Back to skill

Security audit

公告事件速读卡·关键数逐字带出处

Security checks for vulnerabilities and agentic risk

Overview

Review is needed because the connection examples can expose your API key to destinations beyond the documented restrictions.

Use the local card workflow without a key until credential handling is corrected. Before authenticated checks, require validated HTTPS destinations and redirect protection in both the helper and shell examples. Share only public or redacted documents with Cue, and review optional companion skills and pinned installation sources. All bundled files were inspected; remote services and separately installed skills were not audited.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Rogue AgentSelf-Modification, Session Persistence
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
Findings (25)

Tainted flow: 'req' from os.environ.get (line 188, credential/environment) → urllib.request.urlopen (network output)

Critical
Category
Data Flow
Confidence
97% confidence
Finding

Reading the API key and sending it as an authorization header is legitimate for the explicitly requested connectivity test, and the initial URL is guarded. However, the default urllib redirect handling can forward that header to an unvalidated destination, making this credential-to-network flow exploitable rather than inherently critical merely because authentication occurs.

Content

Scanner excerpt · scripts/cue_connectivity_check.py (reported line 193)May include surrounding context.

python
code = None
    raw = ""
    try:
        with urllib.request.urlopen(req, timeout=args.timeout) as resp:
            code, raw = resp.status, resp.read().decode("utf-8", "replace")
    except urllib.error.HTTPError as err:
        detail = ""

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · README.en.md (reported line 84)May include surrounding context.

bash
   CATALOG="https://cuecue.cn/api/mcp-catalog"
   curl -s "$CATALOG" | python3 -c 'import json,sys
   rows=json.load(sys.stdin)["data"]["domains"]
   [print(r.get("id"), r.get("external_status"), (r.get("routing") or {}).get("url", "-")) for r in rows]'
   DOMAIN_ID="<replace with one of the ids listed by the line above>"

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · README.en.md (reported line 88)May include surrounding context.

md
rows=json.load(sys.stdin)["data"]["domains"]
   [print(r.get("id"), r.get("external_status"), (r.get("routing") or {}).get("url", "-")) for r in rows]'
   DOMAIN_ID="<replace with one of the ids listed by the line above>"
   ROUTING_URL="$(curl -s "$CATALOG" | python3 -c 'import json,sys
   rows=json.load(sys.stdin)["data"]["domains"]
   hit=[r["routing"]["url"] for r in rows if r.get("id")==sys.argv[1] and r.get("routing")]
   print(*hit,sep="\n") if hit else sys.exit("Not found: that id is not in the catalog response (or its routing has no url field) - run the line above and pick one of the ids listed")' "$DOMAIN_ID")"

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · README.md (reported line 82)May include surrounding context.

bash
   CATALOG="https://cuecue.cn/api/mcp-catalog"
   curl -s "$CATALOG" | python3 -c 'import json,sys
   rows=json.load(sys.stdin)["data"]["domains"]
   [print(r.get("id"), r.get("external_status"), (r.get("routing") or {}).get("url", "-")) for r in rows]'
   DOMAIN_ID="<把这里换成上一行列出的那个域 id>"

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · README.md (reported line 86)May include surrounding context.

md
rows=json.load(sys.stdin)["data"]["domains"]
   [print(r.get("id"), r.get("external_status"), (r.get("routing") or {}).get("url", "-")) for r in rows]'
   DOMAIN_ID="<把这里换成上一行列出的那个域 id>"
   ROUTING_URL="$(curl -s "$CATALOG" | python3 -c 'import json,sys
   rows=json.load(sys.stdin)["data"]["domains"]
   hit=[r["routing"]["url"] for r in rows if r.get("id")==sys.argv[1] and r.get("routing")]
   print(*hit,sep="\n") if hit else sys.exit("未取到:该 id 不在目录现值里(或它没有 routing 里的 url 字段)——先跑上面那行列出现场的 id 再填")' "$DOMAIN_ID")"

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied code's primary purpose is service connectivity testing, not A-share announcement triage. Its implemented outputs are domain listings and connection diagnostics, with no filing-processing or card-generation logic. Although connectivity checks could support a larger data-backed skill, this script exposes a standalone catalog-discovery and credentialed testing capability without any announcement-specific integration. HTTPS and hostname checks constrain the initial credential destination, but do not make that capability part of the declared purpose.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied chunk implements a generic authenticated MCP discovery workflow. It does not retrieve or analyze an announcement, extract cited figures, compare similar events, or generate the promised card. MCP discovery could support a larger implementation, but this chunk provides no connection between the selected service and announcement analysis. The observable behavior is therefore materially different from the declared purpose. The credential transmission is explicit, although the code alone does not establish malicious intent or an unauthorized recipient.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied shell snippet implements generic MCP discovery and connectivity testing. It does not retrieve or parse an announcement, extract quoted figures, construct the promised card, or perform comparable-event checks. Network requests could support announcement analysis, but this snippet only discovers services and enumerates tools, with no demonstrated connection to that workflow. It also sends the API credential to the selected catalog-provided URL; this behavior is observable, but the snippet alone does not establish malicious credential exfiltration.

Content

No source excerpt is available for this finding.

Automatic redirects can forward credentials beyond the destination restrictions promised by the documentation.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The destination guard checks only the initial URL, while urllib's default redirect handling can preserve the Authorization header when following a POST redirect as a GET. An authorized endpoint that returns an attacker-controlled redirect can therefore leak the key outside the promised host and HTTPS restrictions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

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.

