Letting an Agent Discover Tools at Runtime (and the Risks)

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-agentic-agent-tool-discovery-runtime-creators

1. What is the core idea behind "Letting an Agent Discover Tools at Runtime (and the Risks)"?

   1. Patterns for runtime tool registration vs. static registries — and why runtime is harder than it looks.
   2. prediction
   3. Agent drafts a Depop listing with hashtags and measurements.
   4. Constrain tool permissions so even successful injection has limited blast radius

2. Which term best describes a foundational idea in "Letting an Agent Discover Tools at Runtime (and the Risks)"?

   1. tool-registry
   2. dynamic-tools
   3. MCP
   4. runtime-discovery

3. A learner studying Letting an Agent Discover Tools at Runtime (and the Risks) would need to understand which concept?

   1. dynamic-tools
   2. MCP
   3. tool-registry
   4. runtime-discovery

4. Which of these is directly relevant to Letting an Agent Discover Tools at Runtime (and the Risks)?

   1. dynamic-tools
   2. tool-registry
   3. runtime-discovery
   4. MCP

5. Which of the following is a key point about Letting an Agent Discover Tools at Runtime (and the Risks)?

   1. Pull a fresh tool catalog from MCP or a registry per session
   2. Filter tools by user scope before they enter the prompt
   3. Cache the catalog within a session to avoid re-fetch cost
   4. Log which tools were exposed for every run

6. Which of these does NOT belong in a discussion of Letting an Agent Discover Tools at Runtime (and the Risks)?

   1. Cache the catalog within a session to avoid re-fetch cost
   2. Pull a fresh tool catalog from MCP or a registry per session
   3. prediction
   4. Filter tools by user scope before they enter the prompt

7. Which statement is accurate regarding Letting an Agent Discover Tools at Runtime (and the Risks)?

   1. Predict token cost when the catalog grows unboundedly
   2. Detect a malicious tool definition by description alone
   3. Guarantee a freshly registered tool is safe to call without review
   4. prediction

8. What is the key insight about "Catalog filter rule" in the context of Letting an Agent Discover Tools at Runtime (and the Risks)?

   1. prediction
   2. Agent drafts a Depop listing with hashtags and measurements.
   3. Constrain tool permissions so even successful injection has limited blast radius
   4. Before the agent sees the catalog, filter by user permissions and a deny-list of recently-flagged tools.

9. What is the key insight about "Every new tool is a new vulnerability" in the context of Letting an Agent Discover Tools at Runtime (and the Risks)?

   1. A runtime-discovered tool you have not reviewed is equivalent to running unreviewed code with your credentials.
   2. prediction
   3. Agent drafts a Depop listing with hashtags and measurements.
   4. Constrain tool permissions so even successful injection has limited blast radius

10. Which statement accurately describes an aspect of Letting an Agent Discover Tools at Runtime (and the Risks)?

    1. prediction
    2. Runtime tool discovery sounds elegant but expands the attack surface and the cost surface of every agent invocation.
    3. Agent drafts a Depop listing with hashtags and measurements.
    4. Constrain tool permissions so even successful injection has limited blast radius

11. Which best describes the scope of "Letting an Agent Discover Tools at Runtime (and the Risks)"?

    1. It is unrelated to agentic workflows
    2. It applies only to the opposite beginner tier
    3. It focuses on Patterns for runtime tool registration vs. static registries — and why runtime is harder than it loo
    4. It was deprecated in 2024 and no longer relevant

12. Which section heading best belongs in a lesson about Letting an Agent Discover Tools at Runtime (and the Risks)?

    1. prediction
    2. Agent drafts a Depop listing with hashtags and measurements.
    3. Constrain tool permissions so even successful injection has limited blast radius
    4. What AI does well here

13. Which section heading best belongs in a lesson about Letting an Agent Discover Tools at Runtime (and the Risks)?

    1. What AI cannot do
    2. prediction
    3. Agent drafts a Depop listing with hashtags and measurements.
    4. Constrain tool permissions so even successful injection has limited blast radius

14. Which of the following is a concept covered in Letting an Agent Discover Tools at Runtime (and the Risks)?

    1. tool-registry
    2. dynamic-tools
    3. MCP
    4. runtime-discovery

15. Which of the following is a concept covered in Letting an Agent Discover Tools at Runtime (and the Risks)?

    1. dynamic-tools
    2. MCP
    3. tool-registry
    4. runtime-discovery