Corpus Persona-Extraction
Distill how a specific voice thinks and speaks from a pile of session logs, and persist it as a machine-usable lexicon that downstream agents can write against.
Why this exists
Agents that work with a long-running human collaborator keep re-learning the same voice from scratch: the coinages, the load-bearing metaphors, the sentence shapes that mean "yes, proceed" versus "you've missed the point." A lexicon file makes that knowledge durable and composable — voice-enforcement gates can score against it, drafting skills can imitate it, and translation layers (persona storefronts) can project from it.
Core doctrine
- User coinages outrank standard vocabulary. When the corpus shows an idiosyncratic term for a concept ("vacuum", "organ", "cartridge", "liturgy"), the lexicon records the coinage as primary and the standard term as gloss — never the reverse.
- Append-only. Each extraction run adds a dated block. Prior blocks are evidence of drift, not errors to clean up.
- Separate speakers ruthlessly. A transcript interleaves human, assistant, and tool text. Only the persona's own turns feed frequency analysis; assistant paraphrase contaminates the signal.
Workflow
1. Assemble and filter the corpus
- Gather session files; record provenance (paths, date range, message counts) in the extraction header.
- Parse each format to (speaker, timestamp, text) tuples; keep only
speaker == persona_id turns.
- Strip quoted/pasted material (code blocks, tool output, copied text) — the persona's own prose only.
- Redact before analysis: run the corpus through secret/PII patterns first; a lexicon must be shareable.
2. Vocabulary frequency
- Tokenize; compute frequency ranks for unigrams/bigrams/trigrams.
- Diff against a general-English baseline: the interesting set is terms whose corpus rank vastly exceeds baseline rank (keyness), not raw frequency.
- Bucket results: domain terms (shared with field), coinages (absent from baseline entirely), and inflections (standard words used in non-standard senses).
3. Idioms and structural analogies
- Idioms: recurring multiword expressions with stable meaning ("fix bases, not outputs", "the soul persists").
- Structural analogies: the source domains the persona habitually maps from — music, anatomy, liturgy, architecture. Record each as
analogy: {source_domain} → {target_domain} with 2-3 verbatim examples.
- Register markers: profanity patterns, emphasis habits (caps, em-dashes), sentence-length distribution, imperative density.
4. Synthesize ideal_yearning and archetypal_pattern
These two fields are the lexicon's soul — short synthesized statements, each grounded in ≥3 verbatim quotes:
ideal_yearning — what the persona is reaching toward across all sessions; the recurring wish underneath the tasks. Written as one sentence in the persona's own idiom.archetypal_pattern — the recurring role/shape the persona enacts (e.g. "the conductor who builds the orchestra rather than playing the instrument"). One sentence plus the evidence quotes.
These are hypotheses, marked with extraction date and confidence — later blocks may revise them.
5. Append to the lexicon
# {persona_id}.lexicon.yaml — append block, never rewrite
- extraction:
date: 2026-06-07
corpus: {files: 128, persona_turns: 1432, date_range: [2026-04-17, 2026-06-07]}
redaction: {patterns_run: true, hits_scrubbed: 0}
vocabulary:
coinages:
- term: vacuum
gloss: "an absence that is unrepresented anywhere durable; never a resting state"
examples: ["N/A is a vacuum", "vacuum field burn"]
domain_terms: [...]
inflections: [...]
idioms:
- phrase: "fix bases, not outputs"
meaning: "modify the source/template/pipeline, never the rendered artifact"
analogies:
- source: liturgy/theology
target: configuration management
examples: ["Tier-0 Liturgy", "the Reliquary", "testament"]
register:
profanity: "affection-marker, not hostility"
emphasis: ["em-dash chains", "bold imperatives"]
ideal_yearning: "…" # one sentence, persona's idiom, ≥3 quote citations
archetypal_pattern: "…"
6. Report
Emit a short extraction report: new coinages since last block, terms that dropped out (candidate drift), and any revision to yearning/pattern with the evidence that forced it. Commit lexicon + report together.
Anti-patterns
- Extracting from assistant turns. The most common contamination; the lexicon ends up describing the model, not the persona.
- Overwriting prior blocks. Drift between blocks is the most valuable longitudinal signal; preserve it.
- Unredacted lexicons. Verbatim example quotes can carry credentials/PII from the source sessions; redact at extraction time, not later.
- Synthesizing yearning from one loud session. Require cross-session recurrence (≥3 sessions) before a theme reaches the yearning/pattern fields.
1---2name: corpus-persona-extraction3description: Ingest a corpus of session logs (JSONL transcripts, chat exports) and extract a persona profile — vocabulary frequency, unique idioms and coinages, structural analogies — then synthesize an ideal_yearning and archetypal_pattern and append the result to a {persona_id}.lexicon.yaml. Triggers on "extract persona", "build a lexicon", "map the domain language", or any request to process conversation logs into a voice/vocabulary model.4license: MIT5---6
7# Corpus Persona-Extraction
8
9Distill *how a specific voice thinks and speaks* from a pile of session logs, and persist it as a machine-usable lexicon that downstream agents can write against.
