The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to install the Python package graphifyy via pip. Given the skill is named graphify, the use of graphifyy (with an extra 'y') is a strong indicator of a typosquatting attack intended to install a malicious dependency.- [COMMAND_EXECUTION]: The skill executes shell commands using a binary located in a hidden directory (.tools/graphify-venv/bin/graphify) and a local repository script (scripts/update-graphify.sh).- [OBFUSCATION]: The title uses the identifier XerahS, which is the author's name sharex written in reverse. While stylistic, reversed strings are a recognized technique for concealing identifiers from simple text scanners.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the src/ directory to build architectural relationship graphs, which the agent then uses to answer questions. This creates a vulnerability where malicious instructions embedded in source code could influence agent behavior.
Ingestion points: Scans files within src/ and documentation in docs/architecture/.
Boundary markers: Absent; the tool appears to interpolate symbol data directly into its query engine.
Capability inventory: Executes shell commands for querying, updates, and relationship mapping.
Sanitization: No evidence of sanitization or escaping for data ingested from the project files.

graphify