v1.0.0

Add Siliconflow Provider 1

BenignClawScan verdict for this skill. Analyzed May 1, 2026, 5:38 AM.

Analysis

This is a coherent instruction-only guide for adding SiliconFlow as an OpenClaw model provider, but it persists an API key and changes future model routing, so users should review the configuration before applying it.

GuidanceInstall only if you intend OpenClaw to use SiliconFlow as a model provider. Before applying the patch, confirm the endpoint, referral choice, paid-model pricing, fallback settings, and API key handling; keep the backup so the configuration can be reverted.

Findings (5)

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.

Abnormal behavior control

Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.

Cascading Failures

SeverityLowConfidenceHighStatusNote

SKILL.md

cp ~/.openclaw/openclaw.json ~/.openclaw/openclaw.json.backup.$(date +%Y%m%d_%H%M%S) ... 通过 `gateway config.patch` 添加 SiliconFlow provider ... fallback 链接入

The instructions modify persistent OpenClaw configuration, including provider and fallback behavior. This is the stated purpose and includes a backup step, but the change can affect future sessions and model routing.

User impactFuture agent runs may use SiliconFlow models, including paid models, without the user re-entering the configuration each time.

RecommendationReview the exact patch, keep the backup, and verify default and fallback model settings after applying the configuration.

Human-Agent Trust Exploitation

SeverityInfoConfidenceHighStatusNote

SKILL.md

如果还没有 SiliconFlow 账号，请通过邀请链接注册（双方均获赠额度）：https://cloud.siliconflow.cn/i/ihj5inat

The setup flow includes a referral/invitation link. It is disclosed as mutually beneficial, so this is not deceptive, but users should know it may benefit the referrer.

User impactUsing the provided registration link may credit the referrer as well as the user.

RecommendationUse the referral link only if comfortable with that arrangement, or navigate to SiliconFlow directly.

Agentic Supply Chain Vulnerabilities

SeverityLowConfidenceMediumStatusNote

_meta.json

"ownerId": "kn7axa0czv9a71epzsmhxa10hd80rkx7", "slug": "add-siliconflow-provider"

The embedded metadata does not match the supplied registry metadata for this package, which lists a different owner and the slug `add-siliconflow-provider-1`; the registry also lists the source as unknown. This is a provenance note for a config-changing skill.

User impactUsers have less assurance about who authored or republished the skill.

RecommendationVerify the skill's origin and review the instructions before letting it change OpenClaw configuration.

Permission boundary

Checks whether tool use, credentials, dependencies, identity, account access, or inter-agent boundaries are broader than the stated purpose.

Identity and Privilege Abuse

SeverityLowConfidenceHighStatusNote

SKILL.md

| API Key | 在 [控制台](https://cloud.siliconflow.cn/account/ak) 创建，格式 `sk-xxx` | ... "apiKey": "<YOUR_API_KEY>"

The skill requires a SiliconFlow API key and shows it being placed into OpenClaw provider configuration. This is expected for a model-provider integration, but it grants access to the user's SiliconFlow account and possible billing.

User impactThe configured API key may be used for future model calls and paid usage on the user's SiliconFlow account.

RecommendationUse a revocable API key, monitor usage and billing, and avoid sharing the key outside the intended OpenClaw configuration.

Sensitive data protection

Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.

Insecure Inter-Agent Communication

SeverityLowConfidenceHighStatusNote

SKILL.md

| Base URL | `https://api.siliconflow.cn/v1` | ... | API 协议 | `openai-completions` |

The skill configures OpenClaw to use a third-party model endpoint. This is purpose-aligned, but future prompts and model traffic may be sent to SiliconFlow when these models are selected or used as fallback.

User impactSensitive prompts or project context could leave the local environment and be processed by the SiliconFlow service.

RecommendationConfirm SiliconFlow's data-handling policy and only route data to this provider if it is acceptable for the user's privacy and compliance needs.