coreyhaines31/marketingskills / skills/public-relations
public-relations
VerifiedWhen the user wants press and media coverage, or asks what to say to a reporter, whether a story is newsworthy, or how to handle a crisis: pitches, press releases, media lists, HARO, op-eds, podcast guest prep, newsjacking (public relations, not pull requests). For directories, see directory-submissions. For launches, see launch. For cold outreach to prospects, see cold-email.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 9.6 KB |
| evals/evals.json | config | 5.2 KB |
| references/crisis-communications.md | markdown | 8.4 KB |
| references/journalist-pitching.md | markdown | 10.8 KB |
| references/media-outlets.md | markdown | 9.1 KB |
| references/newsjacking.md | markdown | 8.4 KB |
| references/podcast-guest-prep.md | markdown | 5.5 KB |
| references/press-platforms.md | markdown | 6.7 KB |
| references/press-releases.md | markdown | 7.5 KB |
| references/statement-templates.md | markdown | 5.5 KB |
| references/story-angles.md | markdown | 6.4 KB |
evals/evals.json
View on GitHub{
"skill_name": "public-relations",
"evals": [
{
"id": 1,
"prompt": "We're a tiny 2-person SaaS competing against Notion in the docs space. We don't have any funding news or a big milestone. How do we get press coverage?",
"expected_output": "Should route to references/story-angles.md. Should recognize there's no traditional news hook and instead fit the situation to one of the three story angles, most naturally David vs Goliath (own your size against an incumbent, cite the Superhuman vs Gmail exemplar). Should mention the other two angles — Founding Story (Pieter Levels '12 startups in 12 months,' build-in-public, share revenue) and Have an Enemy (the enemy is a broken system, not the competitor — never 'Notion is bad,' and reference HEY vs Apple's App Store 2020). Should reinforce 'the story is not your product' and that a milestone alone isn't a story — milestone with narrative is. Should note the compound effect (authority, relationships, AI-citation surface) over the traffic bump. May suggest turning proprietary data into a data story and building media relationships before pitching. Should not invent fake news or suggest attacking the competitor by name.",
"assertions": [
"Routes to references/story-angles.md",
"Identifies David vs Goliath as the fitting angle with the Superhuman vs Gmail exemplar",
"Names all three angles (Founding Story, David vs Goliath, Have an Enemy)",
"Clarifies the enemy is a broken system, not a named competitor (HEY vs Apple App Store 2020)",
"Cites Pieter Levels '12 startups in 12 months' for Founding Story",
"Reinforces 'the story is not your product'",
"Notes milestone-with-narrative and/or the compound effect over the traffic bump",
"Does not fabricate news or suggest smearing the competitor by name"
],
"files": []
},
{
"id": 2,
"prompt": "We just found out someone accessed our customer database last night. We don't know yet what data they got. People are starting to ask on Twitter. Write us a statement.",
"expected_output": "Should route to references/crisis-communications.md and references/statement-templates.md. Should classify this as Tier 4 (Severe) and treat it accordingly: legal counsel reviews before anything beyond a holding statement, exec-level spokesperson, dedicated incident page. Should tell the user to loop in legal counsel before admitting fault or publishing, and note that breach notification deadlines and required content vary by jurisdiction without asserting a specific deadline as fact. Should draft a holding statement that acknowledges the incident, says it is being investigated, and commits to a specific next-update time, with no claims about cause, scope, or what data was or wasn't exposed. Should not admit fault or claim 'no sensitive data was affected' before security confirms scope. Should cover channel order (directly affected customers before public posts, internal note in parallel) and offer the full breach notification template for once scope is confirmed.",
"assertions": [
"Classifies the incident as Tier 4 (Severe) using the severity tiers",
"Tells the user to loop in legal counsel before admitting fault or publishing beyond a holding statement",
"Notes breach notification obligations vary by jurisdiction without stating a specific deadline as fact",
"Drafts a holding statement with a specific next-update time",
"Holding statement makes no claims about cause, scope, or what data was or wasn't exposed",
"Does not admit fault before the facts are known",
"Addresses channel order: affected customers and staff before or alongside public posts"
],
"files": []
},
{
"id": 3,
"prompt": "We're launching v2 of our project management app next Tuesday. Write a press release and tell me where to send it. Budget is a few hundred dollars.",
"expected_output": "Should route to references/press-releases.md. Should draft a release in the standard structure (headline naming the company and event, dateline, a lead that stands alone, specific numbers, a quote that gives a reason, About paragraph, named media contact) under 600 words, with no banned words. Should tell the user to post it on their own newsroom page first. Should present distribution as options that fit the budget (free newsroom or PRLog, a low-cost wire such as EIN Presswire, an API option) with tradeoffs, without calling any one option the best. Should say plainly that wire distribution produces syndicated copies rather than reporter-written articles, and suggest pitching journalists alongside the release.",
"assertions": [
"Drafts a release with a headline that names the company and the event",
"Release has a named media contact and an About paragraph",
"Recommends posting on the user's own newsroom page first",
"Lists more than one distribution option within budget, including a free one",
"Does not call any single distribution service the best",
"States that wire distribution yields syndicated copies, not coverage",
"Suggests pitching journalists in addition to the release"
],
"files": []
}
]
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |