The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied token names to resolve entities and fetch metrics across several DefiLlama tools, presenting a surface for indirect prompt injection. \n- Ingestion points: User-provided token names are passed to the defillama:resolve_entity tool as described in Step 1 of the skill workflow. \n- Boundary markers: There are no explicit delimiters or instructions provided in the skill body to warn the agent to ignore instructions embedded within the user-provided content. \n- Capability inventory: The skill utilizes a suite of data-fetching tools (defillama:resolve_entity, get_token_prices, get_token_unlocks, get_token_tvl, get_yield_pools, and get_protocol_metrics) to retrieve information from external sources. \n- Sanitization: The skill logic does not define any input validation or sanitization steps before the user input is interpolated into tool calls.

token-research