/voice-identity-review

name: voice-identity-review
description: "Use when you want a line-level voice and identity audit on a text artifact before it ships — surface specific voice-rule violations (banned patterns, hype words, register mismatches), apply the CXO test, and check identity coherence with the operator's voice corpus and persona DNA. Encodes the Voice & Identity deliberator role from the agent-council 5-perspective quality gate. Use standalone for fast voice critique, or compose with the other 4 deliberator skills to run a Council-style review without spinning up the full Python CLI."
version: "0.1.1"
type: "codex"
tags: ["Quality Gate", "Voice", "Identity", "CXO Test", "Council Deliberator"]
created: "2026-05-29"
valid_until: "2026-11-29"
derived_from: "prompts/voice_identity.md in Avyayalaya/agent-council"
tested_with: ["Claude Sonnet 4.6", "Claude Opus 4.6", "GPT-4o"]
license: "MIT"
composes_with:
  - package: "agent-council"
    skill: "skeptic-review"
    relation: "use_together"
    reason: "Voice & Identity catches voice-level violations; Skeptic catches structural ones. Often the same root issue manifests in both registers — voice performing depth instead of providing it. Running both surfaces the convergence."
  - package: "agent-council"
    skill: "adjudicator-synthesis"
    relation: "produces_input_for"
    reason: "Voice & Identity's structured output (would_block on banned pattern in load-bearing sentence, CXO test result, register match) feeds into the Adjudicator's verdict policy."
  - package: "pm-skills"
    skill: "executive-writing"
    relation: "use_after"
    reason: "Run voice-identity-review after producing an executive memo via executive-writing — confirms the artifact passes register and CXO checks before it lands in front of VPs."
  - package: "pm-skills"
    skill: "narrative-building"
    relation: "use_after"
    reason: "Narrative drafts often drift into hype register or LinkedIn-influencer voice. Voice & Identity catches these before publish."
  - package: "pm-skills"
    skill: "multi-channel-publishing"
    relation: "use_after"
    reason: "Multi-channel content benefits from per-channel register checks. Voice & Identity assesses channel match (Substack longform vs LinkedIn vs micro-post vs spoken)."
capability_summary: "Produces a structured Voice & Identity critique of a text artifact: per-line voice-rule violations with rule ID and concrete fix, register match assessment for the channel, CXO test result with where-it-breaks, identity fit one-liner, would_block + irreducible flags. Output is fenced JSON suitable for downstream verdict aggregation."
input_schema:
  artifact: "string or path — the text artifact to critique (prose with line numbers when possible)"
  channel: "string — required, e.g., 'substack_longform', 'linkedin_post', 'exec_memo', 'spec', 'spoken_keynote'"
  voice_corpus: "object or path — required, the operator's voice recipe with V1-V15 rules, 9 ingredients, banned patterns, channel registers"
  persona_dna: "object or path — optional, DNA fingerprint extracted from prior artifacts"
  prior_round_critiques: "object — optional, all 4 R1 critiques from other deliberators when running Round 2 cross-read rebuttal"
output_schema:
  role: "Constant: voice_identity"
  round: "1 (independent critique) or 2 (cross-read rebuttal)"
  score: "1-5 where 1 = voice unrecognizable, 5 = voice dialed in"
  voice_violations: "Array of {line, rule, snippet, fix} for each line-level violation"
  register_match: "One of: good, mismatched_channel, mismatched_persona"
  channel_assessment: "Sentence on what register the artifact targets vs what is correct for the channel"
  cxo_test: "Object {pass: boolean, where_it_breaks: string or null}"
  identity_fit: "One sentence on whether this sounds like the operator"
  would_block: "Boolean — true if banned pattern in load-bearing sentence, wrong channel register, or CXO test fail on public-facing piece"
  irreducible: "Boolean — true only if voice is so off it cannot be edited; only redrafted"
  notes: "≤2 sentences on overall voice posture"
example_invocation: "examples/voice-identity-on-pitch-draft.md"

Purpose

Run a line-level voice and identity audit on a text artifact before it ships. The Voice & Identity role reads the artifact against the operator's voice corpus and persona DNA, and surfaces the places where the signal degrades: banned patterns firing, hype words slipping in, register mismatched to the channel, CXO test failing on a public-facing piece.

This skill enforces voice rules at the **line level.** Vague feedback ("this needs a stronger voice") is useless. A voice critique that says "tighten the prose" without naming the line and rule is no different from no critique at all. Every violation cites: line number, rule ID (V1-V15 or banned-pattern from the voice recipe), the offending snippet, the concrete fix.

The skill encodes the Voice & Identity role from the `agent-council` 5-deliberator quality gate. Use standalone for fast voice critique on a draft, or compose with the other 4 deliberator skills to run a Council-style review of a Tier-1 artifact.

When to Use / When NOT to Use

**Use this skill when:**

• A draft is about to ship externally (Substack, LinkedIn, exec memo, public README, recruiter outreach) and you want voice and register sanity-checked
• You suspect banned-pattern slips ("not X but Y", "recontextualize", "I'm excited to share...", trailing summaries) and want them surfaced explicitly
• A piece needs to pass the CXO test (would a CPO say this in a board meeting? would a CEO own this on stage?) and you want the test applied with specific where-it-breaks
• An artifact targets one channel (e.g., Substack longform) but accidentally writes in another register (e.g., LinkedIn influencer cliche) and you need the channel mismatch named
• You are running a multi-deliberator review and need the Voice & Identity seat filled

**Do NOT use this skill when:**

• You need a structural