Back to skill

Security audit

Feishu All In One

Security checks for vulnerabilities and agentic risk

Overview

This Feishu skill mostly matches its stated purpose, but it needs Review because its callback server can forward full user interaction data to a configurable Gateway with limited disclosure and weak default scoping.

Install only after reviewing the callback forwarding behavior. Disable Gateway forwarding unless needed, restrict the Gateway URL to a trusted HTTPS or local endpoint, avoid sending raw callback data, protect Feishu App Secret and Gateway tokens, and regenerate the npm lockfile from an approved HTTPS registry with updated dependencies before production use.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/card-callback-server.js:34
Finding

Raw Feishu Callback Data Is Forwarded by Default to a Configurable Gateway

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
scripts/package-lock.json:17
Finding

Dependencies Are Retrieved from a Third-Party Registry over Plaintext HTTP

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (54)

Known Vulnerable Dependency: protobufjs==7.5.4 — 12 advisory(ies): CVE-2026-44294 (protobuf.js: Denial of service from crafted field names in generated code); CVE-2026-44293 (protobuf.js: Code injection through bytes field defaults in generated toObject c); CVE-2026-44289 (protobuf.js: Denial of service through unbounded protobuf recursion) +9 more

Critical
Category
Supply Chain
Confidence
90% confidence
Finding

protobufjs 7.5.4 is flagged with multiple serious issues including denial of service and possible code-injection-related flaws in generated conversion paths. Since the Feishu SDK depends on protobufjs and messaging platforms often parse complex, externally supplied payloads, a vulnerable protobuf implementation in the runtime materially increases risk from crafted input or event data.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The skill claims all features are fully validated and directly usable, but the analyzed behavior appears focused on card callback handling and may forward callback data externally while bypassing expected response logic. Overstated assurance can cause administrators to trust and run a component that handles user interaction data in ways not obvious from the description.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill claims all features are fully validated and directly usable, but the analyzed behavior appears focused on card callback handling and may forward callback data externally while bypassing expected response logic. Overstated assurance can cause administrators to trust and run a component that handles user interaction data in ways not obvious from the description.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The skill claims all features are fully validated and directly usable, but the analyzed behavior appears focused on card callback handling and may forward callback data externally while bypassing expected response logic. Overstated assurance can cause administrators to trust and run a component that handles user interaction data in ways not obvious from the description.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill claims all features are fully validated and directly usable, but the analyzed behavior appears focused on card callback handling and may forward callback data externally while bypassing expected response logic. Overstated assurance can cause administrators to trust and run a component that handles user interaction data in ways not obvious from the description.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill claims all features are fully validated and directly usable, but the analyzed behavior appears focused on card callback handling and may forward callback data externally while bypassing expected response logic. Overstated assurance can cause administrators to trust and run a component that handles user interaction data in ways not obvious from the description.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The skill claims all features are fully validated and directly usable, but the analyzed behavior appears focused on card callback handling and may forward callback data externally while bypassing expected response logic. Overstated assurance can cause administrators to trust and run a component that handles user interaction data in ways not obvious from the description.

Content

No source excerpt is available for this finding.

Ssd 3

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The callback handler packages and sends full callback contents to the Gateway, including raw event data that may contain operator identifiers, message context, and user-submitted form values. This is unnecessary for many integrations and significantly increases privacy and breach impact if the Gateway is compromised, misconfigured, or points to an untrusted endpoint. In this skill type, callback payloads are operationally sensitive because they reflect direct user interactions.

Content

No source excerpt is available for this finding.

Ssd 3

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

At the event entry point, the code explicitly extracts operator, action, context, and full raw event data and relays them to the Gateway asynchronously. That creates a broad data egress path for all card interactions, including appointment form inputs and survey responses processed later in the handler. Because the exfiltration occurs before action-specific validation or minimization, the exposure applies uniformly across all callback types.

Content

No source excerpt is available for this finding.

Known Vulnerable Dependency: axios==1.13.6 — 16 advisory(ies): CVE-2026-44494 (axios Vulnerable to Full Man-in-the-Middle via Prototype Pollution Gadget in `co); CVE-2026-44495 (axios Vulnerable to Credential Theft and Response Hijacking via Prototype Pollut); CVE-2025-62718 (Axios has a NO_PROXY Hostname Normalization Bypass that Leads to SSRF) +13 more

High
Category
Supply Chain
Confidence
95% confidence
Finding

axios 1.13.6 is a network-facing HTTP client used directly by this package and transitively by the Feishu SDK, so known SSRF, MITM, prototype-pollution-gadget, proxy-bypass, or redirect-related issues can be reachable in realistic deployments. A messaging integration skill commonly makes outbound requests using tokens and user-influenced URLs or proxy settings, which increases the practical risk of credential leakage or server-side request abuse.

Content

No source excerpt is available for this finding.

