Security audit

Agent Daily Review

Security checks for vulnerabilities and agentic risk

Overview

This skill reads local daily notes, creates a daily review report, and can append a summary to local memory files as its documented purpose describes.

Install only if you are comfortable with the skill reading local workspace notes and writing derived review files. Use --no-memory when you want a report without changing MEMORY.md, verify the workspace path before running, and avoid the cron example until you are comfortable with automatic daily file reads and writes.

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (4)

Description-Behavior Mismatch

Medium

Confidence: 94% confidence
Finding: The script’s advertised purpose is review/report generation, but it also persists derived data into long-term memory by appending to MEMORY.md. This creates an integrity and privacy risk because running the tool changes a durable knowledge store without strong disclosure, consent, or scoping, and could permanently retain sensitive content inferred from the day’s notes.

Vague Triggers

Medium

Confidence: 83% confidence
Finding: The trigger phrase "Summarize today" is generic and can collide with ordinary user requests that do not intend to invoke this skill. Because this skill performs filesystem scanning and writes review artifacts, accidental invocation could cause unintended access to personal records and modification of workspace files.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The workflow states that the skill will automatically scan journal, memory, and workspace markdown files and later save reports and append to MEMORY.md, but it does not clearly warn users about this data access and modification. In a reflective journaling context, these files may contain sensitive personal or project information, so silent scanning and persistence increase privacy and integrity risk.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The script appends to MEMORY.md automatically unless the user knows to pass --no-memory, meaning a read/analyze workflow causes silent state mutation. This is dangerous because it can poison future agent context, retain sensitive summaries, and make later decisions depend on tool-generated content the user did not explicitly approve.

Static analysis

No suspicious patterns detected.