Back to skill

Security audit

EchoMark

Security checks for vulnerabilities and agentic risk

Overview

EchoMark is a disclosed tool-rating skill, but its default cloud flow sends and prints API credentials in unsafe ways that users should review before installing.

Install only if you are comfortable with a tool that records ratings locally and, by default for submissions, contacts an EchoMark cloud API. Prefer local-only mode, avoid using the default plaintext HTTP endpoint for any cloud feature, and treat any printed API key as exposed unless the package is changed to use HTTPS and suppress secret output.

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

Error
Location
scripts/config.py:4
Finding

Bearer API Credentials and Rating Data Transmitted Over Plaintext HTTP

Content
View full analysis

Vulnerability Details

File Location: scripts/config.py:4, with affected network operations in scripts/register.py:15-23, scripts/submit.py:22-40, and scripts/query.py:24-32
Vulnerability Type: Plaintext transmission of sensitive credentials and data
Risk Level: High

Vulnerable Code

scripts/config.py:4:

python
ECHO_MARK_API_URL = os.environ.get("ECHO_MARK_API_URL", "http://47.109.154.82:9527")

scripts/register.py:15-23:

python
def register(agent_type):
    """Register agent with EchoMark cloud service."""
    url = f"{ECHO_MARK_API_URL}/api/v1/agents/register"

    response = requests.post(url, json={"agent_type": agent_type}, timeout=API_TIMEOUT)
    response.raise_for_status()

    data = response.json()
    api_key = data["api_key"]

scripts/submit.py:22-40:

python
def submit_rating(tool_name, accuracy, efficiency, usability, stability, comment=""):
    """Submit a tool rating to cloud server."""
    api_key = load_api_key()
    url = f"{ECHO_MARK_API_URL}/api/v1/ratings"

    payload = {
        "tool_name": tool_name,
        "accuracy": accuracy,
        "efficiency": efficiency,
        "usability": usability,
        "stability": stability,
        "comment": comment,
    }

    headers = {"Authorization": f"Bearer {api_key}"}
    response = requests.post(url, json=payload, headers=headers, timeout=API_TIMEOUT)
    response.raise_for_status()

    return response.json()

scripts/query.py:24-32:

python
def query_cloud_rating(tool_name):
    """Query ratings for a specific tool from cloud server."""
    api_key = load_api_key()
    url = f"{ECHO_MARK_API_URL}/api/v1/ratings/{tool_name}"

    headers = {"Authorization": f"Bearer {api_key}"}
    response = requests.get(url, headers=headers, timeout=API_TIMEOUT)
    response.raise_for_status()

    return response.json()

Technical Analysis

...[truncated 2858 chars]

Remediation
View remediation

Remediation Suggestions

  1. Replace the default endpoint with an HTTPS URL backed by a valid, trusted TLS certificate.
  2. Reject http:// endpoints whenever registration or authenticated requests are performed. Validate the parsed URL scheme before sending any credential:
    python
    from urllib.parse import urlparse
    
    parsed = urlparse(ECHO_MARK_API_URL)
    if parsed.scheme != "https":
        raise ValueError("EchoMark cloud operations require an HTTPS endpoint")
    
  3. Retain the default certificate and hostname verification provided by requests; do not introduce verify=False or suppress TLS warnings.
  4. Redirecting HTTP to HTTPS is insufficient for credential-bearing requests because the initial request may already be exposed. Construct authenticated requests directly against HTTPS and disable unsafe downgrade behavior.
  5. Rotate or revoke all API keys that may previously have been returned or transmitted through the plaintext endpoint.
  6. Consider restricting environment-based endpoint overrides to HTTPS. If plaintext transport is required solely for local development, require an explicit development-only setting and restrict it to loopback addresses without real credentials.
  7. Add automated tests asserting that registration, submission, and cloud querying fail closed when configured with a non-HTTPS remote endpoint.
  8. Update the security documentation to explicitly state that cloud features require TLS and explain how endpoint validation is enforced.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (11)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The skill is presented primarily as a tool-rating aid, but it also registers the agent with a remote service, generates or retrieves an API key, stores that key locally, and may expose credential material on stdout. This mismatch is dangerous because operators may approve or auto-trigger the skill expecting low-risk local rating behavior while it actually performs sensitive account creation and credential handling.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The skill is presented primarily as a tool-rating aid, but it also registers the agent with a remote service, generates or retrieves an API key, stores that key locally, and may expose credential material on stdout. This mismatch is dangerous because operators may approve or auto-trigger the skill expecting low-risk local rating behavior while it actually performs sensitive account creation and credential handling.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding

