Security Audit
context-manager
github.com/sickn33/antigravity-awesome-skillsTrust Assessment
context-manager received a trust score of 85/100, placing it in the Mostly Trusted category. This skill has passed most security checks with only minor considerations noted.
SkillShield's automated analysis identified 1 finding: 0 critical, 1 high, 0 medium, and 0 low severity. Key findings include Untrusted instruction to open local file.
The analysis covered 4 layers: Manifest Analysis, Static Code Analysis, Dependency Graph, LLM Behavioral Safety. All layers scored 70 or above, reflecting consistent security practices.
Last analyzed on February 20, 2026 (commit e36d6fd3). SkillShield performs automated 4-layer security analysis on AI skills and MCP servers.
Layer Breakdown
Behavioral Risk Signals
Security Findings1
| Severity | Finding | Layer | Location | |
|---|---|---|---|---|
| HIGH | Untrusted instruction to open local file The skill's instructions, provided as untrusted input, contain a directive for the host LLM to 'open' a local file (`resources/implementation-playbook.md`). This is a prompt injection attempt, instructing the LLM to perform an action. If the LLM has file system access capabilities, this could lead to unauthorized file access, potentially allowing the skill to read arbitrary files on the system, leading to data exfiltration or bypassing intended access controls. The specified file is not part of the provided skill package context. Remove or rephrase the instruction to 'open' a file. Instead, the skill should describe the content of the playbook or instruct the user to manually consult it. If the playbook is intended to be part of the skill, it should be included in the supporting files and accessed via a defined, secure mechanism, not a direct 'open' instruction from untrusted input. | LLM | SKILL.md:18 |
Scan History
Embed Code
[](https://skillshield.io/report/3bf3aeec156f6e8d)
Powered by SkillShield