The Agent Skills Directory

[NO_CODE]: The skill is composed entirely of Markdown documentation, references, and examples. It does not include any executable scripts (Python, Node.js, Shell) or binary files.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow that ingests untrusted external data (Form 1095-A enrollment data and household income facts) which could potentially contain injection attempts. However, this is inherent to the skill's primary purpose. The risk is mitigated by a mandatory 'Validation' section that requires the agent to perform strict mathematical checks, input verification against legal standards, and cross-form consistency checks before finalizing the output.
Ingestion points: Data extracted from Form 1095-A Column A, B, and C as described in SKILL.md (Step 7).
Boundary markers: Not explicitly defined for the data input values.
Capability inventory: The skill describes the use of browser automation tools (Playwright, Selenium) for filing at official domains, as detailed in filing.md.
Sanitization: Extensively covered in the 'Validation' and 'Math checks' sections of SKILL.md to ensure data integrity.
[SAFE]: All external URLs referenced in the skill point to official United States government domains (e.g., irs.gov, hhs.gov, healthcare.gov) for forms, instructions, and poverty guidelines. The skill does not contain hardcoded credentials, exfiltration patterns, or persistence mechanisms.

form-8962