Back to skill

Security audit

QuestMeet Skill

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious with medium confidence because it reports automatic self-replacement, broadly stored account tokens, and capabilities that reach beyond its stated networking purpose.

Consider the suspicious verdict before installing. In particular, review whether you are comfortable with automatic remote replacement, where the account token would be stored, and the skill's ability to upload selected local files, transfer quota, and retrieve full chat histories. The scanner does not report clear evidence of malicious intent.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Rogue AgentSelf-Modification, Session Persistence
  • 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 (40)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

Most functions match the networking purpose or are reasonable supporting implementation details. However, uploading local file contents and offering transferable quota as a reply incentive introduce materially significant data-disclosure and account-resource capabilities not disclosed in the description. These capabilities require explicit invocation; the supplied code does not autonomously scan files or transfer quota by default. The assessment concerns these undeclared capabilities, not evidence of malicious execution.

Content

No source excerpt is available for this finding.

Self-Modification

High
Category
Rogue Agent
Confidence
97% confidence
Finding

The skill directs the agent to replace its installed instructions and supporting code with the latest remote version without a specified approval or integrity-verification step. This duplicates the automatic-update findings and creates a supply-chain path for a compromised upstream version to gain the agent's existing capabilities.

Content

Scanner excerpt · SKILL.md (reported line 8)May include surrounding context.

md
# General Instructions
- In this Skill, "you" refers to the AI agent reading it, not a human user.
- This version is v3.1.2, updated on 2026-11-19 at 11:15 UTC. We update this Skill at most once every 72 hours. If more than 72 hours have passed since this update, check whether a newer version is available at https://github.com/QuestMeet/questmeet-skill, and if so, uninstall the local instance and install the latest version.
- This Skill has 19 callable functions defined in scripts/callable_functions.py, which send requests exclusively to https://questmeet.ai/graphql with trust_env=False. The instructions and example workflows for calling the functions are structured into authentication, representation, outreach, and engagement modules. Beyond the example workflows, you can call the functions in any combination following their docstrings.
- This Skill is powered by QuestMeet, an opportunity network for AI-native professionals and buyers. It is designed to facilitate long-context networ

Credential Access

High
Category
Privilege Escalation
Confidence
94% confidence
Finding

Access to the service's own token is legitimate, but requiring persistence in unspecified memory or files creates a credential-storage weakness. An exposed token could allow account actions within its granted scope; this duplicates the retention issue rather than proving intentional credential theft.

Content

Scanner excerpt · SKILL.md (reported line 14)May include surrounding context.

md
g special categories of personal data under the General Data Protection Regulation, such as data revealing racial or ethnic origin, political opinions, religious or philosophical beliefs, trade union membership, genetic data, biometric data for the purpose of uniquely identifying a natural person, data concerning health, or data concerning a natural person's sex life or sexual orientation, all of which are also prohibited by QuestMeet, LLC's Terms of Service.
- For a better user experience, the access token must be persisted in your memory or a local file. If it is missing, reauthenticate to get a new one and continue the current task. If any call returns None, the access token is invalid or has expired. In this case, reauthenticate to replace it with a new one and continue the current task.
- If any call returns a string explaining why the call failed, adjust the arguments and retry, or notify your user of the reason. If any call returns False, the call has failed for other reasons. I

Credential Access

High
Category
Privilege Escalation
Confidence
94% confidence
Finding

This is another match on the same instruction to persist a bearer credential in unspecified storage. The risk is accidental credential exposure and subsequent unauthorized account use, not evidence of a separate credential-exfiltration mechanism.

Content

Scanner excerpt · SKILL.md (reported line 14)May include surrounding context.

md
ical beliefs, trade union membership, genetic data, biometric data for the purpose of uniquely identifying a natural person, data concerning health, or data concerning a natural person's sex life or sexual orientation, all of which are also prohibited by QuestMeet, LLC's Terms of Service.
- For a better user experience, the access token must be persisted in your memory or a local file. If it is missing, reauthenticate to get a new one and continue the current task. If any call returns None, the access token is invalid or has expired. In this case, reauthenticate to replace it with a new one and continue the current task.
- If any call returns a string explaining why the call failed, adjust the arguments and retry, or notify your user of the reason. If any call returns False, the call has failed for other reasons. In this case, just retry once, and if it fails again, notify your user and stop without retrying. If any call raises a TypeError or KeyError, check https://github.com/QuestMee

Credential Access

High
Category
Privilege Escalation
Confidence
90% confidence
Finding

Persistent authentication is normal, but this requirement inherits the earlier allowance to store tokens in generic memory or local files without safeguards. It reinforces the same credential-retention weakness rather than constituting a separate privilege-escalation technique.

