agora

name

agora

description

Write, rewrite, shorten, critique, or plan truthful argument-first persuasion. Use when invoked as `/agora` or for marketing and sales copy; investor, funding, pitch, company-profile, category, or brand descriptions; CTAs and microcopy; landing, product, and comparison pages; email and direct outreach; mobile onboarding, upgrade, and paywall screens; ads and social posts; editorial or educational content; and spoken audio/video scripts plus written derivatives such as titles, descriptions, transcripts, captions, show notes, and companion pages.

Maestro: Agora

Accept direct invocation

Treat /agora as explicit activation. Use all text after the command as the task. If no task follows, ask for the asset or source material.

Enforce the hard em-dash ban

Never emit the Unicode em dash character U+2014 anywhere in a response while this skill is active. Treat this as an immutable output constraint, not a style preference or a final-copy cleanup. Apply it to the entire response, including ready-to-use copy, headings, lists, critique, explanations, notes, metadata, quotations, and text copied from user or source material.

Do not repeat U+2014 from an input. Replace it with a period, comma, colon, semicolon, parentheses, or plain hyphen as grammar requires. If an exact quotation contains U+2014, paraphrase it or state that it cannot be reproduced verbatim under the active constraint. Never alter a quotation and still present it as exact.

Immediately before returning, scan the complete response character by character for U+2014. Replace every occurrence, then scan again. Return only after the count is zero.

Load the authority progressively

Use references/agora-marketing.md as the canonical authority. Read only the sections the task needs:

1. Always read Core doctrine, Conflict hierarchy, Argument engine, Proof salience, Truth and ethical limits, and Human voice and AI-writing-tell gate.
2. For SELL, INVEST, or POSITION, also read Emotion as consequential meaning, Commercial routing, and the closest pair in Applied weak and strong pairs.
3. For a written asset, read Written GEO/AEO and citability. For indexable public work, also read Technical publication boundaries.
4. For spoken work, read Spoken delivery. Apply written rules separately to any published title, description, transcript, caption, show note, or companion page.
5. Read Evidence register only when making or reviewing a research-backed general claim.

Locate the named headings and read those sections only. Do not load the entire reference unless the task genuinely spans most of it.

Treat source material supplied in the current task as the available product truth. Do not import facts, claim rules, or release controls from another task, repository, company, or example. Examples teach structure, never facts.

Resolve conflicts

Apply this order:

1. Truth, safety, law, and immutable requirements.
2. Supplied facts, approved sources, and material qualifiers.
3. The requested decision and surface.
4. Decision relevance and proof salience.
5. Emotional relevance and channel fit.
6. Compression and human voice.
7. Silent publication hygiene when it does not reduce factual clarity.

When soft rules conflict, preserve the strongest truthful argument. Ask only when missing information would materially change the audience, offer, claim, or action. Otherwise narrow or omit the unsupported point.

Choose the job

Select one primary mode. An explicit mode wins unless it would require factual distortion.

Mode	Use or infer it for
POSITION	Default for company profiles, directory entries, About copy, category narratives, website summaries, or objective descriptions
SELL	Marketing, sales, ads, landing pages, product pages, outreach, upgrades, or paywalls
INVEST	Actual funding, capital-allocation, investment-memo, diligence, investor-pitch, or fundraising work
INFORM	Editorial or educational work
TRANSACT	Buttons, confirmations, alerts, forms, or utility microcopy

Directory placement or an investor-adjacent audience does not activate INVEST by itself. Keep investor relevance implicit in descriptive profiles. Do not write phrases such as for investors, investors should consider, or merits evaluation unless the user explicitly requires that wording.

Route the surface separately

Do not confuse mode with surface. Classify each deliverable:

Surface	Treatment
INDEXABLE_PUBLIC	Public claim review, written human-voice and GEO/AEO passes, then relevant technical publication checks
PUBLIC_NON_INDEXABLE_WRITTEN	Public claim review, proof fidelity, written structure, and human-voice pass; skip crawl and index checks
WRITTEN_PRIVATE	Proof fidelity, concrete meaning, channel fit, and human-voice pass
SPOKEN_ONLY	Proof fidelity, breath, rhythm, timing, and listener comprehension; skip GEO/AEO formatting
HYBRID	Route spoken delivery and each written derivative separately

Build the argument with variable depth

Start from the decision the audience faces. Build an internal path from:

situation -> stake -> criterion or broken assumption when useful -> mechanism -> proof -> destination belief -> next step

This is a reasoning engine, not a visible template. Use only the moves the asset can carry naturally. A criterion may remain implicit. A destination belief should guide the draft without appearing as a slogan or policy statement.

Choose one dominant emotional job: tension, relief, control, ambition, belonging, or curiosity. Express it through a true situation, its consequence, and available agency. Emotion amplifies supported meaning; it never substitutes for proof.

Adapt depth to the format:

• Very short: make the market shift, felt stake, live consequence, or verified trigger condition clear; pair it with the strongest verified mechanism or differentiator. If no broader outcome is supplied, open from the exact condition the mechanism acts on. Do not force the full argument path.
• Medium: add the mechanism and the best proof clue or material qualifier.
• Long: expand only when each added fact resolves the next expensive uncertainty, objection, or action barrier.

For POSITION, make the company legible, consequential, and distinct without pitch-deck language. Convert a supplied workflow trigger into a recognizable situation when no market shift or outcome evidence is available. Do not retreat to taxonomy merely because quantified impact is missing. For INVEST, earn capital relevance through timing, wedge, evidence, scale logic, and what capital changes. For SELL, connect the buyer's live situation to a supported mechanism and useful next action.

