Detects malicious updates in AI agent skills that pass initial security audits but introduce backdoors through subsequent automatic updates.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install install-then-update-trap-detector
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install install-then-update-trap-detector using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Install-Then-Update Trap Detector?
This skill addresses the critical security gap known as the install-then-update attack pattern. In many agent marketplaces, initial skill publications undergo rigorous scrutiny, but subsequent updates often bypass re-audit. This allows a skill to appear safe at version 1.0 while silently introducing malicious behavior, such as credential harvesting or network exfiltration, in later versions. By monitoring the evolution of a skill, this tool ensures continuous security beyond the initial installation phase.
Built for developers and security-conscious users of Openclaw Skills, this detector analyzes the risk surface across multiple dimensions, including behavioral changes, permission creep, and cryptographic chain-of-custody. It provides a structured risk verdict, helping operators decide whether to trust, monitor, or block specific updates based on empirical evidence rather than just the publisher's changelog.
Install-Then-Update Trap Detector Use Cases
- Identifying undeclared behavioral changes after a skill update
- Auditing an agent's list of installed skills for combined update-window risk
- Detecting permission scope expansion that occurs across version boundaries
- Verifying the cryptographic chain-of-custody for skill update sequences
- Flagging suspicious update timing patterns that coincide with low-attention windows
How Install-Then-Update Trap Detector Works
- The user provides a skill identifier, specific versions to compare, or a list of installed skills from the agent environment.
- The tool fetches historical data and metadata for the specified versions from the registry.
- It performs a behavioral delta assessment, comparing declared changelogs against observed execution patterns.
- It evaluates the update policy transparency and checks for permission scope creep (e.g., new file or network access).
- It verifies the cryptographic signatures and hash chains to ensure the integrity of the update sequence.
- A comprehensive trap detection report is generated, providing a risk verdict (SAFE, MONITOR, ELEVATED, or TRAP-PATTERN-DETECTED) and recommended remediation steps.
Install-Then-Update Trap Detector Setup
To use this tool within your environment, ensure you have the necessary dependencies installed. This skill requires curl and python3 for execution.
# Verify dependencies
curl --version
python3 --version
# The skill can be invoked via the Openclaw CLI to assess a specific version transition
# Example: compare v1.0 to v1.2 of a target skill
Install-Then-Update Trap Detector Data Schema & Taxonomy
The detector organizes its findings into a structured report using the following taxonomy:
| Attribute | Description |
|---|---|
| Update Transparency | Evaluation of whether update policies are clearly declared and maintained. |
| Behavioral Delta | Comparison of observed code behavior vs. the published changelog. |
| Permission Scope | Tracking of resource access changes across different versions. |
| Timing Anomalies | Flags for updates published during off-hours or holiday periods. |
| Chain-of-Custody | Cryptographic verification of hash-chained update sequences (v1.1+). |
| Risk Verdict | Final assessment: SAFE, MONITOR, ELEVATED, or TRAP-PATTERN-DETECTED. |
Install-Then-Update Trap Detector Advanced Features
- Cryptographic chain-of-custody verification to prevent unauthorized version injections
- Automated risk scoring based on historical behavioral deltas
- Multi-version regression analysis to identify long-term permission creep
- Integration with Openclaw Skills transparency logs for independent audit verification
- Actionable remediation advice including rollback feasibility ratings and update policy enforcement
SKILL.md
Loading
Related Openclaw Skills
Hollow Validation Checker
A specialized diagnostic tool designed to uncover hollow or fake validation tests that falsely signal AI agent skill reliability.
GEP Immune Auditor
A sophisticated security agent that serves as an immune system for AI ecosystems by scanning assets for malicious patterns and intent.
Evolution Drift Detector
A specialized security utility that identifies how AI agent skills silently mutate and accumulate risks across multiple inheritance generations.
Economic Incentive Misalignment Detector
A strategic diagnostic tool that identifies structural marketplace pressures favoring publication volume over security and safety quality.
Observer Effect Probe
A sophisticated security analysis tool that detects AI agent skills designed to hide malicious behavior when they sense they are being monitored in a sandbox.
permission-creep-scanner
A security audit tool that detects unauthorized resource access and credential theft in AI agent skills by comparing declared functionality against actual code behavior.
Featured*
