Aigroup Financial Services Openclaw Release

Security checks across static analysis, malware telemetry, and agentic risk

Overview

Prompt-injection indicators were detected in the submitted artifacts (base64-block, unicode-control-chars); human review is required before treating this skill as clean.

Before installing, make sure you trust this publisher and the companion office/MCP tools you plan to use. Review the preflight/setup commands, install only needed runtimes, and avoid running the template-skill creator or document workflows on confidential files unless the local host skills and bundled binaries are acceptable for your environment. ClawScan detected prompt-injection indicators (base64-block, unicode-control-chars), so this skill requires review even though the model response was benign.

Static analysis

No static analysis findings were reported for this release.

VirusTotal

VirusTotal engine telemetry is currently stale for this artifact.

View on VirusTotal

Risk analysis

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.

#
ASI04: Agentic Supply Chain Vulnerabilities
Low
What this means

The skill may rely on local runtimes and package installation steps when producing Office deliverables.

Why it was flagged

The documentation discloses runtime tools and package installs that are not reflected as registry requirements. These are expected for document/model generation, but users should know they are part of the operating environment.

Skill content
Runtime dependencies (must be on PATH) ... `python3` ... `uvx` ... `node` ... `python-pptx` ... `pptxgenjs` ... `npm install pptxgenjs --omit=dev`
Recommendation

Install only the dependencies you need, prefer trusted package sources, and review any npm, uvx, pip, or shell command before running it.

#
ASI05: Unexpected Code Execution
Low
What this means

Setup can run local shell commands and affect the local OpenClaw gateway process.

Why it was flagged

The quickstart asks the user to restart the gateway and run a bundled shell preflight script. This is disclosed setup behavior and aligned with installing a plugin, but it is still local command execution.

Skill content
launchctl kickstart -k gui/$(id -u)/ai.openclaw.gateway
bash ~/.openclaw/extensions/aigroup-financial-services-openclaw/scripts/preflight.sh
Recommendation

Run these commands only from a trusted checkout/install and inspect the preflight script if you are using this in a sensitive environment.

#
ASI04: Agentic Supply Chain Vulnerabilities
Low
What this means

Document-generation workflows may use bundled executable components rather than only plain markdown instructions.

Why it was flagged

The bundle includes prebuilt .NET binaries/DLLs for DOCX handling. This is consistent with the advertised Office deliverable functionality, but bundled executables are a supply-chain surface users may want to review.

Skill content
listed in file manifest as bundled binary: MiniMaxAIDocx.Cli (124240 bytes), with related .dll files
Recommendation

Use the package from the official ClawHub source you intended, and in higher-assurance environments rebuild or verify bundled binaries before relying on them.

#
ASI02: Tool Misuse and Exploitation
Low
What this means

If invoked, this workflow can create new local skill files from a PowerPoint template.

Why it was flagged

The PPT template workflow can read a user-provided template, write a generated skill directory, and use shell commands. That is central to creating a reusable template skill, but it can modify local files and future skill assets.

Skill content
allowed-tools: ["Read", "Write", "Bash", "Glob"] ... Generate skill directory with assets/ and SKILL.md ... Package the skill
Recommendation

Use this command only with templates you trust and review the generated skill before installing or sharing it.

#
ASI07: Insecure Inter-Agent Communication
Low
What this means

Your documents, models, or financial inputs may be processed by companion tools or host skills if those routes are available.

Why it was flagged

The suite discloses routing some deliverable work through companion MCP or host-provided skills. This is purpose-aligned, but financial source material may be passed to those installed local/host components.

Skill content
Word -> `minimax-docx` -> `aigroup-mdtoword-mcp__markdown_to_docx` ... PPT delegates entirely to host MiniMax PPT skills
Recommendation

Confirm which companion MCP services and host skills are installed and trusted before using the suite with confidential materials.