The Agent Skills Directory

[SAFE]: The skill is a legitimate technical writing assistant. It contains no obfuscated code, hidden instructions, or attempts at privilege escalation. It explicitly forbids the fabrication of data or citations and provides clear handoff rules to other specialized tools.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as funding calls and project concepts, which represents an indirect prompt injection attack surface. However, the risk is negligible as the instructions focus on structured document generation and evidence-based writing rather than executing commands from the data.
Ingestion points: External data from funding calls (RFP) and project concepts are ingested during the Phase 1 and Phase 2 workflows described in SKILL.md.
Boundary markers: No specific delimiters (like XML tags or markers) are instructed for the external content.
Capability inventory: The skill is configured with access to Read, Write, Edit, Bash, Grep, and Glob tools.
Sanitization: No specific sanitization or escaping of external content is defined; the skill relies on the agent's internal logic and explicit instructions to maintain evidence-based claims.

grant-builder