The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script {project-root}/_bmad/scripts/resolve_customization.py during the activation and completion phases to handle configuration merging and workflow customization. This script is part of the vendor's own framework located in the user's project directory.
[PROMPT_INJECTION]: The skill uses dynamic hooks such as activation_steps_prepend, activation_steps_append, and on_complete that execute instructions from local customization files (customize.toml and its overrides). While this allows for extensibility, it also enables the execution of arbitrary instructions defined in these configuration files.
[PROMPT_INJECTION]: The skill is exposed to potential indirect prompt injection due to its core function of retrieving and processing untrusted data from the web.
Ingestion points: Market research and industry data retrieved via web searches across files step-02-domain-analysis.md through step-06-research-synthesis.md.
Boundary markers: Absent; the skill does not use specific delimiters or instructions to isolate external content from the agent's core logic.
Capability inventory: The agent can perform web searches, write to the local file system (creating research reports), and execute subagents or subprocesses as described in the industry analysis step.
Sanitization: None; the skill relies on the underlying model's processing to synthesize information from various sources.
[DATA_EXFILTRATION]: The skill reads project-specific context from all project-context.md files within the project root. While this data is used to tailor the research, it results in project documentation being ingested into the agent's context, which could potentially be included in outbound web search queries.

bmad-domain-research