The Agent Skills Directory

[SAFE]: The skill consists exclusively of instructional guidelines in markdown format. It contains no code, scripts, or direct system calls that could be leveraged for malicious purposes.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external research data and user-supplied context, which represents a potential surface for indirect prompt injection. However, the instructions incorporate strong mitigations through evidence labeling.
Ingestion points: The workflow ingests external information from search tools and user context as described in SKILL.md.
Boundary markers: Present. The 'Guardrails' and 'Workflow' sections of SKILL.md explicitly instruct the agent to separate 'sourced facts' from 'user-supplied context', 'inference', and 'recommendations'.
Capability inventory: The skill coordinates several research-focused modules (exa-search, deep-research, market-research, lead-intelligence, knowledge-ops) to synthesize data.
Sanitization: The skill relies on instructional normalization and structured output formatting rather than programmatic filtering of the ingested content.

research-ops