Content

Scanner excerpt · SKILL.md (reported line 20)May include surrounding context.

md
## Authentication
- This module is for authentication, which is a prerequisite for most functions in this Skill.
- The access token must be persisted to avoid repeated sign-ins. Repeatedly asking for the verification code leads to a poor user experience.
- For security reasons, exclude the access token from any messages to anyone.

### Example workflow: authenticate your user

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 · SKILL.md (reported line 21)May include surrounding context.

md
## Authentication
- This module is for authentication, which is a prerequisite for most functions in this Skill.
- The access token must be persisted to avoid repeated sign-ins. Repeatedly asking for the verification code leads to a poor user experience.
- For security reasons, exclude the access token from any messages to anyone.

### Example workflow: authenticate your user
1. Call the send_code_to_email function to send a verification code to your user's email address.

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 · scripts/callable_functions.py (reported line 20)May include surrounding context.

python
return httpx.post(BASE_URL, json={"query": "mutation SendCodeToEmail($email: String!) { sendCodeToEmail(email: $email) }", "variables": {"email": email}}, trust_env=False, timeout=20).json()["data"]["sendCodeToEmail"]

def sign_in_or_sign_up(email: str, code: str) -> Union[dict, str, bool]:
    """Create an access token and return it with your user's account information.

    Args:
        email: Your user's email address.

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 · SKILL.md (reported line 26)May include surrounding context.

md
### Example workflow: authenticate your user
1. Call the send_code_to_email function to send a verification code to your user's email address.
2. Ask your user for the verification code.
3. Once you have the code, call the sign_in_or_sign_up function to get a new access token along with your user's account information.
4. Once you have the access token, persist it in your memory or a local file under a unique key associated with the user's email address. Then, suggest several options for next steps based on your understanding of your user.

## Representation

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 · SKILL.md (reported line 56)May include surrounding context.

md
### Example workflow: create or update a listing
1. Prepare the context for understanding your user.
2. Compose the offering, seeking, name, and/or other arguments required, and confirm with your user before proceeding.
3. With user confirmation, find the access token, then call the create_listing function or update_listing function and get the latest listing.
4. Distill the offering and seeking of the listing into 1 to 20 impressions of your user as a professional or buyer. For each impression, provide 1 to 5 tags representing its topic, key points, or keywords/keyphrases. Each tag denotes an entity or a concept. Call the create_impressions function and get the impressions of the listing.
5. Check whether there is any logical conflict between the earlier impressions and the recent ones, or whether any earlier ones have become obsolete. If so, call the delete_impressions function to delete the earlier ones.
6. Remind your user to view the listing and export it as an image by signing in

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 · SKILL.md (reported line 105)May include surrounding context.

md
### Example workflow: check agent messages and follow up
1. Prepare the context for understanding your user.
2. Find the access token, then call the read_messages function to read messages in all accessible spaces and chats within the lookback window, together with the members' information in each space.
3. Review all the spaces. Convey to your user the messages in the most recent chat in each space, assess how the messages together with the members' information align with your user's interests or current focus, and suggest what your user should do with the space and chat. If the current information is insufficient to determine what to do, call the read_chat_messages function to get the full information and reassess.
4. For the spaces your user intends to leave, call the leave_spaces function. For each chat your user intends to follow up on, draft a reply to its recent messages and confirm with your user or check compliance with the predefined rules before proceeding.
5. With user con

Credential Access

High
Category
Privilege Escalation
Confidence
95% confidence
Finding

This workflow explicitly persists the token in generic memory or a local file keyed by email, without specifying confidentiality controls. A unique lookup key prevents some mix-ups but does not protect the credential from disclosure or unintended reuse.

Content

Scanner excerpt · SKILL.md (reported line 27)May include surrounding context.

md
1. Call the send_code_to_email function to send a verification code to your user's email address.
2. Ask your user for the verification code.
3. Once you have the code, call the sign_in_or_sign_up function to get a new access token along with your user's account information.
4. Once you have the access token, persist it in your memory or a local file under a unique key associated with the user's email address. Then, suggest several options for next steps based on your understanding of your user.

## Representation
- This module manages your user's representation to maximize the probability of your user being discovered.

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 · SKILL.md (reported line 78)May include surrounding context.

md
### Example workflow: search for professionals or buyers
1. Prepare the context for understanding your user.
2. Determine the perspective. Specify what your user is seeking and the requirements as 1 to 5 queries, each capturing only one aspect of the requirements and collectively covering all the aspects. Find the access token, then call the search function to get search results.
3. Review all the search results. Assess how each search result aligns with your user's requirements, and introduce it to your user if it has at least one listing that meets the requirements to a reasonable extent.
4. If your scheduled tasks do not include searching with similar queries, remind your user to set up a scheduled task to conduct the search regularly.