Known Vulnerable Dependency: form-data==4.0.5 — 1 advisory(ies): CVE-2026-12143 (form-data: CRLF injection in form-data via unescaped multipart field names and f)

High
Category
Supply Chain
Confidence
88% confidence
Finding

form-data 4.0.5 is flagged for CRLF injection through unescaped multipart field names/filenames. Because this skill advertises file and image sending, multipart uploads are plausibly part of its functionality, making it more likely that attacker-controlled metadata could trigger request smuggling or malformed outbound requests against downstream services.

Content

No source excerpt is available for this finding.

Known Vulnerable Dependency: ws==8.19.0 — 2 advisory(ies): CVE-2026-45736 (ws: Uninitialized memory disclosure); CVE-2026-48779 (ws: Memory exhaustion DoS from tiny fragments and data chunks)

High
Category
Supply Chain
Confidence
89% confidence
Finding

ws 8.19.0 is reported vulnerable to memory disclosure and memory exhaustion. Because the Feishu SDK may use WebSocket connectivity for event or stream handling, a messaging skill exposed to remote platform traffic could be susceptible to denial of service or data exposure if a malicious endpoint or intermediary can interact with the WebSocket client/server path.

Content

No source excerpt is available for this finding.

Known Vulnerable Dependency: axios==1.13.6 — 16 advisory(ies): CVE-2026-44494 (axios Vulnerable to Full Man-in-the-Middle via Prototype Pollution Gadget in `co); CVE-2026-44495 (axios Vulnerable to Credential Theft and Response Hijacking via Prototype Pollut); CVE-2025-62718 (Axios has a NO_PROXY Hostname Normalization Bypass that Leads to SSRF) +13 more

High
Category
Supply Chain
Confidence
99% confidence
Finding

The file permits resolution to an axios version identified as having numerous known vulnerabilities, including issues related to SSRF/proxy bypass and prototype-pollution-based exploitation. In a skill intended for message sending/receiving and external network communication, such flaws are more dangerous because they may expose tokens, redirect requests, or allow attacker-controlled network interactions.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

This markdown file includes commands that place App ID and App Secret into environment variables, but it does not warn users that these are sensitive credentials or advise safe handling. Under the markdown-specific missing-warning rule, credentials and privacy-affecting configuration should be accompanied by a user warning when the description instructs users to use them directly.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
77% confidence
Finding

The markdown describes launching a callback server and configuring event subscriptions, but it does not include any warning about network exposure, inbound events, or the need to secure and limit access to the running service. For markdown files, behaviors that may affect system integrity or privacy should be disclosed to the user.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding

The skill declares access to sensitive environment variables, local configuration files, and network functionality, but does not declare any explicit tool scope or permissions boundary. This creates a transparency and least-privilege problem: users may authorize or run a package without understanding that it can read secrets and communicate externally.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

This JSON manifest contains user-facing text exclusively in Chinese, including the title, field labels, placeholders, and button text. The file provides no indication that the skill is region-specific or that users can opt into this locale, which may violate language/locale policy requirements.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The card content is entirely hard-coded in Chinese, including the title and interaction text, with no indication that the skill is region-specific or that users can select another language. This creates a natural-language locale policy concern because the file imposes a specific language without opt-in or documented justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

Most user-visible messages, prompts, and card content are fixed in Chinese, and the interaction flow does not offer the user a language selection or opt-in. Under the stated policy, forcing a specific language without user choice is a natural-language policy violation unless the locale restriction is clearly documented and justified.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The script forwards Feishu card callback data to an OpenClaw Gateway even though the skill is presented as a Feishu messaging solution. That creates an undocumented secondary data flow to another service, and the payload includes user interaction metadata that may contain identifiers and submitted content. In a callback-handling skill, this broad forwarding is more dangerous because users and operators will reasonably expect processing to stay within Feishu/local handling unless explicitly disclosed.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

User callback data is exfiltrated to an external Gateway without any clear user-facing or admin-facing consent/notice in the skill behavior. Because card callbacks can include operator identity, context, and form submissions, silent transmission undermines privacy expectations and can violate data-handling requirements. The skill context makes this more dangerous because it is marketed as a ready-to-use messaging solution, not a telemetry relay.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

Most user-facing strings, logs, prompts, card content, and toasts are fixed in Chinese, and the skill does not generally provide locale selection or user opt-in. This can violate language/locale policy when a skill forces a specific language for all users.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The server forwards full Feishu card callback data, including operator details, action payloads, context, and raw event data, to an external or configurable OpenClaw Gateway. Because the skill is presented as a Feishu messaging solution and does not clearly disclose this secondary data flow, it creates an unexpected exfiltration path for user interaction data and potentially sensitive form contents.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

User interaction details are sent to the Gateway asynchronously without notice, consent, or data minimization. Since callback payloads can include operator identifiers, form values, and card context, this undisclosed transmission increases privacy and compliance risk and could expose business or personal data to another service boundary.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The handler constructs per-action responses but then always returns an empty object, so the intended card updates and user feedback are silently discarded. In a callback-driven workflow this can break confirmation semantics, hide failures, and cause repeated user actions or state confusion, especially where users believe an operation completed or was rejected based on stale UI.

Content

No source excerpt is available for this finding.