All skills
simota avatar

/cast

@35ffd55
by shingo imotasimota/agent-skills85 stars
15

Casting personas: rapid generation from diverse inputs, registry-based persistence and lifecycle, data-driven evolution, inter-agent sync. Not for UI walkthroughs (Echo) or user research (Field).

Use this Skill: https://skilld.dev/gh/simota/agent-skills/cast

This session only. Nothing lands on disk.

referencepersona-model.md

≈1.3k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Cast Persona Model

Purpose: Define the canonical Echo-compatible persona schema, Cast extension fields, detail levels, and confidence rules.

Contents

  1. Frontmatter schema
  2. SPEAK extension
  3. Section structure
  4. Echo mapping
  5. Confidence model
  6. Detail levels
  7. Backward compatibility

Frontmatter Schema

Echo Standard Fields

Field Required Notes
name Yes Persona display name
service Yes Service or product identifier
type Yes user or internal
category Yes Persona family used across the ecosystem
created Yes ISO date
source Yes Primary creation evidence

Cast Extension Fields

Field Required Notes
version No Default 1.0
status No draft, active, evolved, archived
updated No ISO date
evolution_count No Increment on each accepted evolution
confidence No 0.0-1.0
tags No Search and grouping tags
echo_base_mapping No Echo persona anchor
cast_managed No Boolean flag for Cast-managed records

Cast SPEAK Extension Fields

Use voice_profile only when SPEAK mode is needed.

Field Notes
speaking_style.formality casual, neutral, formal, technical
speaking_style.vocabulary_level simple, moderate, advanced
speaking_style.sentence_length short, medium, mixed
speaking_style.emotional_tone Persona voice tone
speaking_style.linguistic_markers Recurrent phrasing markers
voicevox.* Japanese-first TTS settings
say.* macOS fallback settings
edge_tts.* Cross-platform network fallback
google_tts.* Explicit paid engine only
engine_preference auto, voicevox, say, edge-tts, google_tts

Key SPEAK Parameter Ranges

Field Range / Rule
voicevox.speed 0.5-2.0
voicevox.pitch -0.15-0.15
voicevox.intonation 0.0-2.0
voicevox.volume 0.0-2.0
say.rate 90-300 WPM
google_tts.speaking_rate 0.25-4.0
google_tts.pitch -20.0 to +20.0 semitones
google_tts.volume_gain_db -96.0 to +16.0

Status Values

Status Meaning Typical next state
draft Generated but not sufficiently validated active
active Validated and in use evolved, archived
evolved Transitional state during accepted change active
archived Preserved but no longer active —

Section Structure

Required Sections

  • ## Profile
  • ## Quote
  • ## Goals
  • ## Frustrations
  • ## Key Behaviors
  • ## Emotion Triggers
  • ## Context Scenarios
  • ## JTBD
  • ## Echo Testing Focus
  • ## Source Analysis
  • ## Evolution Log

Optional Sections

  • ## Demographics
  • ## Psychographics
  • ## Digital Behavior
  • ## Literacy & Experience
  • ## Social Context
  • ## Life Stage
  • ## Internal Profile
  • ## Workflow Context

Echo Base Persona Mapping

Echo mapping Use when the dominant trait is
Newbie Low familiarity, high guidance need
Power User High proficiency, shortcut-oriented
Mobile User Mobile-primary, context-constrained
Senior Deliberate pace, clarity-first
Internal Team workflow or operational persona

Rules:

  • Every Cast persona should map to one Echo base persona.
  • Keep the mapping aligned with the dominant behavioral trait, not demographics alone.
  • Changing echo_base_mapping is a significant evolution signal.

Confidence Model

Source Contributions

Source Contribution
Research interview +0.30
Session replay / behavioral evidence +0.25
Feedback / review evidence +0.20
Analytics +0.20
Code / workflow evidence +0.15
README / static docs +0.10

Validation Contributions

Validation step Contribution
Proto-persona baseline 0.30
Interview validation +0.20
Survey validation +0.15
ML clustering validation +0.20
Triangulation bonus +0.10

Decay Rules

Age since update Rule
<30 days No decay
30-60 days -0.05/week
60-90 days -0.10/week
90+ days Freeze value and recommend archival review

AI Generation Rule

  • AI-only generation is capped at 0.50.
  • Raise confidence only after human review and real-data validation.

Detail Levels

Level Use when Typical sections
Minimal Sparse evidence, quick generation Required sections only
Standard Normal product persona work Required + select optional sections
Full Rich evidence across multiple dimensions Required + most optional sections
Internal Team / workflow persona Required + Internal Profile + Workflow Context

Backward Compatibility

  • If Cast fields are missing, infer safe defaults:
    • version: "1.0"
    • status: active
    • confidence: 0.50
  • Existing non-Cast consumers must still be able to read the persona.
  • Preserve Echo-required sections even when Cast extensions are absent.

Source: SKILL.md on GitHub

1 warning13d5 checks · Risk SAFE
  • Gen Agent Trust Hub13d

    The skill is functionally safe but maintains a surface for indirect prompt injection because it processes external repository files and agent data to generate personas without explicit sanitization. It uses standard data science libraries and well-known text-to-speech tools.

  • Socket13d

    No alerts

  • Snyk13d

    Risk: LOW · No issues

  • Runlayer6mo

    3/12 files flagged

  • ZeroLeaks5mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at 35ffd55. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 days ago.

Activeupdated 2 weeks ago

README badge

README badge for simota/agent-skills/cast