The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts using the uv run command within a Bash shell to interact with MCP tools.
Evidence includes calls to scripts/mcp/nia_docs.py, scripts/mcp/perplexity_search.py, and scripts/mcp/firecrawl_scrape.py to process research queries.
[EXTERNAL_DOWNLOADS]: The skill is designed to fetch and process data from external sources.
It interacts with established package registries including npm, PyPI, crates.io, and Go modules via the nia-docs tool.
It performs web scraping of arbitrary URLs found during the research process using the firecrawl tool.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection due to its core function of processing untrusted web data.
Ingestion points: Data enters the context via firecrawl_scrape.py (scraped web pages), perplexity_search.py (web search results), and nia_docs.py (external package documentation).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore potentially malicious instructions embedded within the scraped research content during the synthesis phase.
Capability inventory: The skill has Bash access to execute scripts and Write access to generate research documents and implementation handoffs.
Sanitization: The skill lacks explicit sanitization or filtering of the external content before interpolating it into the handoff output, which is specifically intended to guide a implementation agent (plan-agent).

research-external