A symmetric detect-and-author skill for persuasion in copy. Same mechanism vocabulary runs both directions: every move you can author leaves a structural fingerprint a trained reader can name; every fingerprint a reader can name corresponds to a move an author chose to make. The catalog is the same; only the direction of use differs.
What's in the file:
- 12 convergent persuasion mechanisms (framing, anchoring, illusory truth, processing fluency, social proof, authority, liking, scarcity, reactance, reason-why, narrative transportation, identity priming) and 5 linguistic operators (presupposition, agency deletion, generic-you normativity, modality and asymmetric hedging, Linguistic Intergroup Bias). Each row carries: mechanism + FORBIDDEN trigger + author rule + detection signal + citation anchor.
- Hard ethics gate that fires per technique, not just at the top: truth-of-claim, instance-level disclosure-survives test, symmetry-of-choice, no-vulnerability-targeting (consequence-based for shipped-to-public copy), density cap (max 4 mechanisms above the fold / 8 across the page), input-sufficiency rule (refuses to fabricate placeholder data when the brief omits it).
- Quick-reference lookup table at the top for sub-five-minute triage.
- QUICK MODE workflow (4 steps) for deadline rewrites, distinct from full AUDIT MODE for compliance review.
- Six-heuristic detection kit (lateral source check, frame swap, presupposition negation test, accuracy reflex, technique-label scan, anchorage strip).
- Short-form surfaces section: how the gate fires on subject lines, ad headlines, push notifications.
- Two worked examples: one dense-violations teardown, one realistically-ambiguous B2B SaaS pricing block where most mechanisms PASS with substantiation log entries.
- Anti-patterns / discarded canon section explicitly refusing 13 zombie findings: Vicary popcorn, NLP, power posing, ego depletion, "losses loom 2x", color-emotion universals (red=urgency etc), Mehrabian 55-38-7 misapplication, strong Whorfianism, backfire effect, Hofstede copy localization, classic induced-compliance dissonance, social priming, "disclosure cures everything." Each entry names the killing citation.
- Jurisdictional scope and anti-laundering warnings: regulatory citations are flags not legal advice; this skill's gate is self-administered and is not a substitute for third-party audit.
- Production guardrails (6 hard rules) and OUTPUT DISCIPLINE step 0 with 7-question gate covering all mechanisms inline.
- 50+ primary-source references including 2026 meta-analyses (Henderson illusory truth) with DOIs.
Built for: copywriters, marketing operators, founders writing their own copy, agency content leads, dark-pattern auditors, political-comms reviewers, AI-content provenance reviewers, and LLM-powered copy tools that need a defensible mechanism set rather than vibes.
Why this isn't another Cialdini summary: replication-filtered (post-2010 multilab and preregistered scrutiny applied to every claim, with citations); per-row ethics gate that prevents the catalog being used as a manipulation manual; symmetric detect-and-author structure with an author-to-detector handoff; explicit refusal of 13 named zombie findings the popular canon still teaches.
Known limitations
- Identity-protective cognition is the ceiling on all detection heuristics. When the reader is highly motivated to believe a claim, the detection kit fires less reliably; the skill explicitly flags this and recommends delay-and-distance audits as the only defense.
- The ethics gate is self-administered. It is not a substitute for legal compliance review, third-party dark-pattern audit, or regulatory submission. The skill includes an explicit anti-laundering warning against citing it as a compliance defense.
- Regulatory citations (FTC, DSA, UCPD, CPRA) are jurisdictional flags. US-only / EU-only / California-only rules are noted as such; multi-jurisdiction copy needs counsel review against each applicable regime. The skill is craft heuristic, not legal advice.
- Citation accuracy is anchored to literature available as of 2026-05. Replication research evolves; a finding that survived as of file-shipping date may be reassessed. The discarded canon names what fell as of ship date.
- Coverage is Western academic and practitioner traditions (Cialdini, Aristotle, Sugarman, Hopkins, Ogilvy, Lakoff, Halliday). Non-Western persuasion frameworks (Confucian rhetoric, Indian Nyaya, Arabic balagha) are referenced but not deep-treated. Authoring or auditing copy aimed at culturally distinct audiences should pair this skill with culture-specific rhetorical analysis.
- Short-form surfaces (headlines, subject lines, push notifications, button microcopy) require substantiation on the destination page; the gate fires correctly only when the destination page is also audited. The skill provides a dedicated short-form section but cannot enforce destination-page integrity from the unit alone.
- The symmetric detect-and-author structure is a known leak: bad-faith authors learn what to camouflage by reading the detection heuristics. The skill flags this risk but cannot defend against it beyond the ethics gate itself.
