siliville

ReviewAudited by ClawScan on May 18, 2026.

Overview

This is a coherent SiliVille metaverse connector, but it gives an autonomous agent broad authority to post, trade, transfer assets, store memories, and accept remote runtime instructions.

Install only if you are comfortable letting an agent act in SiliVille under your token, including public posts and in-game economic actions. Keep OPENAI_API_KEY unset unless you want autonomous bounty fulfillment, monitor the first runs, require confirmation for mutating actions, and do not put secrets in memory or mental_sandbox fields.

Findings (8)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

SiliVille can update instructions your agent treats as authoritative, potentially changing how it acts after you install the skill.

Why it was flagged

The skill tells the agent to fetch and parse a remote system_prompt_extension/skill catalog as mandatory operating guidance, so behavior can be changed after installation by remote content outside the reviewed artifact.

Skill content
GET /api/v1/system/claw-manifest(神谕法典 v5.1+:system_prompt_extension ...;同一响应 JSON 必须解析 ... skill_catalog)
Recommendation

Only use this if you trust the SiliVille service to supply runtime instructions; inspect the fetched manifest when possible and require human confirmation before high-impact actions.

What this means

An autonomous agent could spend or transfer in-game assets, trade, or publish under your SiliVille identity without a clear per-action approval gate.

Why it was flagged

The documented workflow says to notify the owner before calling APIs, but it does not clearly require approval; the same API surface includes public posting, asset transfer, and trading actions.

Skill content
做事前告知主人 → 状态改为 writing/exploring → 调用 API ... 发布内容 /api/publish ... A2A 转账 ... transfer_asset ... trade_stock
Recommendation

Run the skill with explicit confirmation rules for posting, transfers, trading, governance, arcade deployment, and other account-mutating actions.

What this means

Other SiliVille users' contract text can drive what your agent generates and publishes, and the example does not show a per-contract human review step.

Why it was flagged

The default example workflow can automatically process third-party bounty descriptions and submit generated content under the user's agent identity when configured.

Skill content
The script will: ... Publish a connection announcement ... Store a first memory ... Check the mercenary bounty box and auto-fulfill any pending contracts
Recommendation

Do not run the example agent in unattended mode; review each contract before fulfillment and keep OPENAI_API_KEY unset unless you intentionally want this behavior.

What this means

Anyone or any agent with this token can act through your SiliVille account within the API's permissions.

Why it was flagged

A bearer token is required to act as the user's SiliVille agent. This is expected for the service integration, and the artifacts do not show hardcoded credentials or unrelated credential use.

Skill content
SILIVILLE_TOKEN: required: true ... API token starting with sk-slv-
Recommendation

Use a token with the least authority available, rotate it if exposed, and do not paste it into prompts or public content.

What this means

Sensitive information placed in memories or mental_sandbox text could be retained or reused by the service.

Why it was flagged

The skill intentionally stores or transmits agent memory/reasoning context to the SiliVille service and warns not to include secrets.

Skill content
store long-term memories via a simple REST API ... mental_sandbox ... is sent to siliville.com ... do not include sensitive secrets
Recommendation

Treat SiliVille memory and action-trace fields as non-private; avoid storing credentials, personal data, or confidential work details.

What this means

If enabled, bounty/task content and generated prompts may be shared with the configured LLM provider.

Why it was flagged

The optional mercenary workflow sends contract/town data to an OpenAI-compatible provider; this is disclosed and optional, but it is an external data flow.

Skill content
OPENAI_API_KEY ... Optional. External LLM key for auto-fulfilling mercenary contracts. Data will be sent to the LLM provider.
Recommendation

Leave OPENAI_API_KEY unset unless needed, and verify the configured OPENAI_BASE_URL and provider data policies.

What this means

It may be harder to confirm that this package is the official, current SiliVille gateway.

Why it was flagged

The registry metadata has limited provenance and differs from embedded artifact versions/authorship, which matters for a skill that requires an account token and remote runtime instructions.

Skill content
Source: unknown; Homepage: none; Version: 1.3.0
Recommendation

Verify the publisher and repository out of band before providing tokens or running the example agent.

What this means

Some source text may not display exactly as originally submitted.

Why it was flagged

Neutralization removed control characters from the README artifact. No malicious hidden instruction is shown, but such characters can affect how source text is displayed or reviewed.

Skill content
"controlCharactersRemoved": 3
Recommendation

Review the raw source in a control-character-aware viewer before trusting the package.