Back to skill

Security audit

Press Release

Security checks for vulnerabilities and agentic risk

Overview

The press-release generator is mostly local template tooling, but one included utility silently stores user input and command history on disk.

Review this before installing if you may enter confidential announcements, financing details, incident-response drafts, contact data, or secrets. The template generator itself is local, but the bundled press-release utility can retain user-supplied text and command arguments under the user's data directory in plaintext. The output is also primarily Chinese, so users expecting English-only behavior should check the scripts first.

Vulnerability Patterns
  • 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
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/script.sh:5
Finding

Undisclosed Persistent Storage of Potentially Sensitive User Input

Content
View full analysis
> "$DATA_DIR/history.log"; } ``` ```bash cmd_add() { echo "$(date +%Y-%m-%d) $*" >> "$DB"; echo " Added: $*" _log "add" "${1:-}" } ``` ### Technical Analysis The script creates a persistent data directory under the user's home directory whenever it starts. The `add` command writes all supplied content to `data.log`, while `_log` records command names and their first arguments in `history.log`. This storage behavior is not disclosed in `SKILL.md` and is unrelated to the documented press-release generation workflow. Press-release drafts and command arguments can contain confidential product, financing, event, crisis-response, or contact information. Such content is retained in plaintext without an explicit user retention decision, redaction, encryption, restrictive permission setup, or documented cleanup policy. The script relies on the process's existing `umask` and does not explicitly restrict directory and file permissions. The exposure therefore depends on the host's filesystem permissions and configuration. No network transmission or privilege escalation was identified. ### Attack Path 1. A user invokes the packaged `press-release` utility. 2. The script creates the persistent directory defined by `PRESS_RELEASE_DIR`, `XDG_DATA_HOME`, or the default path `$HOME/.local/share/press-release`. 3. The user supplies confidential content through the `add` command or places sensitive information in the first argument to a logged command. 4. The content is appended in plaintext to `data.log` and/or `history.log`. 5. A local pr ...[truncated 914 chars]
Remediation
View remediation
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (6)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

All generated templates, help text, usage errors, and notices are presented exclusively in Chinese throughout the embedded script. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale constraint is explicitly justified, which is not present here.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The top-level comment and CLI help present this as 'press-release' tooling, implying functionality related to managing or publishing press releases. However, the actual commands merely add, list, search, and export arbitrary entries in a local data file, which contradicts the apparent documented intent rather than implementing press-release-specific behavior.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The script persistently logs user-supplied command arguments to a history file under the user's data directory without notice or consent. This can capture sensitive values a user may pass as arguments, such as names, identifiers, search terms, or accidental secrets, creating a local privacy and confidentiality risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The add command stores arbitrary user-provided content in a persistent local database file without explicitly informing the user that the data will be retained on disk. While this appears to be core functionality rather than overtly malicious behavior, undisclosed persistence can expose sensitive content entered by users and may violate user expectations in an agent skill context.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The document presents the press release tips entirely in Chinese and labels the skill as '新闻稿', which implies a fixed language/locale expectation. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale constraint is explicitly documented and justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The manifest description presents the skill in both Chinese and English, but the document does not state how the language will be chosen for user-facing output or offer the user a preference. Under the policy rule, forcing or assuming a language/locale without opt-in can be a natural-language policy concern.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.