The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs agents and users to install the Hegel library using various package managers, including npm install @hegeldev/hegel, cargo add hegeltest, go get hegel.dev/go/hegel, and opam install hegel. These packages and domains are associated with the primary vendor ('hegeldev') and are required for the library's operation.
[DYNAMIC_EXECUTION]: The library architecture involves loading a native engine binary (libhegel) into the process via Foreign Function Interface (FFI) for languages such as TypeScript, Java, and OCaml. The OCaml documentation describes a dynamic search and download mechanism for this native component, targeting verified releases from the vendor's GitHub repository.
[INDIRECT_PROMPT_INJECTION]: The skill requires the agent to ingest and analyze untrusted third-party source code, docstrings, and existing test suites to derive properties. This ingestion process is vulnerable to indirect prompt injection if project source files contain malicious instructions meant to manipulate the agent.
Ingestion points: Workflow Step 2 in SKILL.md (Explore the Code Under Test) and guidelines in references/evolving-tests.md.
Boundary markers: The instructions include a prerequisite check for anti-AI clauses in project metadata (README, LICENSE, CONTRIBUTING.md) before any code is generated or contributed.
Capability inventory: File system read/write access for code analysis and test creation; external tool execution for package installation and running test suites.
Sanitization: The skill focuses on documentation and does not define automated sanitization logic for the code it reads.

hegel