Back to skill

Security audit

WS QuickStatus

Security checks for vulnerabilities and agentic risk

Overview

This skill is a simple workspace status checklist that only lists local status information, with a limited privacy note around memory filenames.

Installers should know that running this skill may print top-level filenames, git status, deliverable filenames, and up to three recent `memory/*.md` filenames into the conversation or logs. Avoid using it in shared or untrusted contexts if those filenames could reveal private project or client information.

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
Findings (1)

other

Warning
Location
SKILL.md:15
Finding

Agent Memory Metadata Reconnaissance

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 15-24
Vulnerability Type: other: Agent memory metadata reconnaissance
Risk Level: Medium

Vulnerable Code Snippet:

markdown
3. Shows recent memory notes (last 3 files).

## Basic steps

```bash
cd <workspace>
echo "== top level ==" && ls -la
echo "== git ==" && git status -s 2>/dev/null || echo "no git repo"
echo "== recent memory ==" && ls -t memory/*.md 2>/dev/null | head -3

Technical Analysis

The Skill instructs the agent to enumerate the three most recently modified Markdown files in the workspace's memory/ directory. Although it does not read file contents, it discloses filenames and their recency ordering through standard output.

Memory filenames can contain sensitive metadata such as customer names, project identifiers, incident references, task descriptions, or user information. Enumerating this data is not required to perform the Skill's Git-status and deliverable-health checks. The behavior therefore introduces unnecessary access to agent memory metadata and expands the information exposed to the caller.

Attack Path

  1. An agent loads and follows the WS QuickStatus instructions.
  2. The agent changes into a workspace containing Markdown files under memory/.
  3. The command sorts those files by modification time and selects the latest three.
  4. Their paths or filenames are printed to standard output.
  5. A caller able to view the Skill output learns sensitive memory-note metadata and can use it to identify potentially valuable files for subsequent targeted access.

This path exposes metadata only; the audited command does not read memory-file contents or transmit results to an external service.

Impact Assessment

The issue can disclose the names and relative recency of up to three agent memory files available to the current process. It does not grant additional operati ...[truncated 294 chars]

Remediation
View remediation

Remediation Suggestions

  • Remove memory-file enumeration from the default workspace-health workflow.
  • If memory inspection is genuinely required, make it an explicit opt-in operation and obtain user approval before accessing the directory.
  • Return only a count, such as the number of recent notes, rather than exposing filenames.
  • If identifiers must be displayed, redact or hash filenames and avoid exposing modification-order information.
  • Restrict the check to a dedicated, non-sensitive status index instead of scanning the agent's general memory directory.
  • Clearly document the information that will be accessed and displayed before executing the check.
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 (2)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill explicitly lists recent memory note files to stdout, which can expose sensitive filenames and recency metadata such as project names, client identifiers, incident topics, or other confidential labels. Even without printing file contents, metadata disclosure can leak enough context to reveal sensitive workstreams, and the 'one pass' status-check framing makes accidental disclosure more likely in shared logs or transcripts.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The description says to use the skill when the user wants a 'fast overview of the current workspace state' and to 'run it' before or after a task, but it does not define concrete trigger phrases, boundaries, or negative examples. This makes activation scope somewhat ambiguous for a markdown skill description and could overlap with many general requests to inspect or review work.

Content

No source excerpt is available for this finding.