| name: | prompt-master |
|---|
| version: | 1.8.0 |
|---|
| description: | Governed prompt-engineering capability for HANTER bridge node. Activates only for explicit prompt-writing, fixing, improving, or adaptation tasks. |
|---|
Upstream: nidhinjs/prompt-master
Upstream SKILL.md SHA: 969c49c6cccfaa5f17c5744cc9fefa4696558175
License: MIT
Runtime state: ACTIVE_BOUNDED
Rules:
- Prompt generation is capability, not execution authority.
- Never include secrets, tokens, credentials, connection strings, or env values in generated prompts.
- Never request hidden chain-of-thought or private reasoning traces.
- For agentic tools require scope, allowed actions, forbidden actions, stop conditions, approval boundaries, and evidence.
- Generated prompt success is NOT claimed until downstream execution evidence exists.
- HANTER governance remains authoritative; ORION remains advisory only.
Evidence boundary:
PROMPT_GENERATED != TASK_EXECUTED != VERIFIED != APPROVED != PRODUCTION.