The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Provides a CLI alternative using npx @claude-flow/cli@latest. This command fetches and executes the Claude Flow CLI from the npm registry.
[COMMAND_EXECUTION]: Lists Bash in allowed-tools and provides shell commands for querying the knowledge graph via the CLI.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes and synthesizes external knowledge graph data.
Ingestion points: Entity and edge data retrieved from the knowledge graph via agentdb_hierarchical-recall and agentdb_causal-edge (SKILL.md).
Boundary markers: None identified; results are synthesized into a coherent summary without explicit delimiters.
Capability inventory: Includes natural language synthesis via agentdb_context-synthesize and shell access via Bash (SKILL.md).
Sanitization: No validation or escaping of retrieved graph content is implemented prior to synthesis.

kg-traverse