antfu
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use package managers and command-line tools to manage the project lifecycle.
- Utilizes
pnpm,ni, andnpxfor dependency installation and script execution. - Recommends running
pnpm run lint --fixfor automated code formatting. - [EXTERNAL_DOWNLOADS]: The skill incorporates configuration and tools from prominent external repositories and registries.
- Fetches linting configurations from the
@antfu/eslint-configpackage. - Utilizes reusable GitHub Actions workflows from the
sxzz/workflowsrepository for CI/CD tasks like linting and testing. - [DYNAMIC_EXECUTION]: The skill provides a utility script intended to modify local configuration files at runtime.
- Includes an
alias.tsscript that usesfs.writeFileSyncto synchronize path mappings intsconfig.alias.jsonbased on the project's directory structure. - [INDIRECT_PROMPT_INJECTION]: The skill's workflow presents an attack surface where the agent processes external, potentially untrusted project data.
- Ingestion points: The agent reads and updates project-specific configuration files including
package.json,tsconfig.json, andeslint.config.mjs(SKILL.md, references/antfu-eslint-config.md). - Boundary markers: Explicit delimiters or warnings to ignore instructions embedded within the processed project files are not defined.
- Capability inventory: The skill has the capability to write to the file system and execute shell commands via package managers across all its configuration modules.
- Sanitization: Content is processed using standard JSON and YAML parsing; no additional sanitization for data interpolated into the agent's context is specified.
Audit Metadata