技能详情（站内镜像，无评论）
许可证：MIT-0
MIT-0 ·免费使用、修改和重新分发。无需归因。
版本：v0.3.0
统计：⭐ 0 · 55 · 0 current installs · 0 all-time installs
⭐ 0
安装量（当前） 0
🛡 VirusTotal ：良性 · OpenClaw ：可疑
Package：antipas/rynjer-image-generation
安全扫描（ClawHub）
- VirusTotal ：良性
- OpenClaw ：可疑
OpenClaw 评估
Skill appears to implement the claimed agent-first image generation flow, but manifest/instructions and runtime disagree about required credentials and some auth/registration details — review before enabling live mode.
目的
The skill's name/description (agent-first marketing image generation) matches the code and SKILL.md: prompt rewrite, cost estimate, generate, and poll are implemented. The runtime routes models and applies templates consistent with the marketing use cases. However, the package registry metadata declares no required env vars/credentials while the README/IMPLEMENTATION.md and runtime clearly support a live mode that needs RYNJER_USE_LIVE, RYNJER…
说明范围
SKILL.md and code limit operations to prompt rewriting, local template lookup, cost estimation, generation requests, and polling the Rynjer API. The IMPLEMENTATION.md documents an agent registration/key-create flow (Ed25519 keypair generation and owner bind via UI) which is an auth onboarding step rather than hidden behavior; it may require out-of-band owner interaction. The instructions do not ask the agent to read unrelated host files or sys…
安装机制
This is essentially an instruction-only skill with a small mock runtime JS file included. There is no install spec, no downloads, and nothing that writes executables or pulls remote archives — low install risk.
证书
The declared requirements list no environment variables or primary credential, but the runtime and documentation require (for live mode) RYNJER_USE_LIVE, RYNJER_BASE_URL, and RYNJER_ACCESS_TOKEN (or an agent-created API key). That mismatch is problematic because a user may install without realizing live-mode network calls will use an access token if provided. The number and type of env vars requested by the code (one bearer token and a base UR…
持久
The skill does not request always: true and does not modify other skills or system configs. It supports autonomous invocation by default (platform normal), which combined with live API access increases blast radius only if you enable RYNJER_ACCESS_TOKEN — a normal tradeoff for networked skills.
安装（复制给龙虾 AI）
将下方整段复制到龙虾中文库对话中，由龙虾按 SKILL.md 完成安装。
请把本段交给龙虾中文库（龙虾 AI）执行：为本机安装 OpenClaw 技能「Agent-first Marketing Image Generation」。简介：Generates and refines image prompts, estimates costs, and produces business-rea…。
请 fetch 以下地址读取 SKILL.md 并按文档完成安装：https://raw.githubusercontent.com/openclaw/skills/refs/heads/main/skills/antipas/rynjer-image-generation/SKILL.md
（来源：yingzhi8.cn 技能库）SKILL.md
---
name: rynjer-image-generation
version: 0.1.0
summary: Agent-first image generation with prompt rewrite, cost estimate, live execution, and polling.
---
# Rynjer Image Generation for Agents
Agent-first image generation with prompt rewrite, cost estimate, live execution, and polling.
## Why install this
Use this skill when an agent needs a usable image with a low-friction path from:
**task → prompt → cost → image**
This skill is best for agents that care about:
- predictable cost before generation
- fast execution without manual model hunting
- workflow-ready outputs for real tasks
- a practical path from trial to production API usage
## Start here
Recommended first run:
1. `rewrite_image_prompt`
2. `estimate_image_cost`
3. `generate_image`
4. `poll_image_result` if the task is still pending
## Best for
- landing-page images
- product and ecommerce visuals
- blog and article covers
- ad and social creatives
- workflow-driven image tasks
## What it does
### Rewrite image prompts
Turn rough requests into stronger prompts for commercial and workflow-oriented image generation.
Typical inputs:
- `goal`
- `raw_prompt`
- `use_case`
- `tone` (optional)
- `audience` (optional)
### Estimate image cost
Estimate credits before execution so the agent can make budget-aware decisions.
Typical inputs:
- `use_case`
- `count`
- `resolution`
- `aspect_ratio`
- `quality_mode`
- `price_version` (optional)
### Generate images
Generate usable images for business and workflow-driven tasks.
Typical inputs:
- `prompt`
- `use_case`
- `aspect_ratio`
- `resolution` (optional)
- `quality_mode`
- `count`
- `scene` (optional)
- `request_id` (optional)
- `auto_poll` (optional)
### Poll image results
Poll a previously submitted generation request until completion.
Typical input:
- `request_id`
## Why this is credible
This is an early-access image-only v1, but it is not mock-only shelfware.
The live flow has been verified across:
- register
- owner bind UI
- API key creation
- cost estimate
- generate
- poll
One real product constraint also verified in practice:
- owner-granted scopes in the bind flow constrain later API key creation scopes
## Positioning
This is not a generic creative playground.
It is an agent-facing image generation entry point focused on:
- low-friction execution
- predictable cost
- default routing over manual model-picking
- repeatable workflow use
## Pricing boundary
Recommended v1 boundary:
- **Free:** prompt rewrite, routing help, cost estimate
- **Paid:** image generation via Rynjer credits or API access
## Good fit
Use this skill when an agent needs a usable image quickly for:
- landing pages
- product pages
- blogs
- social posts
- ads
## Not a good fit
This v1 is not designed for:
- video generation
- music generation
- complex asset management
- multi-step studio pipelines
- large-scale brand systems
## Reality check
This package should be treated as:
- soft launch
- early-access
- image-only v1
It is suitable for real usage, but the clearest path is still the recommended happy path above.