Skill Claude CodeCodex
Simulate a LinkedIn reaction to a post from your personas — show them a post they'd plausibly see on LinkedIn, have each one comment / react / repost / scroll past the way they actually would. LinkedIn dynamics are professional, longer-form, more sincere, with strong identity-signaling and weaker dunking than X.…
Skill Claude CodeCodex
Simulate a cultural market — give the personas a set of items (songs, products, posts, ideas, messages) and have them pick / rate / upvote across multiple rounds, optionally seeing each other's choices. Run multiple parallel "worlds" to show how social influence makes outcomes unpredictable — the same item can be a…
Skill Claude CodeCodex
Run MaxDiff (Maximum Difference Scaling, aka best-worst scaling) with the personas — show each persona small sets of items and ask them to pick the most and least important from each set. Produces a robust ranked list of importance across the full item set. Use when the user says "/persona-max-diff", "MaxDiff this…
Skill Claude CodeCodex
Engineer and test "memes" — creative / message / positioning units — against your personas as a fast, cheap synthetic RCT. Built on the Concept → Theme → Variant framework: it concentrates rigor at the Concept level (the one-way-door decisions), uses an orthogonal (pairwise) array to test many concepts and elements in…
Skill Claude CodeCodex
Meta-skill — wraps any dynamic-system study (focus group, council, town, market, x-post / linkedin-post / hacker-news with reply rounds) and runs it N times. One run is the control (social-influence dynamics stripped out, personas act independently); the rest are full dynamic runs with the same starting conditions but…
Skill Claude CodeCodex
Manually add real data about an existing persona — a CSV of exchanges, or freeform "here's something I've seen this person say or do." Normalizes the input into universal asset rows ({context, question, answer, source}), appends them to ./.personas/assets/ /observations.jsonl (immutable, append-only), and does a light…
Skill Claude CodeCodex
Personas of Thought (PoT) — answer a question by having each persona respond independently from their own background, then combine the responses into a single anonymous joint answer written as if the personas collaborated on it. A diverse-perspective reasoning technique in the lineage of Chain-of-Thought /…
Skill Claude CodeCodex
Get slide-by-slide audience feedback on a presentation from your personas, then a curated change list. Feed it any mix of a deck (PPTX / PDF / Keynote / Slides export), a run-of-show or table of contents, and/or a transcript or video of you presenting — the personas sit through the whole talk as your audience, react…
Skill Claude CodeCodex
Rebuild a persona's ./.personas/ .md from its existing assets WITHOUT re-pulling any sources — re-synthesize the body and re-select the ## Examples (corrections and observations prioritized). Use when the user says "/persona-refresh", "refresh ", "rebuild ", "re-pick the examples", "regenerate the doc from the…
Skill Claude CodeCodex
Run structured market research with your personas. Parent skill that catalogs the available study types (interview, JTBD interview, survey, focus group, council, user testing, presentation feedback, tree test, A/B test, concept test, TURF, conjoint, MaxDiff, Van Westendorp, brand tracking, social listening…
Skill Claude CodeCodex
Run an asset (marketing copy, landing page, ad, pricing page, feature pitch, value prop, anything aimed at a customer) past a panel of target personas and return code-review-style feedback — specific, line-level, actionable. Use when the user says "persona-review this", "/persona-review", "/persona-review "…
Skill Claude CodeCodex
Get brutally honest, unfiltered, no-punches-pulled feedback from the personas on an idea, plan, asset, pitch, or product. Explicitly prompts personas to tear it apart — every objection they'd ordinarily soften comes out sharp. Output is a roast per persona plus the most lethal critiques synthesized. Use when the user…
Skill Claude CodeCodex
Live rehearsal against a persona who plays the counterpart — a journalist interviewing you, a prospect you're pitching, a boss you're meeting with. You play yourself; the persona stays in character and pushes back; you get coached at the end. Use when the user says "/persona-roleplay", "roleplay", "let me practice a…
Skill Claude CodeCodex
Simulate Slack reactions from your personas to a message — in a specific channel context (team-internal, community Slack, customer Slack), with the right Slack-native dynamics (emoji reactions, thread replies, "🧵", DMs). Outputs the simulated thread plus an analysis of how the message would actually land. Use when…
Skill Claude CodeCodex
Simulate social listening — have personas write social-media-style posts about a topic (not your brand) and measure whether, where, and how your brand surfaces organically in their posts. Outputs mention rate, sentiment when mentioned, share-of-voice vs. named competitors, and the context in which the brand came up.…
Skill Claude CodeCodex
Run a structured survey across multiple personas — a fixed question set asked to each persona independently, then aggregated. Use when the user says "/persona-survey", "survey the personas about X", "ask everyone these questions", or wants breadth across the panel rather than depth on one viewpoint.
Skill Claude CodeCodex
The maximum-effort idea exploration — point the entire persona arsenal at one idea and stretch it in every direction at once. Not a critique and not a vote; a no-holds-barred divergent-then-convergent exploration that generates stretch directions, fans out generative + reactive + adversarial passes across all personas…
Skill Claude CodeCodex
Run a generative-agent simulation — drop the personas into a 2D town, let them move and talk to each other over discrete time ticks, and watch information (an idea, a rumor, a product) propagate through the population. Visualized in real-time via a Phaser.js HTML page. Use when the user says "/persona-town", "simulate…
Skill Claude CodeCodex
Test information architecture / navigation with the personas — given a hierarchical sitemap or menu structure and a list of tasks, see where each persona would click to complete each task. Outputs per-task success rate, average path depth, and the wrong-path patterns that indicate IA problems. Use when the user says…
Skill Claude CodeCodex
Run TURF analysis (Total Unduplicated Reach and Frequency) with the personas — given a list of candidate options (features, messages, channels, plan tiers), find the smallest subset that "reaches" the largest share of personas, where each persona is reached if at least one option in the subset appeals to them. Use…
Skill Claude CodeCodex
Run a usability/user-testing study with personas reacting to a real UI flow — drive through a URL with agent-browser, capture screenshots at each step, and have personas react to each screen. Also supports video (chunked via ffmpeg). Use when the user says "/persona-user-test", "have the personas walk through this…
Skill Claude CodeCodex
Run a Van Westendorp Price Sensitivity Meter with the personas — each persona answers four price questions (too cheap, cheap, expensive, too expensive) for a product or service. Outputs the four price curves, the optimal price point, and the range of acceptable prices. Use when the user says "/persona-van-westendorp"…
Skill Claude CodeCodex
Simulate an X (Twitter) reply thread from your personas — show them a post they'd plausibly see in their feed, have each one reply / quote-tweet / like / scroll past the way they actually would on X. Outputs the reply thread plus an analysis of engagement shape (ratio risk, viral potential, dominant takes). Use when…