10
11## Why this exists
12
13Agents that work with a long-running human collaborator keep re-learning the same voice from scratch: the coinages, the load-bearing metaphors, the sentence shapes that mean "yes, proceed" versus "you've missed the point." A lexicon file makes that knowledge durable and composable — voice-enforcement gates can score against it, drafting skills can imitate it, and translation layers (persona storefronts) can project from it.
14
15## Core doctrine
16
17- **User coinages outrank standard vocabulary.** When the corpus shows an idiosyncratic term for a concept ("vacuum", "organ", "cartridge", "liturgy"), the lexicon records the coinage as primary and the standard term as gloss — never the reverse.
18- **Append-only.** Each extraction run adds a dated block. Prior blocks are evidence of drift, not errors to clean up.
19- **Separate speakers ruthlessly.** A transcript interleaves human, assistant, and tool text. Only the persona's own turns feed frequency analysis; assistant paraphrase contaminates the signal.
20
21## Workflow
22
23### 1. Assemble and filter the corpus
24- Gather session files; record provenance (paths, date range, message counts) in the extraction header.
25- Parse each format to (speaker, timestamp, text) tuples; keep only `speaker == persona_id` turns.
26- Strip quoted/pasted material (code blocks, tool output, copied text) — the persona's *own* prose only.
27- **Redact before analysis**: run the corpus through secret/PII patterns first; a lexicon must be shareable.
28
29### 2. Vocabulary frequency
30- Tokenize; compute frequency ranks for unigrams/bigrams/trigrams.
31- Diff against a general-English baseline: the interesting set is terms whose corpus rank vastly exceeds baseline rank (keyness), not raw frequency.
32- Bucket results: domain terms (shared with field), **coinages** (absent from baseline entirely), and inflections (standard words used in non-standard senses).
33
34### 3. Idioms and structural analogies
35- Idioms: recurring multiword expressions with stable meaning ("fix bases, not outputs", "the soul persists").
36- Structural analogies: the source domains the persona habitually maps from — music, anatomy, liturgy, architecture. Record each as `analogy: {source_domain} → {target_domain}` with 2-3 verbatim examples.
37- Register markers: profanity patterns, emphasis habits (caps, em-dashes), sentence-length distribution, imperative density.
38
39### 4. Synthesize ideal_yearning and archetypal_pattern
40These two fields are the lexicon's soul — short synthesized statements, each grounded in ≥3 verbatim quotes:
41- **`ideal_yearning`** — what the persona is reaching toward across all sessions; the recurring wish underneath the tasks. Written as one sentence in the persona's own idiom.
42- **`archetypal_pattern`** — the recurring role/shape the persona enacts (e.g. "the conductor who builds the orchestra rather than playing the instrument"). One sentence plus the evidence quotes.
43
44These are *hypotheses*, marked with extraction date and confidence — later blocks may revise them.
45
46### 5. Append to the lexicon
47
48```yaml
49# {persona_id}.lexicon.yaml — append block, never rewrite
50- extraction:
51 date: 2026-06-07
52 corpus: {files: 128, persona_turns: 1432, date_range: [2026-04-17, 2026-06-07]}
53 redaction: {patterns_run: true, hits_scrubbed: 0}
54 vocabulary:
55 coinages:
56 - term: vacuum
57 gloss: "an absence that is unrepresented anywhere durable; never a resting state"
58 examples: ["N/A is a vacuum", "vacuum field burn"]
59 domain_terms: [...]
60 inflections: [...]
61 idioms:
62 - phrase: "fix bases, not outputs"
63 meaning: "modify the source/template/pipeline, never the rendered artifact"
64 analogies:
65 - source: liturgy/theology
66 target: configuration management
67 examples: ["Tier-0 Liturgy", "the Reliquary", "testament"]
68 register:
69 profanity: "affection-marker, not hostility"
70 emphasis: ["em-dash chains", "bold imperatives"]
71 ideal_yearning: "…" # one sentence, persona's idiom, ≥3 quote citations
72 archetypal_pattern: "…"
73```
74
75### 6. Report
76Emit a short extraction report: new coinages since last block, terms that dropped out (candidate drift), and any revision to yearning/pattern with the evidence that forced it. Commit lexicon + report together.
77
78## Anti-patterns
79
80- **Extracting from assistant turns.** The most common contamination; the lexicon ends up describing the model, not the persona.
81- **Overwriting prior blocks.** Drift between blocks is the most valuable longitudinal signal; preserve it.
82- **Unredacted lexicons.** Verbatim example quotes can carry credentials/PII from the source sessions; redact at extraction time, not later.
83- **Synthesizing yearning from one loud session.** Require cross-session recurrence (≥3 sessions) before a theme reaches the yearning/pattern fields.
organvm-iv-taxis