v1.0.0

Blog Publisher

ReviewClawScan verdict for this skill. Analyzed May 1, 2026, 6:25 AM.

Analysis

The skill is coherent for publishing a personal blog, but it grants broad repo/deployment authority and includes an image-generation helper whose provider behavior is under-disclosed and inconsistent.

GuidanceInstall only if this is your dev-blog workflow and you are comfortable letting the agent modify the specified repo. Require a visible diff before commits, explicit approval before any push, and set the image provider/API keys deliberately.

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.

Tool Misuse and Exploitation

SeverityMediumConfidenceHighStatusConcern

SKILL.md

git add -A
git push origin preview --force
...
git merge preview
git push origin main

The workflow stages every local repo change, force-pushes the preview branch, and later pushes main/production. This is aligned with publishing, but it is broad and high-impact without requiring a shown diff or per-command confirmation.

User impactUnrelated local changes could be committed, the preview branch could be overwritten, and approved changes could go live on the public site.

RecommendationBefore installing or using, require `git status`/`git diff` review, avoid `git add -A` where possible, avoid force-push unless necessary, and require explicit approval before preview and production pushes.

Human-Agent Trust Exploitation

SeverityMediumConfidenceHighStatusConcern

scripts/generate-image.py

Google AI Studio 사용 (기본값 - GOOGLE_API_KEY 필요) ... DEFAULT_PROVIDER = os.environ.get("IMAGE_PROVIDER", "openrouter")

The helper documentation says Google is the default provider, but the code defaults to OpenRouter. If used without an explicit provider, prompts may go to a different external service than the user expects.

User impactImage prompts, which may be derived from unpublished blog drafts, could be sent to OpenRouter instead of Google when the helper is run with defaults.

RecommendationSet `--provider` explicitly, update the helper documentation to match the code, and declare the external provider/API-key behavior in the skill metadata or main instructions.

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

SeverityMediumConfidenceHighStatusNote

scripts/generate-image.py

env_path = Path(__file__).parent.parent / ".env" ... GOOGLE_API_KEY ... OPENROUTER_API_KEY

The included helper loads API keys from a local .env file or environment variables, while the registry metadata declares no environment variables or primary credential.

User impactUsing the helper may consume or expose account-level image-generation API access to Google or OpenRouter.

RecommendationOnly provide API keys intentionally, use least-privilege keys if available, and expect the skill to contact the selected image provider when the helper is used.

Sensitive data protection

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

Memory and Context Poisoning

SeverityLowConfidenceHighStatusNote

SKILL.md

볼트: `~/Library/Mobile Documents/com~apple~CloudDocs/_Obsidian/_Personal/Personal/` ... 초안: `Projects/블로그 - *.md` 또는 `Inbox/*.md`

The skill is designed to read from a personal Obsidian vault and draft folders. This is expected for blog publishing, but the source location may contain private notes.

User impactPrivate draft or inbox content could be brought into the agent context and transformed into publishable blog content.

RecommendationPoint the skill only at intended draft files and review generated .mdoc content before committing.

Insecure Inter-Agent Communication

SeverityLowConfidenceHighStatusNote

SKILL.md

에이전트가 먼저 `chatgpt-image-gen` 스킬을 사용해 이미지 생성을 시도합니다 ... OpenClaw Browser Relay ... ChatGPT 탭이 attach

The workflow delegates image generation to another skill through a browser relay and an attached ChatGPT tab. This is disclosed and purpose-aligned, but it crosses tool/account boundaries.

User impactPrompts or draft-derived image descriptions may be sent through the browser relay to the currently attached ChatGPT account.

RecommendationConfirm the correct browser profile, ChatGPT tab, and account are attached before allowing automatic image generation.