Security audit

handoff

Security checks for vulnerabilities and agentic risk

Overview

This handoff skill clearly does what it says: it writes a project handoff note and a small memory pointer so future sessions can resume work.

Install this only if you want project-local handoff files and memory pointers. Before committing, syncing, or sharing the generated handoff, review it for secrets, private incident details, unreleased work, or internal reasoning that should not leave the project.

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill directs the agent to persist session state into repository files and project memory, but it does not require an explicit user warning or confirmation that potentially sensitive conversation content, file paths, errors, and reasoning may be written to disk. In a handoff workflow, users may invoke the command reflexively while discussing secrets, internal incident details, or unpublished code changes, causing unintended persistence and later disclosure through commits, sharing, or local access.

Missing User Warnings

Low

Confidence: 89% confidence
Finding: Although the notes mention git-ignore behavior, they understate the risk that handoff files can still capture sensitive in-progress reasoning and may be committed, synced, or shared if ignore rules are absent or bypassed. This creates a data leakage path because the skill encourages writing detailed state summaries whose sensitivity may exceed normal project documentation.

Static analysis

No suspicious patterns detected.