Back to skill

Security audit

QSR Shift Reflection

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed restaurant shift-handoff memory workflow with scoped persistence and no evidence of hidden execution, exfiltration, or destructive behavior.

Before installing, confirm that your OpenClaw host and qsr-store-memory-engine enforce the store namespace, authorization, audit, retention, and deletion controls you expect. Treat exported reflections and issue histories as confidential operational data, especially because they may include staffing, customer incident, maintenance, and food-safety context.

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
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
85% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · SKILL.md (reported line 45)May include surrounding context.

md
Every record is tagged with a single store identifier and lives inside that store's namespace. Records never cross store boundaries. In multi-location deployments, each store has its own isolated reflection archive and open issue list. Cross-store rollups (see `ADAPTING THIS SKILL → Multi-location operations`) are produced by reading each store's namespace independently and combining the results at report time, not by merging the underlying records.

### Sibling skill access

Other skills in the QSR Operations Suite may read from this skill's records *only* through the same store-scoped namespace and *only* in read-only mode. The integrations listed under `CONNECTING TO OTHER SKILLS` are read paths, not write paths. No sibling skill modifies, deletes, or re-exports reflection or open issue records.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The architecture section states that Function 4 'Runs continuously in the background,' which implies autonomous scheduled behavior. Later normative sections explicitly say the skill is not a daemon, does not run on a schedule, and that Function 4 is only operator-triggered or computed at read time, so these statements directly conflict.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The markdown describes export capabilities for reflections and issue records, which can expose confidential store operational data, but it does not include an explicit warning to users that exports may contain sensitive internal information and should be handled carefully. Under the markdown-file warning criterion, behaviors affecting data privacy should be accompanied by a clear caution.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The skill says operators can invoke actions with phrases like "Show open issues" and similar natural-language commands, but it does not define whether these are strict commands, examples, or merely descriptive utterances. Because the command list lacks exclusion conditions or a constrained invocation context, ordinary discussion about open issues could unintentionally trigger command behavior in hosts that auto-route by text.

Content

No source excerpt is available for this finding.