ClawHub
Back to skill

Security audit

Summarize Pro

Security checks across malware telemetry and agentic risk

Overview

This is a local summarization skill with disclosed local history, stats, saved summaries, and templates; it does not show network use or hidden behavior.

Install this only if you are comfortable with a local record of summarization activity. It keeps history and usage stats by default, and saved summaries or templates may contain sensitive details from your documents; review or delete ~/.openclaw/summarize-pro/ if you want to clear that local data.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (7)

Scope Creep

Medium
Confidence
88% confidence
Finding
The skill instructs creating and using a directory under ~/.openclaw/summarize-pro during first-run setup, which is broader than the stated local storage boundary implied by a pure summarization function and creates persistent local state automatically. While still local, automatic filesystem writes increase risk because user content and metadata may be stored without an explicit per-action consent flow.

Context-Inappropriate Capability

Low
Confidence
78% confidence
Finding
Gamification and achievement tracking expand the data collected and retained beyond what is needed to summarize text, such as streaks, favorites, and usage metrics. This is a privacy and scope-minimization issue rather than a direct exploit, but it increases the amount of behavioral data stored locally and the blast radius if those files are later exposed.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The activation triggers include very common phrases such as 'summary', 'compare', 'meeting notes', and 'help', which can cause the skill to activate unexpectedly in normal conversation. Unintended activation is dangerous here because the skill also reads/writes persistent local files, potentially logging or storing content the user did not intend to process with this skill.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill automatically logs summary activity and updates stats, but the user-facing description does not clearly warn that summary content and metadata may be persisted locally by default. This creates a consent and transparency problem, especially because users may paste sensitive emails, reports, or conversations expecting ephemeral processing.

Ssd 3

Medium
Confidence
95% confidence
Finding
Auto-logging every summary to history creates retention of user-derived metadata such as topics, timestamps, and word counts, which can reveal sensitive habits and subject matter over time. Even without network exfiltration, persistent natural-language traces increase exposure to later local compromise, accidental disclosure, or shoulder-surfing by other users of the machine.

Ssd 3

Medium
Confidence
96% confidence
Finding
The saved summaries feature persists summarized user content, topics, and timestamps, which may include condensed versions of confidential source material. Summaries often preserve the most sensitive facts, so storing them long-term can create a high-value local cache of private information even if the original text is no longer present.

Session Persistence

Medium
Category
Rogue Agent
Content
## FEATURE 19: Custom Templates

When user says **"create template [name]"** or **"my templates"**:

Let users define their own summary format:
Confidence
80% confidence
Finding
create template [name]"** or **"my templates"**: Let users define their own summary format: ``` User: "create template standup" Bot: What sections should your 'standup' template include? User: "wha

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.