Back to skill

Security audit

Auto Improving Agent

Security checks for vulnerabilities and agentic risk

Overview

The skill is openly about persistent self-improvement notes, but it allows automatic repository writes, heuristic deletion, and promotion toward agent-behavior files without clear approval gates.

Install only if you want an agent to maintain persistent project learning notes. Use `--dry-run` for retention scoring first, review all proposed writes, and require explicit confirmation before deleting entries or promoting anything into `SOUL.md`, `AGENTS.md`, or `TOOLS.md`. Treat user messages and memory-search results as untrusted until reviewed.

Vulnerability Patterns
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • 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)

T02 · Agent Memory Poisoning

Error
Location
SKILL.md:71
Finding

Untrusted Content Can Be Persisted and Promoted into Long-Term Agent Behavior

Content
View full analysis
90 days old for retention scoring. 3. **Promotion detector**: After logging, scan for entries with the same `[Category]` tag appearing 3+ times. If found, auto-suggest a one-liner promotion to: - behavior/style → `SOUL.md` - workflow/process → `AGENTS.md` - tool/env gotcha → `TOOLS.md` 4. **Cross-session pattern detection**: When `memory_search` returns a daily note describing a workaround, check if `.learnings/` already has it. If not and it passes the write gate, log it. ``` `hooks/openclaw/handler.js:8-25, 46-52`: ```javascript const REMINDER_CONTENT = ` ## Self-Improvement Reminder After completing tasks, evaluate if any learnings should be captured: **Log when:** - User corrects you → \`.learnings/LEARNINGS.md\` - Command/operation fails → \`.learnings/ERRORS.md\` - User wants missing capability → \`.learnings/FEATURE_REQUESTS.md\` - You discover your knowledge was wrong → \`.learnings/LEARNINGS.md\` - You find a better approach → \`.learnings/LEARNINGS.md\` **Promote when pattern is proven:** - Behavioral patterns → \`SOUL.md\` - Workflow improvements → \`AGENTS.md\` - Tool gotchas → \`TOOLS.md\` Keep entries simple: date, title, what happened, what to do differently. `.trim(); // Inject the reminder as a virtual bootstrap file // Check that bootstrapFiles is an array before pushing if (Array.isArray(event.contex ...[truncated 3206 chars]
Remediation
View remediation
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
Findings (8)

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The declared purpose emphasizes learning capture and promotion, but the documented retention policy also allows deletion of entries, which is a materially different and more destructive behavior. Omitting destructive behavior from the high-level contract can mislead users into enabling a skill that silently removes historical data.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared purpose emphasizes learning capture and promotion, but the documented retention policy also allows deletion of entries, which is a materially different and more destructive behavior. Omitting destructive behavior from the high-level contract can mislead users into enabling a skill that silently removes historical data.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared purpose emphasizes learning capture and promotion, but the documented retention policy also allows deletion of entries, which is a materially different and more destructive behavior. Omitting destructive behavior from the high-level contract can mislead users into enabling a skill that silently removes historical data.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill description markets automatic capture and promotion but does not prominently warn that it may write, archive, or delete entries in .learnings/ files. Missing user-facing disclosure around persistent and destructive actions undermines informed consent and can cause loss of data or silent repository modifications in normal usage.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The retention policy explicitly permits deletion of stored entries based on heuristic scores, yet there is no explicit user-facing warning or approval gate for permanent removal. Heuristic deletion is dangerous because scoring mistakes or stale context can erase useful operational history and make the loss hard to detect later.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
84% confidence
Finding

The skill describes behavior that reads and writes project files, including .learnings/ and documentation targets, but does not declare an explicit tool scope or permissions boundary. That makes the write capability implicit and harder for users or a runtime policy layer to review, increasing the chance of unintended persistence or modification of repository files.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill is designed to fire automatically based on broad conditions like corrections, failures, and discoveries, which are common in ordinary sessions. Ambiguous activation criteria can cause the skill to run unexpectedly, leading to unreviewed persistence, context drift, or writes to repo files without a clear user action.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The 'Automated triggers' section authorizes unprompted actions such as post-task scans, session-start sweeps, promotion suggestions, and cross-session pattern handling. Because these triggers are not tightly bounded and can operate across sessions, they increase the risk of surprising behavior, privacy leakage from prior notes, and unauthorized file modifications.

Content

No source excerpt is available for this finding.