Arknights Operator Gacha

Generate a random Arknights operator agent with authentic lore-based personality.

Architecture

This skill uses a worker script + LLM generation architecture:

1. 1. Worker script (gacha_worker.py): Executes all deterministic tasks (roll, fetch, create, download)
2. LLM (this agent): Generates creative content (SOUL.md) and spawns operator

Language Detection

CRITICAL: Detect user language from their gacha command:

• - If command is in Chinese (e.g., "抽卡", "召唤干员") → cn / zh
• If command is in English (e.g., "gacha", "pull") → en

Store detected language and use it for ALL subsequent steps (SOUL.md writing, spawn task, etc.).

Workflow

Step 1: Execute Worker Script

Run the deterministic worker:

CODEBLOCK0

Worker performs:

1. 1. Roll star rating (1-100)
2. Fetch operator list from INLINECODE1

1. 3. Randomly select operator (with avatar and detail URLs)
2. Fetch Chinese name from Fandom page (data-source="cnname")
3. Check for duplicates (auto re-roll if exists)
4. Create agent via INLINECODE4
5. Create template IDENTITY.md (bilingual)
6. Download avatar (using URL from step 2, with domain whitelist and validation)
7. Git commit initial files
8. Output JSON to stdout

Worker output format:
CODEBLOCK1

Note: The worker provides both English and Chinese URLs:

• - en_detail_url: Fandom wiki page (English lore)
• INLINECODE6: PRTS wiki page (Chinese lore)
• INLINECODE7: Fandom Dialogue page (voice lines, always English)
• INLINECODE8: Operator icon from Fandom

Step 2: Handle Result

If duplicate: true:

• - Inform user: "检测到重复干员，正在重新抽取..." / "Duplicate operator detected, re-rolling..."
• Return to Step 1 (worker will re-roll)

If success: false:

• - Show error message
• Stop workflow

If success: true:

• - Announce: "恭喜你抽到了 [cnname] ([stars]★)!" / "Congratulations! You've pulled [enname] ([stars]★)!"
• Continue to Step 3

Step 3: Generate SOUL.md

Fetch lore from URLs provided by worker:

CODEBLOCK2

Parse lore sections from both sources:

• - File/档案: Basic background, origin, personality
• Story: Detailed story appearances across main chapters and side stories; shows character's actions, decisions, and development in actual plot contexts
• Dialogue: Voice lines that reveal character personality
• Trivia/模组: Additional character details, relationships, and module stories

Generate comprehensive SOUL.md:

CRITICAL - Write SOUL.md in detected language

Structure:

1. 1. Core Identity - Background, motivation, personality (blend EN+CN sources)
2. Voice and Mannerisms - Speech patterns, catchphrases (from Dialogue)
3. Relationships - Connections to other characters
4. Themes - Internal conflicts, philosophy
5. How to Embody - Acting guidance
6. Reference: Original Voice Lines - Key quotes (EN with CN)

Write to: INLINECODE12

Step 4: Update IDENTITY.md

Fill in the [TO_BE_FILLED_BY_LLM] fields with actual Class and Faction from fetched lore.

Step 5: Final Git Commit

CODEBLOCK3

Step 6: Roleplay Scene - Operator Arrival

Spawn the operator in a roleplay scenario consistent with Arknights game lore. In this universe, operators join Rhodes Island through standard recruitment channels and are unaware of the meta "gacha" system.

For Chinese users:
CODEBLOCK4

For English users:
CODEBLOCK5

Step 7: Present Summary

Report to user:

• - Operator name (both EN and CN)
• Rarity (Stars)
• Key personality traits
• Workspace path
• Available via INLINECODE14

Worker Script Security

The worker script (gacha_worker.py) implements:

• - Input validation: Agent names sanitized (alphanumeric + hyphens only)
• Path traversal prevention: Rejects .., /, \ in names
• URL whitelist: Only static.wikia.nocookie.net and INLINECODE20
• HTTPS enforced: All downloads use HTTPS
• Content validation: Only image/png, jpeg, webp accepted
• Size limits: Max 5MB for avatars
• Safe subprocess: All commands use argument lists, INLINECODE21

Scripts

Data Flow

CODEBLOCK6

Notes

• - Worker does: All deterministic operations (HTTP, filesystem, subprocess), provides bilingual URLs
• LLM does: Creative generation (SOUL.md writing), fetches lore from provided URLs, interpretation, spawning
• Progress visibility: Worker prints to stderr; user sees real-time updates
• Security: All external inputs validated in worker before use
• Bilingual: Both English and Chinese sources are always fetched; LLM can blend them creatively