Ultimate Research

• Outcome: Review
• Latest audit: Sep 16, 2026 · 8:15 AM UTC
• Version: 1.0.2

Overview

The skill is a disclosed research-orchestration helper, but it mandates unscoped persistent-memory use for broad and ambiguous queries, so users should review it before installing.

Install only if you are comfortable with a research skill that routinely pulls prior memory context. For safer use, narrow when it activates and make supermemory conditional on explicit user need, with project-level scope and no forwarding of unrelated memory-derived details.

A.I.G

Vulnerability Patterns

• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs
• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally

Findings (1)

T05 · Unauthorized Access and Privilege Escalation

Warning

• Location: SKILL.md:16

Finding

Mandatory Overbroad Access to Persistent Agent Memory

Content

View full analysis

Vulnerability Details

File Location: SKILL.md:16, SKILL.md:34-35, and references/routing.md:5
Vulnerability Type: Excessive access to persistent memory in violation of least-privilege principles
Risk Level: Medium

Vulnerable Code Snippets

SKILL.md:16:

markdown

- Always use the mandatory core: `web-scraper`, `self-improvement`, `supermemory`, `brainstorming`, `zo-research-topic`, and `market-research`.

SKILL.md:34-35:

markdown

2. **Load context**
   - Use `supermemory` for prior decisions, preferences, projects, and recurring context.

references/routing.md:5:

markdown

1. `supermemory` — recover prior decisions, preferences, and related work.

Technical Analysis

The Skill requires supermemory to be invoked for every query, regardless of whether the current task requires historical user context. It directs the Agent to retrieve prior decisions, preferences, projects, recurring context, and related work.

This mandatory retrieval violates the principle of least privilege because a simple or unrelated research request does not necessarily justify access to persistent memory. Retrieved information may also enter the broader research workflow, which invokes several other skills.

The implementation of supermemory is not included in this project, so the exact storage permissions, retrieval scope, filtering behavior, and downstream disclosure controls cannot be verified. The finding therefore concerns the unconditional and overbroad access requirement in this Skill, rather than a demonstrated defect in the external memory implementation.

Attack Path

1. A user or untrusted workflow submits any research query, including one that does not require historical context.
2. The Skill loads and applies its mandatory routing instructions.
3. The Agent invokes supermemory because it is required for every query.
4. The memory skill retrieves prior decisions, preferences, projects, recurring context, or rela ...[truncated 1265 chars]

Remediation

View remediation

Remediation Suggestions

1. Remove supermemory from the mandatory core and invoke it only when historical context is materially necessary for the current request.
2. Require explicit user consent before retrieving persistent context, particularly for sensitive or cross-project research.
3. Define narrow retrieval scopes, such as a named project, a specified date range, or an explicit category of preferences.
4. Exclude credentials, secrets, private communications, and unrelated project data from retrieval.
5. Minimize retrieved content and pass only the specific facts required for the current task to downstream skills.
6. Add a policy that memory-derived data must not be forwarded to web scraping, market research, or other external capabilities unless the user explicitly authorizes it.
7. Record which memory categories were accessed and disclose that access in the response when appropriate.
8. Update the instructions to use conditional language, for example:

markdown

- Use `supermemory` only when the user requests continuity with prior work or when historical context is essential.
- Ask for consent and scope retrieval to the relevant project or preference category.
- Do not retrieve or forward secrets, credentials, or unrelated historical context.

9. Apply the same conditional-access rule consistently in both SKILL.md and references/routing.md.

SkillSpector

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Vague Triggers

Medium

• Category: Not specified by scanner
• Confidence: 96% confidence

Finding

The skill’s activation criteria are overly broad, explicitly covering 'ambiguous query' and general structured synthesis use cases. That creates unnecessary routing into a powerful orchestration skill that mandates multiple other skills, increasing attack surface, tool use, and the chance of inappropriate data access or unsafe downstream actions for requests that should stay narrow or require explicit user confirmation.

Content

No source excerpt is available for this finding.