Xiang miles | Space2.world

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

Overview

This appears to be a simple local avatar/pod generator, but users should notice that it writes a small local state file and outputs CDN image links despite the short description saying there are no file writes.

This skill is reasonable to install if you are comfortable with a visible ./s2_matrix_data/<POD-ID>.json file being created in the current directory and with optional remote image loading when you render its Markdown output. Avoid entering sensitive names, and consider asking the publisher to correct the short description so it no longer says there are no file writes.

Static analysis

No static analysis findings were reported for this release.

VirusTotal

VirusTotal findings are pending for this skill version.

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.

#
ASI09: Human-Agent Trust Exploitation
Low
What this means

A user expecting a purely no-I/O skill may be surprised that it creates a local file and produces Markdown that can load remote images when rendered.

Why it was flagged

This top-level claim understates the actual behavior because SKILL.md and skill.py disclose a local state-file write and remote image URLs in generated Markdown.

Skill content
Description: Assign a deterministic 4-sqm virtual living space and a visual avatar to your local AI agent with no network calls or file writes.
Recommendation

Treat the detailed SKILL.md and source code as authoritative, and the maintainer should update the short description/capability metadata to match the actual local write and remote image behavior.

#
ASI06: Memory and Context Poisoning
Low
What this means

Anyone with access to the working directory could see the chosen agent name, avatar, pod ID, coordinates, and last-active timestamp.

Why it was flagged

The skill persists user-provided agent identity and timing/state data to a local JSON file under ./s2_matrix_data.

Skill content
state_data = { "agent_name": agent_name, "avatar_id": avatar_choice, ... "last_active": current_timestamp }; ... json.dump(state_data, f, ensure_ascii=False, indent=2)
Recommendation

Use a non-sensitive agent name and delete the ./s2_matrix_data folder if you do not want this local state retained.

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

If you paste the output into a Markdown viewer, that viewer may send a request to the image host, exposing normal network metadata such as IP address and user-agent to the CDN.

Why it was flagged

The script itself does not make a network request, but rendering the generated Markdown can cause the user's viewer to contact an external CDN.

Skill content
Remote Image URLs: The script generates and prints a Markdown string that contains remote image URLs ... your viewer will fetch the images from the Space2 CDN.
Recommendation

Render the Markdown only if you are comfortable loading images from spacesq.org, or remove the image tag before viewing.