Bmad Brainstorming Coach
Analysis
This is mostly a brainstorming assistant, but one included workflow contradicts the self-contained claim by pointing to unreviewed external BMad files that could change how the agent behaves.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
methods="{project-root}/_bmad/core/workflows/advanced-elicitation/methods.csv" agent-party="{project-root}/_bmad/_config/agent-manifest.csv" ... <action>Load and read {{methods}} and {{agent-party}}</action>This workflow tells the agent to load method and agent-party data from external `_bmad` paths that are not included in the packaged artifacts, so behavior can depend on unreviewed local files.
本 Skill 的所有逻辑、模板和引擎均已包含在本地 `references/` 和 `assets/` 目录下,严禁引用外部目录(如 `_bmad`)。
The skill assures users that all logic is bundled locally and that external `_bmad` directories are forbidden, but an included workflow still references `_bmad` method and agent-manifest files.
每遇到 `template-output` 标签,必须生成内容并保存到 `{output_folder}` 下 ... `[y] YOLO`: 自动完成当前文档。The skill is designed to write generated brainstorming output to a user-provided folder and can continue document generation automatically when the user selects YOLO.
Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.
Check if `context_file` is provided ... If context file exists and is readable, load it ... Update the document frontmatter: `session_topic`, `session_goals`, `ideas_generated`, `context_file`
The workflow can read a user-provided context file and persist brainstorming state in the output document for later continuation.
