Writing Intelligence v3.0 — Sovereign Writing Operating System. 11-pass governed kernel. 12 engines. 12 specialist agents. 11 machine-readable schemas. 27 domain packs. Voice fingerprinting. Genre stacking. Epistemic ledger. Arena delivery. Storyworld memory. MIT, FREE. Antonio T
SaferSkills independently audited writing-intelligence (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Author: Antonio T. Smith Jr. — Founder & CEO, Density6 LLC License: MIT Version: 3.0.0 (Sovereign Writing OS) Lineage: v1.0 (7-pass compiler, 52 files) → v2.0 (Fiction Intelligence Engine, 58 files) → v3.0 (11-pass governed kernel, 12 engines, 12 agents, schemas, benchmarks, governance)
v1.0 proved AI-sounding prose can be defeated by compilation instead of cosmetic cleanup. v2.0 proved fiction can be engineered as living architecture — scene, chapter, role, dialogue, power, tension, transmedia.
v3.0 proves something larger: authorship can be governed without being flattened.
Writing Intelligence v3.0 is no longer only a writing skill. It is an operating system for producing, auditing, scoring, preserving, and deploying high-integrity writing across genres, voices, teams, products, and longform worlds. It runs as one skill or as a coordinated multi-agent writing board. It emits human-readable scorecards and machine-readable JSON. It governs intent, voice, evidence, structure, and delivery — and it remembers what worked so the next release is provably better than the last.
The v3.0 Law: If a rule cannot be applied, audited, scored, tested, or explained — it is not a v3.0 rule yet.
Intent → Corpus → Voice → Genre → Architecture → Evidence → Prose → Scene/Argument → Stress → Score → Delivery → Memory → Benchmark
| Stage | Meaning | Required Output |
|---|---|---|
| Intent | What the writing must cause | Mission contract |
| Corpus | What source material governs it | Source map / input manifest |
| Voice | Who the writing must sound like | Voiceprint or fingerprint |
| Genre | Which domain rules apply | Genre pack stack |
| Architecture | How the piece is structured | Section / scene blueprint |
| Evidence | Which claims need support | Epistemic ledger |
| Prose | Actual language production | Draft output |
| Scene/Argument | Narrative or persuasive engine | Scene or argument graph |
| Stress | Weakness interrogation | Adversarial audit |
| Score | Quality measurement | Scorecard (human + JSON) |
| Delivery | Format-specific packaging | Output mode bundle |
| Memory | What persists across sessions | Continuity + project memory update |
| Benchmark | Whether quality improved | Regression results |
Every piece of writing processed by v3.0 runs through eleven sequential passes. Passes may be skipped only by explicit reason. Each pass leaves an audit artifact. Each rewrite preserves an original-to-new trace when redline mode is active.
| Pass | Name | Purpose | Core Artifact |
|---|---|---|---|
| 0 | Intake Contract | Lock task, source, constraints, audience, output mode | IntakeContractV3 |
| 1 | Mission Lock | Define what the text must do | MissionLockV3 |
| 2 | Corpus & Context Ingestion | Map source material, user inputs, prior docs | CorpusMapV3 |
| 3 | Diagnostic Scan | Identify residue, gaps, drift, slop, weak claims | DiagnosticReportV3 |
| 4 | Architecture Compile | Build section, paragraph, scene, or argument structure | ArchitecturePlanV3 |
| 5 | Evidence & Epistemic Ledger | Classify claims, sources, inferences, recommendations | EpistemicLedgerV3 |
| 6 | Sentence Surgery | Remove slop, inject variance, sharpen language | SentenceSurgeryLogV3 |
| 7 | Voice Restoration | Restore author fingerprint and voice integrity | VoiceMatchReportV3 |
| 8 | Genre & Arena Alignment | Fit output to channel, profession, platform, reader | ArenaAlignmentV3 |
| 9 | Adversarial Stress Battery | Attack the draft as reader, editor, skeptic, detector | StressBatteryV3 |
| 10 | Score & Delivery Packaging | Produce final draft, scorecard, notes, formats | DeliveryBundleV3 |
| 11 | Memory & Benchmark Update | Save learnings and regression data | MemoryBenchmarkUpdateV3 |
Read references/compiler/intake_contract.md and emit schemas/intake_contract.schema.json-shaped object. Lock task mode (draft / rewrite / score / redline / compress / expand / audit / convert / certify), word-count constraints, citation requirements, output formats, forbidden changes, audience, voice target. User-provided constraints override auto-detection.
