Back to skill

Security audit

Agent Memory

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed persistent memory skill, with privacy-sensitive behavior that fits its purpose but should be installed only with awareness of retention and auto-injection behavior.

Before installing, confirm the missing implementation files and timer unit come from a trusted source, review what the memory plugin stores, and make sure you have commands or procedures to inspect, disable, and delete stored memories. Avoid storing secrets or sensitive personal data unless you are comfortable with cross-session retention and prompt recall.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • 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 (5)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The manifest description advertises "budgeted German-aware query retrieval," which introduces a language-specific behavior in the skill's core description. The file does not clearly present this as user-selectable or region-justified at that point, so it may violate language/locale neutrality expectations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

This skill persistently stores conversation-derived memory across sessions, including facts, snippets, lessons, entities, and relations, but the markdown introduction does not present a prominent user-facing warning or consent notice. Users may unknowingly have sensitive conversational data retained and later resurfaced via recall or plugin injection, creating privacy, retention, and inadvertent disclosure risk.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · SKILL.md (reported line 22)May include surrounding context.

md
- Authority Lanes (identity / preference / evidence / authorization / procedural)
- Rebound-Protection after idle phases (signalfoundry / Moltbook pattern)
- Class-specific TTL and forget_stale()
- Finer source trust: five sources with per-lane write policy (`tool`/`external` quarantined to `evidence`; `identity`/`authorization` protected)
- Raw recall snippets kept separate from semantic facts
- Auto-injection plugin with per-lane budgets and German-aware, score-ranked query retrieval (token-prefix FTS + synonyms, deterministic, no embeddings)
- Conflict detection on single-valued lanes (`identity`, `authorization`) with explicit resolution; open conflicts auto-reconcile when a referenced fact becomes inactive

Session Persistence

Medium
Category
Rogue Agent
Confidence
88% confidence
Finding

The skill is explicitly designed for persistent memory across restarts and includes a systemd user timer to maintain retained state, which increases the duration and operationalization of stored data. Persistent storage is not inherently malicious, but in a memory skill it materially raises privacy and data-retention risk if sensitive session content is captured without strong consent, minimization, and deletion controls.

Content

Scanner excerpt · SKILL.md (reported line 77)May include surrounding context.

bash
# Runs forget_stale() daily
systemctl --user enable --now hermes-memory-cleanup.timer

5. Verify

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 103)May include surrounding context.

md
Behavioral rules ("how to respond"), kept separate from facts ("what is true")
in a `procedural_rules` table. Observation-only writes via `propose_rule()`;
every rule starts `pending` and only `approve_rule()` (a mandatory human gate, no
auto-approve) makes it active. Deterministic, stdlib-only conflict detection on
approval hard-blocks direct contradictions and soft-blocks interactions /
artifact bloat / budget overflow (override with `ack_interactions`). The plugin
injects only active, trigger-matching rules in a sanitized, budgeted