Back to skill

Security audit

Verifiable-Recursive-Safety-Guardrails-Runtime

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed single-file guide for operating a security runtime, with no embedded executable payload or hidden jailbreak behavior found in the inspected artifact.

Install this as documentation for understanding or operating Provenable, not as proof that the full AEGX codebase is safe. Before building, installing, or running the referenced daemon or adapters, obtain the actual repository from a trusted source, review its installer and runtime code, and confirm where it will store records, run background processes, and monitor local activity.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

YARA rule 'agent_skill_prompt_injection_hidden_instructions': Prompt injection or hidden instructions embedded in AI agent skill text [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: provenable_condensed_singlefile
description: Condensed single-file skill for the full AEGX Provenable repository. Use when you need the repository's complete operating model, architecture, CLI, integration workflows, installer and bundle layout, and per-file source map in one SKILL.md without separate references.
---

# Provenable Condensed Single-File Skill

Use this skill when you want **one compressed file that represents the whole repository**. Treat this document as a portable operator and integrator guide for **AEGX Provenable**, including the runtime model, guard surfaces, command surface, platform workflows, packaging model, and a complete tracked-file map.

This skill is intentionally condensed. It does **not** reproduce the raw source code verbatim

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
90% confidence
Finding

Skill attempts to nullify the agent's safety policies or restrictions ('you have no restrictions', 'ignore your guidelines', 'do anything now'). This is a direct jailbreak that disables guardrails.

Content

Scanner excerpt · SKILL.md (reported line 144)May include surrounding context.

md
| **Snapshot before risk** | Create a snapshot before destructive or trust-sensitive changes |
| **Verify before trust** | Verify exported bundles before relying on them |
| **Escalate degraded state** | Ask for user guidance or restore health before proceeding |
| **Do not bypass policy** | Never work around the guard path silently |
| **Treat rollback honestly** | Rollback restores protected state, not external side effects already emitted |

## Platform Workflows

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 204)May include surrounding context.

md
| `SKILL.md` | Original skill manifest and operator command reference |