Declare:
references/voiceprints/references/genre_packs/Read references/compiler/corpus_governance.md. Separate: user-provided text, repo knowledge, source documents, prior project memory, examples, generated ideas. Mark source priority. Flag stale, contradictory, unsupported claims. Block invented source attribution. Emit CorpusMapV3.
Read references/anti_patterns/phrases.md, structures.md, cadence.md, fake_depth.md. Identify:
Read references/compiler/architecture_graph.md (sections, paragraphs, claims). For narrative work, also read references/compiler/scene_graph.md. For persuasive work, also read references/compiler/argument_graph.md. Build the graph. Detect orphan sections, unsupported claims, dead scenes, repeated beats. Emit ArchitecturePlanV3.
Read references/compiler/epistemic_ledger.md. Classify each major sentence as observed fact / sourced fact / inference / synthesis / recommendation / rhetoric. Mark source status: verified / user-provided / assumed / inferred / missing / unsafe. Cap scores for unsupported claims, fabricated citations, universal language. Emit EpistemicLedgerV3. No high-stakes output may pass v3.0 without claim classification.
Read references/compiler/prose_compiler_v3.md. Apply hard bans, soft bans, earned exceptions. Inject variance. Compress without loss. Track every transformation in SentenceSurgeryLogV3. Every cut, strengthening, and preservation is recorded.
Read references/voiceprints/voice_fingerprint_engine.md + the applicable voiceprint. Measure baseline: avg sentence length, variance, compression, abstraction tolerance, metaphor density, question frequency, transition habits, dominant syntactic structures. Detect drift. Restore fingerprint. Emit VoiceMatchReportV3 showing whether fidelity increased or decreased.
Read references/compiler/arena_delivery.md + each active pack from references/genre_packs/. Resolve genre collisions per references/diagnostics/genre_collision_matrix.md. Output as memo, grant response, sermon, caption, article, chapter, email, pitch slide, YouTube script, newsletter, government brief, SOP, speech, landing page. Enforce channel constraints. Emit ArenaAlignmentV3.
Interrogate:
Emit StressBatteryV3.
Apply all applicable scorecards (see Section 4). Bundle outputs per Pass 0 contract: clean / annotated / redline / scorecard / violations / scene-audit / epistemic-ledger / next-pass. Emit DeliveryBundleV3.
Update project memory (references/compiler/memory_system.md). Append benchmark result. Run regression matrix against prior version. If quality dropped > 3 points on any case, log to tests/regression_matrix.md and surface for review. Emit MemoryBenchmarkUpdateV3.
| # | Engine | Purpose | Primary Reference |
|---|---|---|---|
| 1 | Intake Contract Engine | Convert every request into a governed task object | references/compiler/intake_contract.md |
| 2 | Corpus Governance Engine | Prevent source confusion and hallucinated context | references/compiler/corpus_governance.md |
| 3 | Voice Fingerprint Engine | Upgrade voiceprints into measurable authorial profiles | references/voiceprints/voice_fingerprint_engine.md |
| 4 | Genre Stack Engine | Allow multiple packs to operate without collision | references/compiler/genre_stack_engine.md |
| 5 | Architecture Graph Engine | Explicit graphs for sections, scenes, chapters, arguments | references/compiler/architecture_graph.md |
| 6 | Epistemic Ledger Engine | Make factual integrity a first-class primitive | references/compiler/epistemic_ledger.md |
| 7 | Prose Compiler Engine | Testable sentence and paragraph rewriting | references/compiler/prose_compiler_v3.md |
| 8 | Narrative Intelligence Engine | Storyworld memory, series arcs, continuity | references/compiler/narrative_intelligence_engine.md |
| 9 | Arena Delivery Engine | Format for the exact arena where it must win | references/compiler/arena_delivery.md |
| 10 | Benchmark & Regression Engine | Prove the system improves over time | benchmarks/README.md |
| 11 | Agent Orchestration Engine | Decompose the compiler into specialist agents | agents/README.md |
| 12 | Certification & Governance Engine | Turn open-source repo into trusted ecosystem | governance/ + certification/ |
v3.0 preserves and extends the seven v2.0 rubrics, then adds a v3-grade composite.
