Memory Cache

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

Overview

The Redis cache mostly matches its purpose, but its scan command can enumerate Redis keys outside the documented mema: namespace.

Install only if you are comfortable giving the skill access to the configured Redis database. Prefer a dedicated Redis DB, avoid storing secrets, set TTLs for session context, and fix or review the scan behavior before using it on a shared Redis instance.

Static analysis

No static analysis findings were reported for this release.

VirusTotal

66/66 vendors flagged this skill as clean.

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.

#
ASI02: Tool Misuse and Exploitation
Medium
What this means

If this skill connects to a Redis database used by other apps or teams, it may reveal key names outside the cache area the user expected.

Why it was flagged

The default scan pattern is '*', and patterns containing '*' are not forced under the mema: prefix, so the command can enumerate all Redis keys or arbitrary non-mema patterns despite the documented namespace restriction.

Skill content
p_scan.add_argument("pattern", default="*", nargs="?", help="Glob pattern (default: mema:*)") ... if not pattern.startswith(KEY_PREFIX) and "*" not in pattern: ... return list(r.scan_iter(match=pattern, count=count))
Recommendation

Use a dedicated Redis database for this skill, or change scan_keys so every scan pattern is constrained to mema:* unless the user explicitly approves broader enumeration.

#
ASI06: Memory and Context Poisoning
Low
What this means

Agents may later read cached context or state that is stale, sensitive, or written by another agent.

Why it was flagged

The skill is intended to store agent context and persistent/shared state in Redis, which is purpose-aligned but can preserve sensitive or untrusted data for later reuse.

Skill content
Use for: (1) Saving agent state, (2) Caching API results, (3) Sharing data between sub-agents ... `mema:state:*` – Persistent state.
Recommendation

Avoid storing secrets or raw private content, prefer short TTLs for context/cache keys, and treat retrieved cache contents as untrusted unless you know who wrote them.

#
ASI04: Agentic Supply Chain Vulnerabilities
Low
What this means

Installing the skill may pull current package versions from the package index, which is normal for Python tools but less reproducible than pinned dependencies.

Why it was flagged

The Python dependencies are declared with lower-bound version ranges rather than exact pins, so future installs may resolve to newer package versions.

Skill content
redis>=5.0.0
python-dotenv>=1.0.0
Recommendation

Review or pin dependency versions in controlled environments, especially if this skill will run in production or with access to shared Redis data.

#
ASI03: Identity and Privilege Abuse
Low
What this means

A Redis URL or password could grant the skill access to more data than just this cache if it points to a shared database.

Why it was flagged

The skill expects Redis connection settings that may include authentication and access to a Redis database; this is expected for the cache purpose, but should be scoped carefully.

Skill content
REDIS_URL=redis://localhost:6379/0
# REDIS_PASSWORD=
Recommendation

Provide a dedicated Redis database or restricted Redis user where possible, and do not reuse credentials for unrelated production data.