Back to skill

Security audit

AI Daily Intelligence Digest

Security checks for vulnerabilities and agentic risk

Overview

The skill is coherent, but it automatically writes AI-generated RSS digests to a Feishu Wiki and relies on unpinned external skills in that publishing path.

Install only if you are comfortable with scheduled Feishu Wiki writes from AI-generated summaries. Use a tightly scoped Feishu credential, verify the destination space and permissions, review or pin the dependency skills before use, and avoid enabling cron until you have tested a manual run and confirmed the digest content is appropriate.

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

T08 · Insecure Dependencies

Warning
Location
workflow.json:6
Finding

Unpinned External Skill Dependencies Create a Supply-Chain Risk

Content
View full analysis

Vulnerability Details

File Location: workflow.json:6-10, workflow.json:15, workflow.json:33, workflow.json:54; SKILL.md:13; README.md:140-144
Vulnerability Type: T08: Insecure Dependencies
Risk Level: Medium

Vulnerable Code

workflow.json:6-10:

json
"skills": [
  "rss-aggregator",
  "summarize-pro",
  "feishu-wiki"
],

The workflow subsequently invokes these dependencies by unversioned names:

json
"skill": "rss-aggregator"
json
"skill": "summarize-pro"
json
"skill": "feishu-wiki"

SKILL.md:13:

json
"skills": ["rss-aggregator", "summarize-pro", "feishu-wiki"],

README.md:140-144 permits any compatible version at or above the stated minimum:

markdown
| Skill | Version | Purpose |
|---|---|---|
| `rss-aggregator` | ≥1.0.0 | Multi-source RSS aggregation |
| `summarize-pro` | ≥1.0.0 | AI batch summarization and translation |
| `feishu-wiki` | ≥1.0.0 | Feishu Wiki knowledge-base writing |

Technical Analysis

The workflow identifies three executable third-party Skills only by package name. It does not pin exact versions, provide integrity hashes, identify a trusted registry, or enforce publisher identity. The documented ≥1.0.0 ranges further allow future versions that were not included in this audit.

These dependencies occupy security-sensitive positions in the data flow:

  • rss-aggregator performs outbound requests to configured feed URLs.
  • summarize-pro receives fetched article data and may transmit it to an AI service.
  • feishu-wiki receives generated content, a Wiki space identifier, and a parent-node token and operates using Feishu credentials available to the runtime.

The network transfers themselves are necessary for the declared aggregation, summarization, and Wiki-publication functionality. No direct transmission of app secrets, private keys, or credentials to an attacker-control ...[truncated 1914 chars]

Remediation
View remediation

Remediation Suggestions

  1. Pin each external Skill to an exact, reviewed version rather than a package name or open-ended minimum version.
  2. Record and verify cryptographic integrity hashes or signed provenance metadata before execution.
  3. Resolve dependencies only through an explicitly trusted registry and require verified publisher identities.
  4. Maintain a lock file or equivalent immutable dependency manifest and require review before updating it.
  5. Audit the source and release artifacts of rss-aggregator, summarize-pro, and feishu-wiki.
  6. Run each dependency in a restricted sandbox:
    • Allow the RSS component to access only configured feed hosts.
    • Allow the summarizer to access only the approved AI endpoint.
    • Allow the Feishu component to access only required Feishu API endpoints.
    • Deny unnecessary filesystem, shell, process, and environment-variable access.
  7. Use a Feishu credential restricted to creating pages in the intended Wiki space or parent node. Do not provide tenant-wide administrative permissions.
  8. Avoid exposing credentials directly to components that do not require them, and redact tokens, page content, and identifiers from logs.
  9. Add automated dependency provenance, signature, vulnerability, and unexpected-permission checks to the release process.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (9)

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 38)May include surrounding context.

md
## Prerequisites

1. **RSS Feeds configured** — Add your target RSS feed URLs to the config
2. **Feishu Wiki access** — Valid Feishu access token and target Wiki space/node
3. **OpenClaw environment variables**:
   - `FEISHU_APP_ID` / `FEISHU_APP_SECRET` (for Wiki API)
   - `RSS_FEED_URLS` (comma-separated, optional override)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The README recommends a cron-based automation that publishes to Feishu Wiki on an ongoing basis, but it does not clearly warn users that this creates recurring external writes to a third-party service. In an enterprise context, this can lead to unintended disclosure, persistent noisy updates, or compliance issues if users enable the schedule without understanding that content will be automatically pushed outside the local environment every day.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill clearly states it will publish aggregated articles and AI-generated summaries to Feishu Wiki, but it does not prominently warn users about the data disclosure implications of sending externally sourced content and generated output into a shared knowledge base. This can lead to unintended publication of sensitive, copyrighted, inaccurate, or policy-violating material if users assume the workflow is only local summarization.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The config hard-codes language: en, which is a natural-language locale choice applied by default. Under the policy, forcing a specific language without offering user choice or documenting a justified region-specific constraint is a violation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The workflow automatically publishes AI-generated content to Feishu Wiki using configured space and parent node identifiers, with no approval, preview, or explicit user confirmation step. Because the content originates from external RSS feeds and an LLM summarization stage, this can cause unintended disclosure of inaccurate, sensitive, manipulated, or prompt-injected content directly into a shared knowledge base.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

On publish failure, the workflow prints the digest content to the console, which may expose summarized internal or externally sourced content in logs, terminals, CI output, or monitoring systems not intended for content storage. Error paths are often less protected, so this fallback can become an unintended data disclosure channel.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

The config example hard-codes language: en, which indicates a locale constraint in the skill's documented behavior. The README does not present this as optional, offer alternative languages, or justify why English-only processing is required, which can violate language/locale policy expectations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The natural-language description includes Chinese text for core behavior and use cases, while the rest of the document is primarily in English, and there is no indication that the skill is intentionally limited to Chinese-speaking users or that users may choose a language. This can violate language/locale policy expectations when a specific language is imposed without opt-in or justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The configuration sets the default article language to "en", which imposes a language preference in the workflow without any indication that the user can opt into another locale or that the restriction is required. This can violate language-choice policy when a specific language is forced by default in user-facing behavior.

Content

No source excerpt is available for this finding.