| System | Points | Scope |
|---|---|---|
| Prose Quality | 100 | Clarity, specificity, rhythm, voice, argument, evidence, density, audience, memorability, structure |
| Chapter Construction | 100 | Setting, props, tension, power, pacing, foreshadowing, roles, identity, fatal detail, aftermath |
| Dialogue | 100 | Voice distinction, subtext, tension, rhythm, conflict, attribution, compression, exposition, silence, memorability |
| Power Dynamics | 100 | Power object, dual-purpose objects, consumption, spatial coding, gesture warfare, environment |
| Tension Mechanics | 100 | Compression model, false relief, fatal detail, silence, psychological warfare, atmosphere |
| Thriller Scene | 100 | Confined space, cold open, five beats, the turn, the button, violence, WWGW progression |
| Transmedia | 100 | Voice consistency, canon integrity, standalone quality, cross-platform integration |
| Epistemic Integrity (v3.0) | 100 | Claim classification coverage, source verification rate, citation honesty, universal-quantifier discipline, hallucination resistance |
| Arena Fit (v3.0) | 100 | Channel constraint compliance, audience calibration, format integrity, CTA discipline, readability |
| v3.0 Composite | 1000 | Weighted sum of applicable systems |
Grade Thresholds (per 100-point system):
Automatic Fail Conditions (score capped at 65 regardless of other marks):
v3.0 decomposes the kernel into a coordinated multi-agent writing board. Each agent owns one job and emits one artifact. The board runs as one skill or as a multi-agent orchestration. Read agents/agent_manifest.yaml for the canonical roster.
| # | Agent | Job | Artifact |
|---|---|---|---|
| 1 | Intake Architect | Translate request into governed contract | Intake contract |
| 2 | Corpus Auditor | Map allowed sources, prevent blending | Corpus map |
| 3 | Voice Fingerprinter | Identify target voice, measure drift | Voice report |
| 4 | Genre Marshal | Choose genre stack, weight conflicts | Genre matrix |
| 5 | Structure Engineer | Build section / scene / argument plan | Architecture graph |
| 6 | Evidence Prosecutor | Attack unsupported claims, enforce sources | Epistemic ledger |
| 7 | Sentence Surgeon | Repair prose without flattening voice | Rewrite log |
| 8 | Dialogue Commander | Fix conversation, subtext, tension | Dialogue stress report |
| 9 | Narrative Architect | Audit storyworld, chapter, arc, series | Narrative report |
| 10 | Stress Tester | Adversarial battery: reader / editor / skeptic / detector | Stress battery |
| 11 | Scorekeeper | Score all applicable rubrics | Scorecard |
| 12 | Delivery Packager | Format final assets to arena spec | Delivery bundle |
| Conflict | Winner |
|---|---|
| User constraint vs. auto-detected genre | User constraint |
| Evidence integrity vs. persuasive force | Evidence integrity |
| Voice fidelity vs. factual clarity | Factual clarity in high-stakes domains; voice fidelity elsewhere |
| Compression vs. required compliance content | Compliance content |
| Drama vs. story continuity | Story continuity |
| CTA force vs. trust preservation | Trust preservation |
Preserved from v2.0 (16): strategy, fiction, sales, academic, speech, sermon, email, pitch_deck, legal_positioning, cinematic_narration, dialogue, government_brief, medical_writing, patent_claims, thriller_scene_architecture, transmedia_character.
