The Agent Skills Directory

[COMMAND_EXECUTION]: The utility script scripts/patch_wikilink.py is designed to modify the locally installed graphifyy Python package.
The script locates wiki.py within the package's site-packages directory, applies regex transformations to its source code, and overwrites the file.
Instructions in references/build-and-fallback-recipes.md suggest automating this script by wiring it into post-implementation hooks (e.g., ~/.jeo/config.json), leading to silent modification of the runtime environment.
[REMOTE_CODE_EXECUTION]: The script scripts/patch_wikilink.py contains a dynamic execution pattern identified by static analysis.
It utilizes the exec() function within its self_test() routine to execute code generated via string manipulation and AST parsing. Although the test source is hardcoded, this pattern of generating and executing code at runtime is a known risk vector.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download of additional instruction sets from an external repository.
It provides commands such as npx skills add https://github.com/akillness/jeo-skills --skill graphify -a universal to fetch components from the author's GitHub profile.
[DATA_EXFILTRATION]: The skill exhibits an attack surface for indirect prompt injection through data ingestion.
Ingestion points: The graphify update and graphify add <url> commands ingest arbitrary external data (source code, documents, web content) to build knowledge graphs.
Boundary markers: The instructions lack explicit delimiters or warnings to ignore embedded instructions within the processed corpus.
Capability inventory: The agent is granted access to the Bash tool, which could be exploited if malicious instructions inside a processed document mislead the agent during a graphify query session.
Sanitization: There is no evidence of sanitization or safety filtering for the content before it is processed into the graph and subsequently queried.

graphify