The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous examples of shell commands involving redis-cli, docker, python3, and cargo. These are legitimate commands used for database administration, container orchestration, and data migration, which align with the skill's primary purpose. Evidence includes usage of falkordb-bulk-insert, docker run, and cargo build in SKILL.md and the ingestion-skills/ directory.
[EXTERNAL_DOWNLOADS]: The documentation references cloning the toolset from the official FalkorDB GitHub repository (github.com/FalkorDB/skills.git) in README.md. This is a trusted vendor resource.
[INDIRECT_PROMPT_INJECTION]: The skill identifies ingestion points for external data from CSV, SQL, Neo4j, and Neptune in ingestion-skills/. While processing such data can be a vector for indirect prompt injection, the skill acts as a documentation guide and does not automate the processing of untrusted remote data. Evidence includes falkordb-bulk-insert and various migration scripts.
[CREDENTIALS_UNSAFE]: Examples in SKILL.md and operations-skills/set-authentication/skill.md use obvious placeholders such as '--password secret' and '--requirepass yourpassword'. These are documentation conventions and do not represent actual sensitive data exposure.
[DYNAMIC_EXECUTION]: The skill demonstrates the use of User-Defined Functions (UDFs) via db.udf_load() in udf-skills/load-javascript-udf/skill.md. This allows for loading and executing JavaScript within the FalkorDB environment, which is a core feature of the database. The examples provided are benign and instructional.

falkordb-skills