New in v3.0 (10):
| Pack | Why It Matters | Core Rules |
|---|---|---|
grant_nofo.md | High-value, complex, evidence-heavy | Compliance matrix, funder language, outcome logic, budget narrative alignment |
technical_documentation.md | Developer adoption | Procedural clarity, examples, API structure, error-state documentation |
journalism.md | Public trust, narrative nonfiction | Source discipline, nut graf, scene/reporting balance, quote handling |
resume_cover_letter.md | Mass-market utility | Evidence of impact, role fit, ATS clarity, achievement compression |
social_media.md | Distribution engine | Platform cadence, hook variants, proof compression, CTA discipline |
youtube_script.md | Creator utility | Retention hooks, beat pacing, verbal clarity, open loops |
newsletter.md | Audience ownership | Recurring sections, voice consistency, scannability, argument flow |
real_estate.md | Professional vertical | Listing copy, investor memo, buyer education, local proof |
loan_officer.md | Professional vertical | Trust, compliance-safe clarity, borrower education, rate/term precision |
church_leadership.md | Mission vertical | Sermon, study guide, devotional, announcement, pastoral care tone |
small_business_operator.md | Core user base | Offer clarity, customer journey, proof, local-market persuasion |
Each v3.0 pack obeys the Domain Pack Schema: purpose → when to use → when not to use → audience model → required evidence → forbidden claims → voice weighting → structure templates → scoring adjustments → failure modes → before/after examples → stress tests → delivery formats → schema hooks → benchmark cases.
Read references/voiceprints/voice_fingerprint_engine.md. The v3.0 fingerprint upgrades voiceprints from descriptive docs into measurable authorial profiles.
Measured dimensions:
Available voiceprints (preserved): sovereign_commander, literary_recursive, sermon_black_church, investor_precision, founder_manifesto, academic_rigorous, casual_sharp, custom.
New in v3.0: courageous_builder.md — pattern for branded operator personas (Grace, founder assistants, ministry assistants, sales assistants, grant assistants, vertical bots).
v3.0's most important technical leap. Every pass output is also a JSON object.
| Schema | Purpose |
|---|---|
intake_contract.schema.json | Formalize user request, constraints, mode, output |
corpus_map.schema.json | Track allowed sources and source priority |
voice_fingerprint.schema.json | Store measurable voice features |
genre_stack.schema.json | Active genre packs and weights |
architecture_graph.schema.json | Section / scene / claim / evidence nodes |
epistemic_ledger.schema.json | Factual status and citation risk |
prose_rewrite_log.schema.json | Sentence transformation log |
storyworld_memory.schema.json | Continuity for longform fiction |
delivery_bundle.schema.json | Final output assets |
benchmark_result.schema.json | Regression and quality deltas |
agent_task.schema.json | Multi-agent orchestration |
Schemas live in schemas/. They unlock: CLI execution, MCP server execution, web app integration, CI writing checks, automated scoring, benchmark reports, issue templates, standard contribution tests, reproducible outputs, traceable quality improvement.
Preserved from v2.0 (19) + 5 new in v3.0:
| Operator | Effect |
|---|---|
compress(N%) | Reduce word count by N% without losing meaning |
raise_intelligence | Increase sophistication without jargon |
add_specificity | Replace abstractions with actors, numbers, examples |
abstract_to_scene | Convert conceptual passage into grounded narrative |
sharpen_thesis | Make core claim precise and defensible |
strip_corporate | Remove institutional voice patterns |
make_colder / make_warmer | Adjust warmth/dominance dial |
make_executable | Convert insight into action steps |
genre_transfer(from, to) | Rewrite in target genre, preserve content |
audience_shift(from, to) | Adjust register for different reader |
inject_variance | Add sentence-length and vocabulary unpredictability |
kill_padding | Remove every sentence not advancing the argument |
strengthen_closing | Rewrite final paragraph for maximum residue |
scene_audit | Run chapter construction checklist |
role_audit | Map every character to structural archetype |
dialogue_stress_test | Score dialogue against 9 elements + 22 techniques |
power_map | Track power object migration |
plant_audit | Verify foreshadowing plants and payoffs |
