Free skills · open source
Skills for AI agents, tested in public.
34 skills across 7 capabilities from the AI Marketing Bench, plus 2 standalone repos. Each skill is a folder your agent loads when a task calls for it. By Ashleah Wilson.
All skills
All 36 skills
- Outboundv3.2.1
Reach out
Drafts B2B prospecting outreach, including cold and warm emails, LinkedIn connection notes and DMs, follow-ups and short sequences, for a named contact, an account, a list or a buying signal.
MIT - Outboundv1.2.1
Handle reply
Handles replies to B2B prospecting outreach.
MIT - Outboundv1.2.1
Competitor intel
Builds a competitor profile and a sales battlecard from material the user supplies, such as saved competitor pages, pricing pages, review excerpts, win/loss exports and call notes.
MIT - Outboundv1.2.1
ICP
Builds and maintains the ideal customer profile and buyer personas from evidence (brand memory, a customer list or CRM export, win/loss and call notes) and writes a points-based scoring rubric another skill can apply to a…
MIT - Outboundv1.2.2
Build list
Builds a scored prospect account list from data the user already has, such as CRM or database CSV exports, directory or event attendee exports, and lists of URLs.
MIT - Outboundv1.2.2
Map contacts
Maps the buying committee at one target account from information the user supplies, such as a saved team page, an org chart, CRM contacts, a LinkedIn export, call notes or job postings.
MIT - Marketing Auditv1.2.2
Marketing audit
Audits a company's marketing from its website and public footprint and returns a scored Marketing Health report (0–100, graded A–F) across positioning, acquisition, conversion, brand and measurement, calibrated to the business…
MIT - Marketing Auditv1.2.1
Positioning deep dive
Runs a positioning deep dive on a company's website and writes a positioning report with the three findings that matter most, quoted evidence for each, two or three rewritten hero headlines and subheads built only from facts on…
MIT - Marketing Auditv1.2.2
Acquisition audit
Audits how people find a business, from its web pages and writes an acquisition report that leads with the three findings that matter most and a 48-hour / 2-week / 3-month plan.
MIT - Marketing Auditv1.2.2
Conversion audit
Audits a website's conversion paths page by page from saved pages and returns a conversion report with the three findings that matter most, a 48-hour / 2-week / 3-month plan, a short test backlog (hypothesis, change, metric) and…
MIT - Marketing Auditv1.1.3
Brand audit
Audits how a company's brand shows up in words across the pages and channel pieces supplied (site pages, posts, emails, newsletters) and writes a brand report with the three findings that matter most, a 48-hour / 2-week / 3-month…
MIT - Marketing Auditv1.2.1
Measurement audit
Audits what a website can actually measure and returns a measurement report that leads with the top three findings, a 48-hour / 2-week / 3-month plan, a tracking plan (events, their parameters, which are conversions, where each…
MIT - Brandv1.2.1
Brand memory
Sets up and maintains a brand memory, a brand/ folder of small plain-text files (offer, audiences, positioning, proof, voice, claims, competitors, senders, rules) where every fact carries a source and a status, so every other…
MIT - Brandv1.2.1
Brand voice codify
Builds or updates a brand's voice guide (brand/voice.md, the brand-memory format) from evidence, such as samples of its best writing, examples it dislikes, before-and-after edits a founder made to drafts, and feedback like "we…
MIT - Brandv1.1.1
De-slop
Removes what makes writing read as AI-generated, from empty claims and borrowed rhetorical moves down to stock words, without inventing a single detail, and catches the overcorrections (fragments, one-line paragraph stacks,…
MIT - Brandv1.5.1
Brand voice gate
Checks a draft before it ships and fixes it in two passes, first removing what makes writing read as AI-generated or generic, then applying the brand's own voice, words, channel rules and approved claims from brand memory, and…
MIT - Content Opsv1.2.1
Content brief
Turns a content request (a topic, a goal, a rough idea or a campaign need) plus brand memory into a brief a writer or another skill can work from, saved in briefs/ in the format content-qa-gate later checks the finished piece…
MIT - Content Opsv1.2.2
Content QA gate
Runs the last check on a finished content piece before it's published, such as a blog post, landing page, newsletter or article, and returns a verdict (Publish, Fix first, or Hold) with every problem located and a concrete fix.
MIT - Content Opsv1.2.1
Editorial review cycle
Runs the review loop on a content draft between writing and publishing.
