shellcode-dev
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONOBFUSCATIONINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [METADATA_POISONING]: The skill's YAML frontmatter lists the author as 'malskill', which differs from the recognized author 'aeondave' provided in the system context. This represents a discrepancy in identity metadata.
- [COMMAND_EXECUTION]: The skill instructs the agent to use several powerful command-line tools and programming libraries including pwntools, Keystone, Capstone, and Unicorn for payload generation, assembly, and emulation tasks.
- [DYNAMIC_EXECUTION]: The core purpose of the skill involves the development and execution of shellcode, featuring advanced techniques like reflective/manual in-memory loading (PE/ELF/Mach-O), section overlays, and runtime decryption of payloads using key splitting over environment state.
- [OBFUSCATION]: The skill contains detailed implementation guidance for obfuscating shellcode payloads, specifically mentioning Mixed Boolean-Arithmetic (MBA), chunked ASCII containers, and key splitting to defeat static analysis and ML-based classifiers.
- [INDIRECT_PROMPT_INJECTION]: The skill has a high-risk surface for indirect prompt injection because it is designed to ingest untrusted assembly or C code from users or files and has the capability to assemble, emulate, and execute it. 1. Ingestion points: User-provided assembly/C code and shellcode bytes (SKILL.md, references/platform-workflows.md). 2. Boundary markers: Absent. 3. Capability inventory: Assembly via Keystone, emulation via Unicorn, and execution via C/Go wrappers. 4. Sanitization: Absent; the skill focuses on raw engineering without input validation layers.
Audit Metadata