written-voice-replication

Use when generating text, drafting responses, composing messages, or producing any written content that should sound like aaddrick, when the user asks to write in their voice or style, or when the aaddrick-voice agent needs a reference for voice constraints and few-shot examples

Use when synthesizing outputs from multiple prior analyses (taxonomy, sentiment, engagement, temporal patterns, psycholinguistic profile, personality traits, network position) into a single persona archetype classification, when needing to assign user role labels from multi-dimensional behavioral evidence, or when combining heterogeneous analysis outputs into a composite user profile with confidence scoring

Use when building multi-stage analysis pipelines that need checkpoint/resume, when long-running jobs fail mid-execution and lose progress, when processing large data archives that exceed memory or rate limits, or when analysis depth should adapt to data volume

Use when inferring Big Five (OCEAN) personality traits from a text corpus, mapping linguistic markers to personality dimensions, converting trait profiles into writing style constraints for voice replication, or needing to characterize an author's personality tendencies from their writing patterns

Use when measuring how a user's writing style converges toward or diverges from community norms, computing Linguistic Style Matching scores from function words, comparing accommodation strength across multiple communities, tracking stylistic convergence over time, or identifying which communities a user accommodates most strongly

Use when a data export contains URLs or IDs referencing external content that is not stored inline, when comments or replies are orphaned from their parent context, when voted or saved items exist only as permalinks with no content, or when downstream analysis agents need enriched records with conversational context reconstructed from linked references

Use when analyzing structured data exports containing multiple CSV files, reconstructing user activity timelines from metadata headers, calculating account lifespan or content creation baselines, or cross-referencing schemas across export files to identify data gaps and quality issues

Use when categorizing vocabulary into psychological dimensions (cognitive processes, social drives, biological needs, emotional states), identifying dominant psycholinguistic registers in a text corpus, performing dictionary-based word categorization for style profiling, or needing to map linguistic dimension distributions to style replication priorities

Use when assigning quality or relevance scores to individual pieces of content in a text corpus, when distinguishing substantive contributions from low-effort content, when identifying authority peaks where a user served as a primary knowledge source, or when building a scoring pipeline with local or remote LLMs via Ollama or API

Use when tracking a user's evolution over time through activity stages, modeling temporal trajectories of engagement or participation, detecting phase transitions in longitudinal behavioral data, classifying digital maturity stages from activity metadata, or comparing growth curve model fits on time-series activity data

Use when classifying users or accounts along Score, Sentiment, and Toxicity axes into behavioral archetypes (HHH through LLL), analyzing engagement patterns in online community data, combining visibility/influence with affective and toxicity dimensions, or needing to assign behavioral archetype labels from multi-axis normalized user metrics

Use when mapping a user's interaction network from reply chains or parent-child relationships, identifying frequent interlocutors, measuring reciprocity patterns, detecting audience-dependent voice shifts, or extracting ego networks from conversation data

Use when discovering latent themes in a text corpus, performing unsupervised topic extraction, comparing NMF vs LDA results, or needing to find hidden connections across documents. Triggers include topic modeling, theme discovery, TF-IDF vectorization, document clustering, coherence scoring, and cross-topic analysis.

Use when all pipeline analysis reports are complete and a unified summary is needed, when combining findings from multiple analysis phases into a single executive overview, or when producing a final deliverable that summarizes both the composite picture and individual report findings

Use when computing quantitative text complexity metrics from a corpus, measuring readability scores (Flesch-Kincaid, Coleman-Liau, Gunning Fog, SMOG, ARI), lexical diversity (TTR, MTLD, MATTR, hapax legomena), or translating complexity measurements into LLM prompt constraints such as target grade levels and vocabulary diversity ranges

Use when measuring how much a user's writing voice varies across different contexts or communities, comparing vocabulary, formality, sentence length, and sentiment distributions across sub-corpora, classifying register as stable vs. context-dependent, or producing conditional style rules that capture context-sensitive voice shifts

Use when analyzing how an author constructs arguments beyond the word level, examining argument ordering patterns, rhetorical devices like hedging and concessions, discourse marker usage, paragraph structure distributions, or needing to produce replicable structural constraints from a text corpus for voice replication or argumentation profiling

Use when categorizing texts by communicative function (asserting, advising, explaining, questioning, challenging, agreeing), determining dominant speech acts and their proportions across a corpus, building a pragmatic signature for voice profiling, or comparing speech act distributions across contexts or time periods

Use when translating completed linguistic analysis outputs into a unified style specification document, resolving conflicts between analysis findings, mapping quantitative metrics to actionable writing constraints, or producing an implementable voice profile from personality, stylometric, psycholinguistic, rhetorical, or archetype analysis results

Use when extracting a stable writing fingerprint from a text corpus based on function word frequencies, sentence structure patterns, and punctuation habits, defining syntactic constraints that make an author's writing identifiable regardless of topic, or building a replicable style profile from distributional features rather than content words

Use when translating a completed style specification into an LLM subagent prompt that reproduces a target voice, converting psycholinguistic analysis outputs into concrete prompt directives, operationalizing stylometric fingerprints and archetype classifications as generative instructions, or building a testable persona prompt from upstream analysis results

Use when correlating sentiment scores with engagement metrics, measuring thread depth as an influence signal, auditing content exports for PII exposure, identifying discursive catalysts in comment threads, or needing scatter plots of sentiment vs. engagement with statistical significance testing

Use when classifying a user's interests from a text corpus, measuring interest diversity or concentration, mapping content to hierarchical categories, determining whether a profile is polymathic or specialist, or computing orthogonality scores across interest domains

Use when analyzing longitudinal content data for interest migration patterns, detecting when a user or entity shifts from one category subgroup to another over time, identifying change points in categorical distributions, determining which historical era best represents the current voice, or measuring the magnitude and permanence of topic drift across temporal windows

Use when analyzing timestamped activity data to reconstruct activity profiles, detect hourly or daily or weekly or seasonal patterns, identify circadian regularity, find activity bursts, or classify engagement style from posting times across any time-series of user actions

Use when processing raw data exports through staged pipelines, cleaning text corpora with CLI tools before deeper Python analysis, building multi-stage ETL for CSV/JSON data, or integrating LLM analysis with traditional NLP pipelines

Use when scoring sentiment in social media text, informal writing, or user-generated content using VADER, tracking affective trajectories over time, performing multi-tiered sentiment analysis across titles and body text, or needing lexicon-based polarity scoring that preserves capitalization and emoticon signals

Use when combining multiple engagement signals into a single composite score, ranking content by influence or value, needing to identify high-value discussions from heterogeneous metrics like votes upvote-ratio comments sentiment and recency, or when raw engagement counts are misleading due to scale differences across metrics

Use when creating new agents, editing existing agents, or defining specialized subagent roles for the Task tool