The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the openbkn command-line interface, npm, and git to perform environment checks, manage platform resources, and install the @openbkn/bkn-sdk package. These operations are performed at various stages of the creation and validation pipelines.
[INDIRECT_PROMPT_INJECTION]: The skill defines a large surface area for processing untrusted data, specifically when extracting object-relation schemas and business rules from documents provided by users. 1. Ingestion points: internal/bkn-extract/SKILL.md (processes source_text) and internal/_plugins/bkn-rules/SKILL.md (processes prd_content). 2. Boundary markers: The instructions rely on role-based context to differentiate between document content and extraction logic, but lack formal delimiters or sanitization protocols to prevent the agent from following instructions embedded within the business documents. 3. Capability inventory: The agent has high-privilege access to the OpenBKN platform via the CLI, including the ability to push new network definitions, create agents, and query data views.
[PERSISTENCE]: The internal/_plugins/bkn-distribute/SKILL.md plugin is designed to write generated skill files to local AI agent configuration paths, specifically .cursor/skills/, .claude/skills/, and .openclaw/workspace/skills/. While functional for its stated purpose of distributing business rules, this mechanism provides a way to persist instructions across agent environments.

bkn-creator