MIT - Content Opsv1.2.2
Content audit
Audits an existing content library (blog posts, guides, landing pages) from what the user supplies, such as a CSV export of URLs with metrics, a sitemap or a folder of the posts, and decides for every piece whether to keep,…
MIT - Content Opsv1.2.1
Email sequence
Writes lifecycle email sequences for people who opted in or are customers, including welcome and onboarding series, nurture after a lead magnet, trial-to-paid, re-engagement, post-purchase, event follow-up and newsletters.
MIT - Content Opsv1.2.1
LinkedIn content
Writes LinkedIn posts for a named person or a company page from the user's own material (a point of view, a story they tell, a result, notes from a call or talk) as a single post, a short series or a carousel outline.
MIT - Paid Adsv1.2.2
Google Ads audit
Audits a Google Ads account from CSV exports (campaigns, keywords, search terms, ads, conversion actions, settings), or read-only through a connected Google Ads tool, and returns a scored report (0–100, graded A–F) with the top…
MIT - Paid Adsv1.2.2
Google Ads build
Builds a Google Ads search campaign ready to import, or rebuilds one from a Google Ads audit's fixes list.
MIT - Paid Adsv1.2.1
Meta ads audit
Audits a Meta (Facebook and Instagram) ads account from the Ads Manager and Events Manager exports the user supplies (or read-only from a connected tool) and returns a scored report (0–100) that leads with the top three findings…
MIT - Paid Adsv1.2.1
Meta ads build
Builds a Meta (Facebook and Instagram) ad campaign ready to load, or rebuilds one from a Meta ads audit's fixes list, or checks and fixes existing ad drafts.
MIT - Team Opsv1.2.1
Skill author
Turns work that went well into a reusable Agent Skill that meets the Bench Standard, and checks an existing skill folder before it's shared.
MIT - Team Opsv1.2.1
AI workload map
Maps a marketing team's work to who or what should do it, from what the user supplies (a task inventory, job descriptions, a time log or survey, process docs, the tools in use).
MIT - Team Opsv1.2.1
Campaign intake
Turns a messy campaign request (an email, Slack thread, meeting notes or half-filled form) into a structured intake saved in intake/, covering objective and the business result it serves, audience matched to brand memory, offer,…
MIT - Team Opsv1.2.1
Approval routing
Works out who has to approve a piece of marketing before it goes out, in what order and by when, and tracks where it stands.
MIT - Team Opsv1.2.2
Agency brief
Writes the brief a company sends to an outside agency, freelancer or contractor, and reviews what comes back.
MIT - Claims & Compliancev1.2.2
Claims review
Reviews marketing material (web pages, ads, emails, decks, social posts, packaging and labels) for claims substantiation.
MIT - Claims & Compliancev1.2.1
Accessibility review
Reviews the HTML of a marketing web page, landing page or HTML email for accessibility against WCAG 2.2 level AA, and returns a recommendation with every issue tied to its success criterion, the exact element and a concrete fix.
MIT - Claims & Compliancev1.2.1
Disclosure check
Checks marketing content for the disclosures it needs and whether each one is clear and placed where a reader will see it, then gives the exact fix text and placement.
MIT - Searchv5.2.0
NoW Search Visibility Audit & Growth Roadmap: SEO · GEO · AEO
Audits any website for SEO, GEO (getting cited by ChatGPT, Perplexity, Google AI Overviews and AI Mode, Gemini, Claude) and AEO (featured snippets, People Also Ask, voice), then builds a growth roadmap: an AI Visibility Score out…
CC BY 4.0 - Chief of Staffv1.3.0
AI Chief of Staff
An AI chief of staff for a leader: morning briefs, meeting prep, meeting-to-actions, a commitment ledger, and a weekly review, grounded in files the leader controls.
MIT
How the Bench skills were tested
Every AI Marketing Bench skill is built and tested to the Bench Standard and ships with its test results:
- It reads one shared brand memory, traces every claim to a source, checks its own output before you see it, and knows when to push back.
- It’s tested with the skill, without it, and against the public skill it replaces, on a large model and a small one, with blind grading.
- The scorecard is published in each skill’s evals folder, including the tests it lost. Where a skill hasn’t fully cleared the bar yet, the scorecard says so.
Install a capability or one skill
Claude Code, one capability
/plugin marketplace add ashleahwilson/ai-marketing-bench
/plugin install outbound@ai-marketing-benchAny agent that supports Agent Skills, one skill
npx skills add ashleahwilson/ai-marketing-bench --skill reach-outEach skill page has the exact lines for that skill. The two standalone repos have their own install steps on their pages.