Back to skill

Security audit

Email Swipe

Security checks for vulnerabilities and agentic risk

Overview

This email-training skill is mostly upfront about its purpose, but it needs review because its local web server and command helpers can expose or alter private mail-related data.

Install only if you are comfortable reviewing and hardening it first. Use demo mode or trusted local-only operation for testing; before real mail, bind the UI to localhost or add authentication, do not expose it to public networks, and avoid untrusted EMAIL_SWIPE_EMAIL, EMAIL_ACCOUNT, GOG_PATH, EMAIL_SWIPE_GOG_PATH, or config.json values until the shell command construction is fixed.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Findings (3)

T05 · Unauthorized Access and Privilege Escalation

Error
Location
scripts/serve-ui.py:57
Finding

Unauthenticated LAN API Exposes Email Data and Permits Persistent State Modification

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/fetch-emails.py:14
Finding

Shell Command Injection Through Configurable Email Account and Executable Path

Content
View full analysis
&1") ``` The interpolated values can originate from configuration or environment variables: ```python if os.environ.get('EMAIL_SWIPE_EMAIL'): cfg['email'] = os.environ['EMAIL_SWIPE_EMAIL'] if os.environ.get('EMAIL_SWIPE_GOG_PATH') or os.environ.get('GOG_PATH'): cfg['gogPath'] = os.environ.get('EMAIL_SWIPE_ ...[truncated 2552 chars]
Remediation
View remediation
&1`; captured stderr is already available through `subprocess`. 4. Validate account identifiers using a conservative email-address parser and reject control characters or shell metacharacters. 5. Resolve `gog_path` to an existing regular executable file and reject unexpected command fragments. 6. Constrain `limit` to a reasonable positive range. 7. Avoid loading executable paths from project-controlled UI configuration where possible. 8. Add regression tests using values containing spaces, semicolons, command substitutions, pipes, redirects, and newline characters. ]]>

T08 · Insecure Dependencies

Note
Location
scripts/fetch-emails-html.py:22
Finding

Unpinned Third-Party Python Dependency Installation Guidance

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • 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 (115)

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · .gitignore (reported line 2)May include surrounding context.

text
# OAuth secrets - never commit
client_secret.json
**/client_secret.json

# OS

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · .gitignore (reported line 3)May include surrounding context.

text
# OAuth secrets - never commit
client_secret.json
**/client_secret.json

# OS

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · AGENTS.md (reported line 150)May include surrounding context.

md
- When fetching training mail, prefer full sanitized HTML/body content; subject+snippet alone is a fallback, not the target.
- **Always guess** on every injected email (`predictedAction` + `predictionConfidence`) — even bootstrap, even low confidence. Blank guesses hide the score.
- Advanced folder routes persist in `settings.json` — don't re-ask each session; build out per `platform-rules.folderRoutes`.
- Use brief + training-pack at runtime — not full swipe history in prompts.

## CLI cheat sheet

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · README.md (reported line 166)May include surrounding context.

md
- When fetching training mail, prefer full sanitized HTML/body content; subject+snippet alone is a fallback, not the target.
- **Always guess** on every injected email (`predictedAction` + `predictionConfidence`) — even bootstrap, even low confidence. Blank guesses hide the score.
- Advanced folder routes persist in `settings.json` — don't re-ask each session; build out per `platform-rules.folderRoutes`.
- Use brief + training-pack at runtime — not full swipe history in prompts.

## CLI cheat sheet

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · SKILL.md (reported line 195)May include surrounding context.

md
- When fetching training mail, prefer full sanitized HTML/body content; subject+snippet alone is a fallback, not the target.
- **Always guess** on every injected email (`predictedAction` + `predictionConfidence`) — even bootstrap, even low confidence. Blank guesses hide the score.
- Advanced folder routes persist in `settings.json` — don't re-ask each session; build out per `platform-rules.folderRoutes`.
- Use brief + training-pack at runtime — not full swipe history in prompts.

## CLI cheat sheet

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · scripts/session-intake.py (reported line 184)May include surrounding context.

python
- When fetching training mail, prefer full sanitized HTML/body content; subject+snippet alone is a fallback, not the target.
- **Always guess** on every injected email (`predictedAction` + `predictionConfidence`) — even bootstrap, even low confidence. Blank guesses hide the score.
- Advanced folder routes persist in `settings.json` — don't re-ask each session; build out per `platform-rules.folderRoutes`.
- Use brief + training-pack at runtime — not full swipe history in prompts.

## CLI cheat sheet

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The stated purpose understates the skill's real capabilities: email access detection, account management, Gmail label reading, folder-based learning, settings synchronization, and background compilation. Because the target domain is email, that understatement materially increases privacy and integrity risk.

Content

No source excerpt is available for this finding.