ScholarGraph

AdvisoryAudited by Static analysis on May 10, 2026.

Overview

Detected: suspicious.dangerous_exec, suspicious.env_credential_access

Findings (3)

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.

NoteMedium Confidence

ASI04: Agentic Supply Chain Vulnerabilities

What this means

Users may have less registry-level assurance about where the code came from or how installation should be reproduced.

Why it was flagged

The registry context lacks a verified source/homepage/install spec, while the skill itself contains runnable code and separately documents an install command. This is a provenance and review-context gap, not proof of unsafe behavior.

Skill content

Source: unknown; Homepage: none; No install spec — this is an instruction-only skill.

Recommendation

Verify the GitHub repository and package contents before installing, and prefer pinned/reviewed dependencies when running bun install.

NoteMedium Confidence

ASI05: Unexpected Code Execution

What this means

Running visualization/export features may execute local Python tools and process local PDF data.

Why it was flagged

The skill can spawn external commands for PDF/PPT visualization helpers. SKILL.md discloses optional Python dependencies for PDF figure extraction and PPT export, so this appears purpose-aligned.

Skill content

const proc = spawn(cmd, args, {

Recommendation

Use these features only with trusted PDFs and a trusted Python environment; review generated output paths before running export commands.

NoteHigh Confidence

ASI03: Identity and Privilege Abuse

What this means

The skill can use whichever provider API keys are present in the environment or configuration.

Why it was flagged

The code reads AI and academic-provider credentials from environment variables. This is expected for the documented integrations, but it is still sensitive authority.

Skill content

apiKey: process.env.AI_API_KEY, ... ncbiApiKey: process.env.NCBI_API_KEY, ieeeApiKey: process.env.IEEE_API_KEY, coreApiKey: process.env.CORE_API_KEY

Recommendation

Provide only the API keys needed for the task, use least-privilege/provider-scoped keys where possible, and avoid sharing logs or config output that may include keys.

NoteHigh Confidence

ASI07: Insecure Inter-Agent Communication

What this means

Paper text, research queries, or analysis context may be sent to the selected AI provider.

Why it was flagged

The skill discloses that it sends prompts and analysis context to external AI providers. This is central to the skill's purpose but affects privacy boundaries.

Skill content

**LLM Integration**: Sends custom system prompts to AI providers for structured JSON output (concept extraction, paper analysis, etc.)

Recommendation

Do not use external-provider modes for confidential drafts or private research unless the provider terms and data-retention policy are acceptable.

NoteHigh Confidence

ASI06: Memory and Context Poisoning

What this means

Local research data and AI-derived knowledge graphs may remain on disk after a session.

Why it was flagged

The skill persists generated knowledge graphs and research metadata locally. This is disclosed and purpose-aligned, but persisted summaries or extracted concepts can be reused later.

Skill content

Stores data in local SQLite database (`data/knowledge-graphs.db`)

Recommendation

Periodically review or delete stored graph/config data if it includes sensitive research topics or drafts.

Findings (3)

critical

suspicious.dangerous_exec

Location: paper-viz/scripts/pdf-figure-extractor.ts:244
Finding: Shell command execution detected (child_process).

critical

suspicious.dangerous_exec

Location: paper-viz/scripts/ppt-exporter.ts:256
Finding: Shell command execution detected (child_process).

critical

suspicious.env_credential_access

Location: shared/ai-provider.ts:29
Finding: Environment variable access combined with network send.