Back to skill

Security audit

meta-recursive-capability-discovery

Security checks for vulnerabilities and agentic risk

Overview

This skill is a small local capability-gap analysis tool with a contained learning log, but its documentation overstates the amount of autonomous self-improvement it actually implements.

Before installing, be aware that the skill is less capable than its marketing language suggests: it performs local capability-gap analysis and optional local learning logs, not a full autonomous super-agent loop. Review whether you are comfortable with learner.py creating learned_patterns.json when used.

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
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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 (6)

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The implementation narrowly performs recursive capability discovery and blind-spot analysis over provided JSON inputs. While this aligns somewhat with the phrase 'recursive-capability-discovery,' it does not implement the stronger declared claims: no agent orchestration, no persistent self-improvement loop, no self-modification, no reflective optimization cycle, and no substantive self-verification beyond deterministic assertions in a test routine. The primary behavior is capability-gap analysis, not a broader transcendence/meta-agent system as described.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The description presents a sophisticated autonomous meta-skill with recursive capability discovery, self-validation, reflection, orchestration, and ongoing self-improvement. The actual code chunk only implements a lightweight persistence layer for recording successes, failures, and notes in a local JSON file. While such logging could support a broader meta-evolution loop, this snippet does not itself perform the declared advanced behaviors. Therefore the declared purpose materially overstates and misrepresents what the code actually does.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
80% confidence
Finding

The skill advertises behaviors that imply stateful learning and reflection loops, and the static analyzer detected file-write capability, but the manifest declares no explicit tool scope or permissions boundary. In an agent ecosystem, undeclared write capability weakens reviewability and can enable persistence or modification of local state without operator awareness.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The natural-language content of the skill description and usage instructions is entirely in Chinese, and there is no indication that the user can opt into another language or locale. Under the policy, forcing a specific language without user choice is a natural-language policy violation unless the locale constraint is explicitly justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The entire report is presented in Chinese, including headings and operational descriptions, with no indication that language selection is optional or that the skill is explicitly region-specific. This creates a natural-language policy concern because it imposes a locale/language preference without user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
97% confidence
Finding

This Python file contains its primary docstring and operational comments entirely in Chinese, including usage context and behavior descriptions, with no indication that the language is optional or region-specific. Under the stated policy, forcing a specific language without user opt-in is a natural-language locale violation.

Content

No source excerpt is available for this finding.