Content

Scanner excerpt · README.en.md (reported line 44)May include surrounding context.

bash
# 1. Put it in the skills directory
mkdir -p ~/.workbuddy/skills
cp -r event-brief-cn ~/.workbuddy/skills/

# 2. Copy the card template and fill it following section 1

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The recommended installation command executes an unpinned npm CLI and installs skills from a mutable repository reference. Although this is not an MCP server definition, compromised or unexpectedly changed upstream content could execute with the installer's privileges or introduce malicious agent instructions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

Selecting an individual skill reduces installation scope but does not pin either the executable npm CLI or the repository revision. An upstream compromise could therefore substitute executable installer code or adversarial skill instructions during setup.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
99% confidence
Finding

The copyable shell block sends CUE_API_KEY to a URL taken directly from catalog data without validating its scheme or destination host. A malicious or compromised catalog entry could direct the bearer credential to an attacker or an unencrypted endpoint; the separate script's documented egress checks do not protect this shell block.

Content

Scanner excerpt · README.en.md (reported line 93)May include surrounding context.

md
hit=[r["routing"]["url"] for r in rows if r.get("id")==sys.argv[1] and r.get("routing")]
   print(*hit,sep="\n") if hit else sys.exit("Not found: that id is not in the catalog response (or its routing has no url field) - run the line above and pick one of the ids listed")' "$DOMAIN_ID")"
   [ -n "$ROUTING_URL" ] || { echo "Fetch failed: ROUTING_URL is empty - run the line above, then set DOMAIN_ID to one of the ids listed" >&2; exit 1; }
   curl -s -X POST "$ROUTING_URL" \
     -H "Authorization: Bearer $CUE_API_KEY" \
     -H "Accept: application/json, text/event-stream" \
     -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' \

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The package's own recommended installation command leaves both the npm CLI version and source repository revision unspecified. Checking that the skill name appears afterward verifies installation presence, not the integrity of the executable or installed instructions.

Content

No source excerpt is available for this finding.

Attachment parsing lacks a warning that document contents are sent to an external service.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The README routes attachments through an external parsing channel without explicitly explaining that document contents leave the local environment. Public announcements reduce typical sensitivity, but arbitrary attachments are accepted, and consent concerning time or billing does not establish informed consent to disclose private annotations or other confidential content.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

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.

Content

Scanner excerpt · README.md (reported line 42)May include surrounding context.

bash
# 1. 放到技能目录
mkdir -p ~/.workbuddy/skills
cp -r event-brief-cn ~/.workbuddy/skills/

# 2. 复制卡模板,按 §一 三步填

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The installation command uses an unpinned npx package and a mutable skill repository reference. Although this is ordinary optional setup rather than evidence of malicious intent, a compromised upstream release could execute installer code or introduce adversarial agent instructions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

Selecting an individual skill does not pin either the npx installer or the upstream collection revision. Future upstream changes can therefore alter the code and agent instructions installed by the same documented command.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
99% confidence
Finding

The copyable shell example sends the API key to a URL taken directly from remote catalog data without validating its scheme or destination host. A compromised or misconfigured catalog could direct the credential to an attacker or an unencrypted endpoint; the safeguards described for the separate Python helper do not protect this curl command.

Content

Scanner excerpt · README.md (reported line 91)May include surrounding context.

md
hit=[r["routing"]["url"] for r in rows if r.get("id")==sys.argv[1] and r.get("routing")]
   print(*hit,sep="\n") if hit else sys.exit("未取到:该 id 不在目录现值里(或它没有 routing 里的 url 字段)——先跑上面那行列出现场的 id 再填")' "$DOMAIN_ID")"
   [ -n "$ROUTING_URL" ] || { echo "取值失败:ROUTING_URL 为空——先跑上面那行列出现场 id,再把 DOMAIN_ID 换成其中之一" >&2; exit 1; }
   curl -s -X POST "$ROUTING_URL" \
     -H "Authorization: Bearer $CUE_API_KEY" \
     -H "Accept: application/json, text/event-stream" \
     -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' \

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The self-install command resolves a mutable repository and an unversioned npx installer. This creates a supply-chain exposure even though installing the requested skill is a legitimate operation.

Content

No source excerpt is available for this finding.

The attachment-parsing workflow lacks an explicit third-party data-sharing warning.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The documented attachment workflow delegates parsing to an external service without explicitly warning that document contents will leave the user's environment. Public announcements usually have low confidentiality, but uploaded attachments can contain private annotations or unrelated sensitive information, and billing confirmation is not informed data-sharing consent.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, network.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The workflow sends attachments and long announcement text to an external parsing service without a privacy warning.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

The skill directs attachments and long text to the external omni parsing channel without an explicit privacy warning or content-sharing consent requirement. Although official announcements are public, user-supplied files and pasted text can include confidential additions that should not automatically be disclosed.

Content

No source excerpt is available for this finding.

The documented default 'no packets' behavior contradicts the unconditional catalog GET.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The claimed default 'no packets' behavior is false because execution retrieves the remote catalog unless a local catalog file is supplied, even without --post. This can cause unexpected network contact and disclose ordinary connection metadata, although this anonymous GET does not attach the API key or announcement contents.

Content

No source excerpt is available for this finding.