The Agent Skills Directory

[SAFE]: The skill is designed to process external financial data and web content, which inherently introduces a surface for indirect prompt injection. However, the skill implements robust defensive measures by explicitly instructing the agent to treat all external content as untrusted evidence and to ignore any embedded commands, role changes, or tool-use requests.
Ingestion points: As noted in SKILL.md, the skill reads data from user-provided financial information, uploaded filings, and public web sources through research commands.
Boundary markers: The SKILL.md file contains a 'Data Rules' section that specifically directs the agent to ignore instructions found within source content.
Capability inventory: The skill uses the file_write capability exclusively to save generated reports in the finance/ directory as specified in SKILL.md and the YAML frontmatter.
Sanitization: The agent is instructed to use external content strictly for factual grounding and citations, ensuring it does not influence the agent's logic or behavior.

dcf-valuation