The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple external sources including Reddit, YouTube, Product Hunt, and various competitor websites via WebSearch, WebFetch, and Playwright tools.
Ingestion points: Steps 5, 6, and 7 in SKILL.md involve fetching content from external URLs.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore potential commands within the fetched content.
Capability inventory: The agent has access to Write, Edit, and Bash tools, which could be targeted if the agent is manipulated by injected instructions.
Sanitization: No explicit sanitization or filtering of the retrieved web content is specified before the agent processes and summarizes it into research.md.
[DYNAMIC_CONTEXT_INJECTION]: The skill uses dynamic context placeholders to retrieve local environment information when the skill is initialized.
Evidence: SKILL.md includes !git branch --show-current 2/dev/null and !git log --oneline -5 2/dev/null to provide branch and history context.
[COMMAND_EXECUTION]: The skill provides and instructs the execution of bash scripts to perform multi-tier domain availability checks.
Evidence: references/domain-check.md contains scripts utilizing whois, dig, and curl to interact with registration databases and DNS.
[EXTERNAL_DOWNLOADS]: The skill recommends installing external dependencies and mentions a specific GitHub repository for self-hosted search infrastructure.
Evidence: SKILL.md suggests pip install praw for Reddit interaction and references github.com/fortunto2/searxng-docker-tavily-adapter (a vendor-associated resource).

solo-research