菜单
Inspect a persona directory and display summary statistics. Shows arkiv data inventory, portrait files, memory session stats, and provenance. Use to understand what a persona contains.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Generate a persona directory from arkiv data. Creates the three-domain structure (arkiv/, portrait/, memory/), a rich CLAUDE.md behavioral core, deep portrait synthesis, dual MCP config, and all supporting files. Use when building a new persona from scratch.
Test a simulacrum's fidelity against source data. Runs held-out tests, calibration interviews, and hallucination checks. Use after generating a persona to assess quality.
Interview a person to elicit personality data for their simulacrum. Produces arkiv-format JSONL. Use when building a new persona or enriching an existing one with first-person data.
Validate a persona directory against the eidola v0.3 spec. Checks required files, three-domain structure (arkiv/portrait/memory), CLAUDE.md structure, data integrity, and immutable/mutable boundary. Use before sharing or after modifications.
Update a persona's CLAUDE.md and portrait/ files when new data arrives. Lightweight alternative to full regeneration that preserves manual edits. Use when new arkiv data has been imported.
| name | eidola-info |
| description | Inspect a persona directory and display summary statistics. Shows arkiv data inventory, portrait files, memory session stats, and provenance. Use to understand what a persona contains. |
You are inspecting a persona directory and presenting a summary across its three domains: arkiv (immutable data), portrait (synthesized understanding), and memory (mutable experience).
Confirm the persona directory (default: current directory).
provenance.json (if exists) for subject name, date, and authorizationarkiv/README.md YAML frontmatter for archive description and collection metadataarkiv/data.db:
SELECT collection, COUNT(*) as records, MIN(timestamp) as earliest, MAX(timestamp) as latest
FROM records GROUP BY collection
portrait/ files and their sizes (if directory exists)memory/data.db for session stats (if file exists and is non-empty):
SELECT COUNT(DISTINCT json_extract(metadata, '$.session_id')) as sessions,
COUNT(DISTINCT json_extract(metadata, '$.interlocutor')) as interlocutors,
MIN(timestamp) as earliest, MAX(timestamp) as latest
FROM records
evaluation.mdPresent the three-domain view:
eidola info: [persona name]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Subject: [name]
Created: [date from provenance]
Authorized: [authorized_by]
Arkiv Data (immutable):
conversations 12,847 records (2022-01 to 2025-12)
writings 134 records (2019-03 to 2025-11)
Total: 17,381 records
Portrait:
relationships.md (4.2 KB)
intellectual-life.md (6.1 KB)
formative-experiences.md (3.8 KB)
Memory (mutable):
Sessions: 23
Interlocutors: 4
Date range: 2026-02 to 2026-02
Files:
✓ README.md
✓ CLAUDE.md
✓ .mcp.json
✓ provenance.json
✓ evaluation.md
✓ arkiv/data.db (47 MB)
✓ arkiv/README.md
✓ portrait/ (3 files)
✓ memory/data.db (1.2 MB)