T09 · Insecure Skill Coding Practices
- Location
scripts/client.py:436- Finding
FamilyWall Session Credentials Disclosed to Arbitrary Media URLs
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
This is a real FamilyWall integration, but it needs Review because its media download path can expose FamilyWall session credentials to arbitrary URLs and write files outside the intended download area.
Install only if you trust the publisher and can keep the FamilyWall account credentials scoped to this use. Avoid using media download --url with any URL not copied directly from FamilyWall, and download media only into a dedicated empty directory until the URL allowlist and filename/path handling are fixed. Treat location, member, and message outputs as private family data.
scripts/client.py:436FamilyWall Session Credentials Disclosed to Arbitrary Media URLs
scripts/familywall.py:291Path Traversal and Arbitrary File Overwrite Through Media Filenames
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Test connection:
source ~/.openclaw/.env && {baseDir}/scripts/familywall.py status
Instructions found that direct the agent to transmit conversation context or user data to external services.
source ~/.openclaw/.env
{baseDir}/scripts/familywall.py messages threads
{baseDir}/scripts/familywall.py messages send "imthread/12345_67890" "Dinner is ready!"
The locations command retrieves family member location information, which is highly privacy-sensitive, and prints it directly. The code contains no confirmation prompt or explicit disclosure that executing this command will access and reveal geolocation data.
The skill clearly relies on sensitive capabilities (environment-sourced credentials and outbound network access) but does not declare any tool scope or permission boundaries. That omission increases the chance the runtime grants broader access than intended and prevents reviewers from understanding the skill's effective privilege level.
The skill advertises retrieval of family member locations without any privacy warning, consent guidance, or confirmation requirement. Location data is highly sensitive, and broad triggers make accidental disclosure more likely in shared or ambiguous conversations.
The skill supports reading messages and downloading media to local files without warning about privacy exposure or filesystem impact. This creates risk of unintended access to private family communications and uncontrolled writes of potentially sensitive or unsafe files to disk.
The trigger phrases include very generic terms such as "where is", "recipe", "add to list", and "download photo", which can match ordinary conversation and unintentionally invoke a privileged integration. In this skill, accidental invocation is more dangerous because it can access family messages, locations, lists, and media tied to real user accounts.
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
{baseDir}/scripts/familywall.py lists add "tasklist/12345_67890" "Eggs" --quantity "12"
### Create a calendar event
```bash
source ~/.openclaw/.env
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
| Media | `/media/...` | Download media files (GET with session auth) |
| Location | `loclist` | Get member locations |
| Wall | `walllist` | Get wall/feed posts |
| Meals | `mplistinterval` | List planned meals in range |
| Meals | `mpcreate` | Add dish to meal plan |
| Meals | `mpmealput` | Create/edit a meal |
| Recipes | `mprecipeput` | Create/edit a recipe |
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
class FamilyWallClient:
"""Client for interacting with the FamilyWall API."""
BASE_URL = "https://api.familywall.com/api"
PARTNER_SCOPE = "Family"
def __init__(self, timezone=None):
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
class FamilyWallClient:
"""Client for interacting with the FamilyWall API."""
BASE_URL = "https://api.familywall.com/api"
PARTNER_SCOPE = "Family"
def __init__(self, timezone=None):
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
class FamilyWallClient:
"""Client for interacting with the FamilyWall API."""
BASE_URL = "https://api.familywall.com/api"
PARTNER_SCOPE = "Family"
def __init__(self, timezone=None):
The delete_event method performs a remote deletion operation via the FamilyWall API, but the code provides no confirmation prompt, logging, or explicit warning to the user before or during the action. Because this is an irreversible state-changing operation and this file contains no user-facing disclosure around it, the deletion lacks the safety warning required for code files.
download_media writes remote content to an arbitrary caller-supplied output_path with no path validation, filename restrictions, or safe-directory enforcement. In an agent context, an attacker could influence the destination path and cause overwrite or placement of untrusted files in sensitive locations, making this substantially more dangerous than a normal SDK helper.
The members command fetches and outputs personally identifiable information including email addresses, phone numbers, account IDs, and last login data. Although the code prints the resulting JSON, there is no disclosure in the command help, comments, or docstrings warning that sensitive family data will be exposed to the terminal.
The media download command accepts an arbitrary user-supplied URL and passes it to the client downloader, allowing the skill to fetch resources not obtained from FamilyWall message metadata. In an agent context, this broadens the skill from 'download FamilyWall attachments' into a generic network fetch primitive, which can enable SSRF-style access to internal services or unintended exfiltration/retrieval depending on where the client runs.