The skill invokes shell commands, performs local file writes, accesses environment-derived credentials, and can communicate with a remote service, but it declares no explicit tool scope or permissions boundary. That makes the skill easier to invoke with more privilege than users or orchestrators may expect, increasing the chance of unintended network calls, credential handling, and filesystem modification.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill says to trigger before selecting a tool and after using any external tool, which is extremely broad. In practice this can cause frequent or automatic invocation across unrelated workflows, leading to unnecessary network submissions, repeated local writes, and wider exposure of tool usage metadata than users intended.

Content

No source excerpt is available for this finding.

Known Vulnerable Dependency: requests==2.31.0 — 6 advisory(ies): CVE-2024-47081 (Requests vulnerable to .netrc credentials leak via malicious URLs); CVE-2024-35195 (Requests `Session` object does not verify requests after making first request wi); CVE-2026-25645 (Requests has Insecure Temp File Reuse in its extract_zipped_paths() utility func) +3 more

Medium
Category
Supply Chain
Confidence
98% confidence
Finding

The dependency is pinned to requests==2.31.0, and the supplied finding indicates multiple published advisories affecting that version, including credential leakage and TLS/session verification issues. In a skill that explicitly interacts with external tools, APIs, and potentially attacker-controlled endpoints, using a known-vulnerable HTTP client increases the chance of credential exposure or insecure network communication.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This file implements agent registration with a cloud service and provisions a persistent API key, behavior that is not reflected in the stated skill purpose of rating/querying external tools. That mismatch increases supply-chain risk because operators may invoke the skill expecting local reputation functionality while it silently enrolls the environment with a remote service and creates credentials.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
80% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · scripts/register.py (reported line 17)May include surrounding context.

python
"""Register agent with EchoMark cloud service."""
    url = f"{ECHO_MARK_API_URL}/api/v1/agents/register"

    response = requests.post(url, json={"agent_type": agent_type}, timeout=API_TIMEOUT)
    response.raise_for_status()

    data = response.json()

External Transmission

Medium
Category
Data Exfiltration
Confidence
80% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · scripts/submit.py (reported line 40)May include surrounding context.

python
"""Register agent with EchoMark cloud service."""
    url = f"{ECHO_MARK_API_URL}/api/v1/agents/register"

    response = requests.post(url, json={"agent_type": agent_type}, timeout=API_TIMEOUT)
    response.raise_for_status()

    data = response.json()

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script silently persists the returned API key to disk without prior warning, confirmation, or disclosure of where and why the secret is being stored. Even with chmod 600 on Unix, unexpected secret persistence can violate user expectations, increase forensic footprint, and expose the credential on shared hosts, backups, sync folders, or systems where permission hardening is ineffective.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

The script prints the newly issued API key directly to stdout after registration. Secrets written to terminal output can be captured by shell history tools, CI logs, terminal recorders, shared sessions, or higher-level agent transcripts, causing immediate credential exposure.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The README states that ratings are always written to local SQLite and, by default, also submitted to the cloud, but the workflow does not present this as a prominent warning before use. In an agent skill context, silent persistence and network submission can cause unintended data-affecting behavior, especially if agents invoke the tool automatically after using external tools.

Content

No source excerpt is available for this finding.