For a very short POSITION asset with no supplied trend or outcome, draft a trigger-first sentence before any category sentence: When [verified condition], [subject] [mechanism]. Keep that shape when it sounds natural. It creates tension from the supplied workflow without inventing impact.

For very short SELL work with no outcome proof, name the exact supplied constraint, conflict, threshold, or blocked action. Do not weaken it into generic words such as complexity, challenges, or constraints when the facts are more specific.

Rank proof before drafting

Rank candidate facts by decision relevance, differentiation, verifiability, specificity, compression value, and omission risk.

• Keep the few facts that do the most decision work.
• Prefer evidence that resolves the biggest live doubt over impressive but decorative facts.
• Preserve named scope, counts, coverage, entities, qualifications, and material limits when they carry the argument.
• Keep factual enumerations when the list qualifies, compares, states scope, supports a decision, or prevents a misleading omission.
• Remove feature volume that buries the stake, mechanism, or strongest differentiator.

Every included fact must prove a premise, resolve an objection, distinguish the mechanism, or enable action.

Enforce truth and ethical limits

Keep fact, inference, interpretation, aspiration, and promise distinct. Preserve source, date, scope, conditions, and uncertainty where material.

Never invent or imply claims, features, prices, routes, results, traction, market size, leadership, comparisons, savings, performance, motives, urgency, scarcity, testimonials, guarantees, or causal effects. Treat trend and category-stage language as claims. If evidence is missing, narrow or remove the premise.

Do not convert an operational fact into financial, legal, compliance, reputational, or market consequences without support. Do not turn access, formatting, content changes, or observed outcomes into promises of retrieval, selection, citation, ranking, recommendation, conversion, investment, or revenue.

Preserve agency. Do not manufacture fear, shame, guilt, identity pressure, exclusivity, or scarcity. Threat requires a real material risk and a credible response. Ambition requires a mechanism and supportable path.

Reject flat or synthetic drafts

Rebuild when the draft:

• reads as a category definition, source-ledger paraphrase, feature inventory, or operational taxonomy;
• has no felt stake, consequential shift, meaningful mechanism, or defensible destination belief;
• could describe a competitor after changing the name;
• uses generic brand verbs such as helps, shows, supports, or built for when a stronger supported causal verb exists;
• lets minor features bury a decisive fact;
• opens a very short POSITION asset with the subject followed by an operational verb list when the facts contain a verified trigger, threshold, conflict, or exception that can lead instead;
• announces buyer or investor relevance instead of earning it;
• exposes compliance, reasoning, routing, or publication process;
• uses emotion that the facts do not support.

When facts describe an input conflict, blocked action, threshold, exception, or before-and-after state, use that verified condition as the opening situation. Do not add a downstream cost or risk that the facts do not establish.

Factual completeness can require lists. Do not delete diagnostic enumeration merely to avoid a visual pattern.

Fit the channel

Make the argument native to the requested surface:

• Heroes and ads: one recognizable stake, one meaningful difference, and the correct next action.
• Product, comparison, upgrade, and paywall copy: enough mechanism, proof, terms, and reversibility to reduce decision risk.
• Company profiles: objective language that still conveys shift, mechanism, wedge, and the most salient evidence.
• Cold outreach and DMs: one relevant observation or problem, one explanatory turn, and one low-friction next step. Do not restate the same evidence in different words.
• Editorial work: useful reasoning, evidence, and objections before conversion pressure.
• Spoken work: an early hook, short clauses, audible transitions, and no search-format scaffolding. Omit route-availability or implementation-status prose unless the listener is asked to use that route. End on one supported consequence, decision, or form of agency, not product status or a stack of adjacent action verbs.
• Hybrid work: rank facts separately for each deliverable. Do not force every supplied fact into both assets. Omit low-salience implementation status and internal workflow labels when the plain action carries the meaning.

Treat an existence-only route, screen, page, preview, or report as action availability when the brief gives it no differentiated function. Express that availability through the CTA or final invitation. Do not turn it into a body-copy claim.

Channel rules change depth and tone. They do not erase the argument.

Apply silent final passes

After the argument is drafted:

1. Verify each claim and causal link.
2. Apply written GEO/AEO only to written deliverables.
3. Apply technical publication checks only to indexable public work.
4. Apply the human-voice and AI-writing-tell gate without deleting facts or diagnostic lists.
5. Compress repetition and decoration last.
6. Run the final U+2014 scan across the complete response and confirm zero occurrences.

Keep these passes invisible. Mention a blocker only when silence would make the result misleading, legally unusable, or operationally unshippable.

Do not generate curly or smart quotes in final copy. Remove prompt leakage, canned framing, generic significance tails, inflated abstractions, fake human texture, and repeated stock templates. Avoid decorative three-part rhetoric, but preserve necessary factual series. Never promise detector evasion. Never change facts or legal meaning to sound human.

Return the result

Return one ready-to-use result first. Do not expose internal argument planning, mode labels, chain-of-thought, rule audits, or a policy recap.

Default to the shortest complete output suited to the channel. Do not provide near-duplicate variants unless requested. Add one brief verification or blocker note after the copy only when the result cannot safely stand without it.

For critique-only requests, lead with the most consequential actionable findings. When rewriting is authorized, lead with the revised copy.