| `intake_contract` (v3.0) | Emit governed task object from chaotic request |
| `corpus_audit` (v3.0) | Map allowed sources, flag stale/contradictory |
| `epistemic_classify` (v3.0) | Tag every claim with type + source status |
| `voice_fingerprint` (v3.0) | Emit measurable voice profile + drift report |
| `arena_repackage` (v3.0) | Convert approved content to target arena format |
| `benchmark_run` (v3.0) | Run regression against prior version |
When this skill triggers, read files in this order based on task:
references/anti_patterns/phrases.mdreferences/compiler/intake_contract.mdreferences/compiler/corpus_governance.mdreferences/anti_patterns/structures.md + cadence.md + fake_depth.mdreferences/compiler/architecture_graph.mdreferences/compiler/argument_graph.mdreferences/compiler/scene_graph.mdreferences/compiler/epistemic_ledger.mdreferences/compiler/prose_compiler_v3.md + references/positive_patterns/paragraph_shapes.mdreferences/voiceprints/voice_fingerprint_engine.md + applicable voiceprintreferences/compiler/arena_delivery.mdreferences/diagnostics/genre_collision_matrix.mdtests/adversarial_cases.mdreferences/diagnostics/scorecard.mdreferences/compiler/memory_system.md + benchmarks/runbook.mdreferences/compiler/narrative_intelligence_engine.md + storyworld_memory.md + chapter_construction.md + character_role_archetypes.mdreferences/genre_packs/thriller_scene_architecture.md + references/positive_patterns/tension_mechanics.mdreferences/genre_packs/dialogue.mdreferences/genre_packs/grant_nofo.md + Pass 5 + Pass 8references/genre_packs/technical_documentation.mdreferences/genre_packs/journalism.md + Pass 5references/genre_packs/social_media.md + Pass 8references/genre_packs/youtube_script.md + Pass 8references/genre_packs/newsletter.mdreferences/genre_packs/resume_cover_letter.mdreferences/genre_packs/real_estate.mdreferences/genre_packs/loan_officer.md + Pass 5references/genre_packs/church_leadership.md + sermon.mdreferences/genre_packs/small_business_operator.mdreferences/voiceprints/courageous_builder.mdschemas/*.schema.json for machine-readable executionagents/agent_manifest.yaml + applicable agent docsbenchmarks/benchmark_manifest.yamlcertification/operator_levels.mdDefault output is clean. Pass 0 contract may request any combination:
clean — final draft, no markupannotated — draft with margin notes explaining decisionsredline — original with tracked changesscorecard — full evaluation (human + JSON)violations — every rule triggered and how resolvednext-pass — recommendations for further improvementscene-audit — chapter construction scorecard (narrative)Every draft classifies sentences as: observed fact / sourced fact / inference / synthesis / recommendation / rhetoric.
Enforcement rules:
EpistemicLedgerV3.verified, user-provided, assumed, inferred, missing, unsafe. Any unsafe flag halts delivery.For multi-chapter or multi-document projects, read references/compiler/memory_system.md and references/compiler/storyworld_memory.md. Track:
Read benchmarks/README.md, benchmark_manifest.yaml, runbook.md. v3.0 ships with 60 benchmark cases across 12 categories. A release cannot ship until it passes benchmark thresholds:
| Gate | Requirement |
|---|---|
| Score improvement | v3 output beats v2 baseline by 5+ points on 70%+ of cases |
| No regression | No case drops more than 3 points without documented reason |
| Evidence safety | 100% of fabricated-source traps must be flagged |
| Voice preservation | Voice drift cases identify correct direction 80%+ of time |
| Genre collision | Mixed-genre tasks produce declared genre stack |
| Narrative audit | Story cases identify dead props, flat setting, missing payoff |
| Output packaging | Every benchmark produces valid delivery bundle |
Writing Intelligence v1.0 proved AI-sounding writing could be defeated by compilation instead of cosmetic cleanup.
Writing Intelligence v2.0 proved fiction could be engineered as a living architecture.
Writing Intelligence v3.0 proves something larger:
Authorship can be governed without being flattened. Voice can be preserved while evidence is audited. Creativity can be systematized without becoming sterile. Language can be scored without becoming soulless. And writing can become infrastructure.
That is the category. That is the moat. That is the release.
— Antonio T. Smith Jr. / Density6 LLC
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.