Back to skill

Security audit

Software Data Self Improving Developer Helper

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill benign with high confidence: the completed scans found no reported malicious behavior, while flagging broad activation wording as a limited concern.

The verdict is benign, but broad activation wording may lead to unintended use in unrelated conversations. Consider whether that behavior suits you before installing; the scan does not guarantee the skill is risk-free.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (6)

The activation keywords include common words without scope constraints.

Medium
Category
Trigger Abuse
Confidence
97% confidence
Finding

The usage section lists standalone keywords such as self, improving, errors, and corrections, which can occur in many unrelated requests. Although specific invocation examples follow, the file does not restrict these broad keywords to self-improving agent development or provide exclusion conditions.

Content

No source excerpt is available for this finding.

Activation keywords include broad everyday terms without a clear domain restriction.

Medium
Category
Trigger Abuse
Confidence
99% confidence
Finding

The trigger list includes standalone terms such as self, improving, captures, and errors, which occur in many requests unrelated to self-improving agent development. The description at line 4 also treats several of these terms as activation conditions, and no exclusions require an agent-development context.

Content

No source excerpt is available for this finding.

Common standalone keywords can activate the skill outside its intended self-improving-agent context.

Medium
Category
Trigger Abuse
Confidence
99% confidence
Finding

The trigger list includes broad words such as self, improving, captures, and errors, which commonly occur in unrelated requests. Neither this list nor the activation description on line 4 requires these terms to appear within a self-improving-agent task, and no exclusions are provided.

Content

No source excerpt is available for this finding.

Activation keywords include overly broad terms such as `self`, `improving`, `errors`, and `corrections`.

Low
Category
Trigger Abuse
Confidence
97% confidence
Finding

These standalone keywords overlap with ordinary conversation and unrelated troubleshooting requests. The usage section provides no exclusion conditions or requirement that these keywords occur within a self-improving-agent context, so they could cause unintended invocation.

Content

No source excerpt is available for this finding.

Implicit invocation is enabled, but the description does not define a clear activation scope.

Low
Category
Trigger Abuse
Confidence
93% confidence
Finding

The description and default prompt discuss demand for self-improving agents and generic help fixing bugs rather than specifying when this skill should activate. With implicit invocation enabled, there is no clear boundary distinguishing relevant self-improvement tasks from ordinary debugging requests.

Content

No source excerpt is available for this finding.

Usage keywords include broad everyday terms that can cause unrelated skill invocations.

Low
Category
Trigger Abuse
Confidence
93% confidence
Finding

The usage signals include standalone terms such as "self", "captures", "errors", and "corrections", which occur in many requests unrelated to self-improving agent workflows. No contextual matching constraints or exclusions clarify when these broad keywords should activate the skill.

Content

No source excerpt is available for this finding.