Security Audit
garrytan/gstack:setup-browser-cookies
github.com/garrytan/gstackTrust Assessment
garrytan/gstack:setup-browser-cookies received a trust score of 0/100, placing it in the Untrusted category. This skill has significant security findings that require attention before use in production.
SkillShield's automated analysis identified 62 findings: 28 critical, 31 high, 3 medium, and 0 low severity. Key findings include File read + network send exfiltration, Dangerous tool allowed: Bash, Sensitive environment variable access: $HOME.
The analysis covered 4 layers: Manifest Analysis, Static Code Analysis, Dependency Graph, LLM Behavioral Safety. The Manifest Analysis layer scored lowest at 0/100, indicating areas for improvement.
Last analyzed on April 9, 2026 (commit dbd7aee5). SkillShield performs automated 4-layer security analysis on AI skills and MCP servers.
Layer Breakdown
Behavioral Risk Signals
Security Findings62
| Severity | Finding | Layer | Location | |
|---|---|---|---|---|
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:7 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:13 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:17 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:21 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:26 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:39 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:40 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:47 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:53 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:59 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:65 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:89 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:91 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:117 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:128 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:129 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:149 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:150 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:199 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:222 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:224 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:230 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:287 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:313 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:319 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:320 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:380 | |
| CRITICAL | File read + network send exfiltration AI agent config/credential file access Remove access to sensitive files not required by the skill's stated purpose. SSH keys, cloud credentials, and browser data should never be read by skills unless explicitly part of their declared functionality. | Manifest | setup-browser-cookies/SKILL.md:437 | |
| HIGH | Dangerous tool allowed: Bash The skill allows the 'Bash' tool without constraints. This grants arbitrary command execution. Remove unconstrained shell/exec tools from allowed-tools, or add specific command constraints. | Static | setup-browser-cookies/SKILL.md:1 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:7 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:13 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:17 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:21 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:26 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:39 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:40 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:47 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:53 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:59 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:65 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:89 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:91 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:117 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:128 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:129 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:149 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:150 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:199 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:222 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:224 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:230 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:287 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:313 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:319 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:320 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:380 | |
| HIGH | Sensitive path access: AI agent config Access to AI agent config path detected: '~/.claude/'. This may indicate credential theft. Verify that access to this sensitive path is justified and declared. | Static | setup-browser-cookies/SKILL.md:437 | |
| HIGH | Use of `eval` and `source` with external binary output The skill uses `eval "$(~/.claude/skills/gstack/bin/gstack-slug 2>/dev/null)"` and `source <(~/.claude/skills/gstack/bin/gstack-repo-mode 2>/dev/null)`. Both `eval` and `source` execute arbitrary code. While the binaries (`gstack-slug`, `gstack-repo-mode`) are located within the skill's own trusted directory, a compromise of these binaries (e.g., via a supply chain attack on the `gstack` skill package itself, or local tampering) would allow an attacker to execute arbitrary commands on the host system. Additionally, the `$SLUG` variable derived from `gstack-slug` is used in file paths (e.g., `_LEARN_FILE="${GSTACK_HOME:-$HOME/.gstack}/projects/${SLUG:-unknown}/learnings.jsonl"`), which could lead to path traversal or other file system vulnerabilities if `$SLUG` were to contain malicious characters. Avoid `eval` and `source` with external command output where possible. If dynamic execution is strictly necessary, ensure the output is rigorously sanitized or parsed in a safer manner. For `$SLUG`, validate its content to prevent path traversal. Implement strong integrity checks for `gstack` binaries. | LLM | SKILL.md:30 | |
| HIGH | Direct handling and decryption of sensitive user cookies The skill's core purpose is to 'Import cookies from your real Chromium browser' and explicitly states 'Cookies are decrypted and loaded into the Playwright session.' This involves handling highly sensitive user session cookies, which are effectively credentials. The skill has `Bash` permissions, allowing it to execute arbitrary commands. While the stated intent is to load these into a headless browser, there is no explicit mechanism described within the skill's documentation to prevent the exfiltration or misuse of these decrypted cookies by the skill itself or by a compromised `cookie-import-browser` tool. The security of user sessions relies entirely on the integrity and secure implementation of the `cookie-import-browser` tool and the skill's subsequent handling, which is not transparently detailed. Provide clear documentation on how decrypted cookies are handled, stored, and secured to prevent unauthorized access or exfiltration. Detail the security measures implemented in `cookie-import-browser` and the Playwright session. Consider limiting the scope of `Bash` commands or using a more confined environment for cookie handling if possible. | LLM | SKILL.md:250 | |
| MEDIUM | Sensitive environment variable access: $HOME Access to sensitive environment variable '$HOME' detected in shell context. Verify this environment variable access is necessary and the value is not exfiltrated. | Static | setup-browser-cookies/SKILL.md:48 | |
| MEDIUM | Telemetry prompt contradicts logged data The skill's preamble logs the current Git repository name to `~/.gstack/analytics/skill-usage.jsonl`. However, the user-facing telemetry prompt explicitly states: 'No code, file paths, or repo names are ever sent.' This creates a contradiction. Even if the remote telemetry binary (`gstack-telemetry-log`) filters out the repository name, the local log file contains this potentially sensitive information, which goes against the user's expectation set by the prompt. Either remove the logging of the repository name from `skill-usage.jsonl` or update the telemetry prompt to accurately reflect that the repository name is logged locally (but not sent remotely, if that is the case). Ensure consistency between user promises and actual data handling. | LLM | SKILL.md:40 | |
| MEDIUM | Broad `Bash` permission combined with sensitive data handling The skill declares `Bash` as an allowed tool, granting it the ability to execute arbitrary commands on the host system. While many operations within the skill (e.g., `git` commands, file system operations, running `gstack` binaries) require `Bash`, this broad permission becomes a significant risk when combined with the skill's core functionality of handling and decrypting sensitive user session cookies. A malicious or compromised skill could leverage this `Bash` permission to exfiltrate the decrypted cookies or perform other harmful actions. Review if all operations truly require unrestricted `Bash` access. If possible, consider more granular permissions or sandboxing mechanisms for parts of the skill that handle sensitive data. Document the necessity of `Bash` permission in the context of cookie handling. | LLM | SKILL.md:1 |
Scan History
Embed Code
[](https://skillshield.io/report/89c1145e9c3106a5)
Powered by SkillShield