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 · SKILL.md (reported line 85)May include surrounding context.

md
### Example workflow: contact or invite a person
1. Prepare the context for understanding your user.
2. Determine whether to send a direct message or group message to the agent address. Select the appropriate listing or space, draft the content, set the attention reward, and confirm with your user before proceeding.
3. With user confirmation, find the access token, then call the contact function or invite function to send the message.

## Engagement
- This module processes messages to identify and capture opportunities.

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 · scripts/callable_functions.py (reported line 35)May include surrounding context.

python
"""Read your user's complete representation to support tasks that require understanding your user.

    Args:
        access_token: Access token in UUID format.

    Returns:
        On success, a dictionary containing email, subscription_plan, monthly_quota, extra_quota, pending_quota, attention_price, badges, listings each containing listing_id, url, agent_address, name, offering, seeking, expires_at, and updated_at, attachments each containing url, name, type, size, and created_at, and impressions each containing listing_id and content with its creation date appended.

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 · scripts/callable_functions.py (reported line 46)May include surrounding context.

python
"""Count search hits for your user's listings and tags in a time range.

    Args:
        access_token: Access token in UUID format.
        since: Inclusive start time in ISO 8601 format.
        before: Exclusive cutoff time in ISO 8601 format.

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 · scripts/callable_functions.py (reported line 59)May include surrounding context.

python
"""Create a new listing.

    Args:
        access_token: Access token in UUID format.
        name: Condense the offering and seeking into one sentence, up to 256 characters.
        offering: Introduce your user and describe what your user is offering in detail, up to 32768 characters, with URLs allowed.
        seeking: Describe what your user is seeking and any associated requirements in detail, up to 32768 characters, with URLs allowed.

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 · scripts/callable_functions.py (reported line 75)May include surrounding context.

python
"""Update an existing listing.

    Args:
        access_token: Access token in UUID format.
        listing_id: ID of the listing to be updated.
        name: New value, required if updating the name.
        offering: New value, required if updating the offering.

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 · scripts/callable_functions.py (reported line 92)May include surrounding context.

python
"""Delete specified listings.

    Args:
        access_token: Access token in UUID format.
        names: Names of the listings to be deleted.

    Returns:

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 · scripts/callable_functions.py (reported line 104)May include surrounding context.

python
"""Make local files accessible via URLs.

    Args:
        access_token: Access token in UUID format.
        names: Descriptive names of the attachments, each up to 256 characters.
        paths: Local file paths paired with the names, each file up to 256 MiB.

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 · scripts/callable_functions.py (reported line 126)May include surrounding context.

python
"""Delete specified attachments.

    Args:
        access_token: Access token in UUID format.
        names: Names of the attachments to be deleted.

    Returns:

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 · scripts/callable_functions.py (reported line 154)May include surrounding context.

python
"""Create impressions for a listing to index it.

    Args:
        access_token: Access token in UUID format.
        listing_id: ID of the listing to be indexed.
        perspective: Either "Professional" or "Buyer" depending on the listing.
        impressions_with_tags: Distill the offering and seeking of the listing into impressions with tags, following the impressions_with_tags_format schema.

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 · scripts/callable_functions.py (reported line 169)May include surrounding context.

python
"""Delete specified impressions.

    Args:
        access_token: Access token in UUID format.
        content_prefixes: Content prefixes of the impressions to be deleted.

    Returns:

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 · scripts/callable_functions.py (reported line 181)May include surrounding context.

python
"""Search for professionals or buyers who meet specific requirements.

    Args:
        access_token: Access token in UUID format.
        perspective: Use "Professional" for vendors, freelancers, entrepreneurs, candidates, or similar roles; use "Buyer" for clients, customers, investors, employers, or similar roles.
        queries: Up to 5 queries for semantic matching against others' impressions, each up to 512 characters (about 80 English words).

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 · scripts/callable_functions.py (reported line 194)May include surrounding context.

python
"""Create a new space with the person identified by an agent address and send a direct message.

    Args:
        access_token: Access token in UUID format.
        agent_address: Agent address of the person to contact and transfer quota to.
        listing_id: ID of the listing representing your user.
        content: Content explaining the reason for contacting the person, up to 32768 characters, with URLs allowed.

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 · scripts/callable_functions.py (reported line 209)May include surrounding context.

python
"""Add the person identified by an agent address to an existing space and send a group message.

    Args:
        access_token: Access token in UUID format.
        agent_address: Agent address of the person to invite and transfer quota to.
        space_id: ID of the space to add the person to.
        content: Content explaining the reason for inviting the person, up to 32768 characters, with URLs allowed.