tong-jincheng-relationship-skill
Warn
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's documentation guides users to install external code from an untrusted source via the command
npx skills add hotcoffeeshake/tong-jincheng-skill. This repository is not associated with a trusted vendor and represents an unverifiable third-party dependency. - [COMMAND_EXECUTION]: The installation instructions involve the execution of shell commands to download and integrate the skill into the local environment.
- [INDIRECT_PROMPT_INJECTION]: The skill architecture is designed to ingest and process external data from research files, which presents a vulnerability surface if those files are modified by untrusted contributors.
- Ingestion points: The skill explicitly reads content from
references/research/(e.g.,01-writings.md,03-expression-dna.md) to build the system prompt. - Boundary markers: The integration examples show direct string interpolation of these files into the
systemPromptvariable without the use of delimiters or 'ignore' instructions to separate data from instructions. - Capability inventory: The documented implementation uses the Anthropic SDK to execute messages based on the assembled prompt.
- Sanitization: No sanitization, filtering, or validation logic is present to ensure that the content of the research files does not contain malicious instructions meant to hijack the agent's behavior.
Audit Metadata