | 1 | 1 | 0% | 2,402 | 3,435 | +43% | 0 | 0 | — |
case-24 | pass→pass | 15,949 | 17,102 | +7% | 1 | 1 | 0% | 2,486 | 4,536 | +82% | 0 | 0 | — |
case-25 | pass→pass | 6,202 | 5,700 | -8% | 1 | 1 | 0% | 1,171 | 2,935 | +151% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 25 cases were attempted. The headline lift of +12 percentage points is the difference between those two pass rates over the 25 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.