AWS bedrock-agentcore documentation change
Summary
Updated typography (straight apostrophes to curly apostrophes) and changed one heading from markdown level 6 to bold text.
Security assessment
The changes are purely typographical and formatting updates. There is no indication of a security issue being addressed. The document already contained security-related evaluators (like Harmfulness and Stereotyping) but the changes do not alter the security content.
Diff
diff --git a/bedrock-agentcore/latest/devguide/prompt-templates-builtin.md b/bedrock-agentcore/latest/devguide/prompt-templates-builtin.md index 3dcf0e89a..762bbe556 100644 --- a//bedrock-agentcore/latest/devguide/prompt-templates-builtin.md +++ b//bedrock-agentcore/latest/devguide/prompt-templates-builtin.md @@ -19 +19 @@ Details on the placeholder values used by our current evaluators: - * `context` – All information from previous turns, including user prompts, tool calls, and assistant responses, plus the current turn's user prompt and tool call. + * `context` – All information from previous turns, including user prompts, tool calls, and assistant responses, plus the current turn’s user prompt and tool call. @@ -27 +27 @@ Details on the placeholder values used by our current evaluators: - * `context` – All information from previous turns (user prompts, tool call details, assistant responses) plus the current turn's user prompt and any tool calls made before the tool call being evaluated. + * `context` – All information from previous turns (user prompts, tool call details, assistant responses) plus the current turn’s user prompt and any tool calls made before the tool call being evaluated. @@ -34 +34 @@ Details on the placeholder values used by our current evaluators: -###### Topics +**Topics** @@ -67 +67 @@ Details on the placeholder values used by our current evaluators: -The Goal success rate evaluator assesses whether an AI assistant successfully completed all user goals within a conversation session. This session-level evaluator analyzes the entire conversation to determine if the user's objectives were met. +The Goal success rate evaluator assesses whether an AI assistant successfully completed all user goals within a conversation session. This session-level evaluator analyzes the entire conversation to determine if the user’s objectives were met. @@ -105 +105 @@ The Goal success rate evaluator assesses whether an AI assistant successfully co -The Coherence evaluator assesses the logical consistency and cohesion of an AI assistant's response. This trace-level evaluator examines whether the response maintains internal consistency without contradictions or logical gaps. +The Coherence evaluator assesses the logical consistency and cohesion of an AI assistant’s response. This trace-level evaluator examines whether the response maintains internal consistency without contradictions or logical gaps. @@ -215 +215 @@ The Context relevance evaluator assesses whether the provided context contains t -The Correctness evaluator assesses the factual accuracy and correctness of an AI assistant's response to a given task. This trace-level evaluator focuses on whether the content and solution are accurate, regardless of style or presentation. +The Correctness evaluator assesses the factual accuracy and correctness of an AI assistant’s response to a given task. This trace-level evaluator focuses on whether the content and solution are accurate, regardless of style or presentation. @@ -240 +240 @@ The Correctness evaluator assesses the factual accuracy and correctness of an AI -The Faithfulness evaluator assesses whether an AI assistant's response remains consistent with the conversation history. This trace-level evaluator identifies conflicts between current responses and previous interactions within the same conversation. +The Faithfulness evaluator assesses whether an AI assistant’s response remains consistent with the conversation history. This trace-level evaluator identifies conflicts between current responses and previous interactions within the same conversation. @@ -285 +285 @@ The Faithfulness evaluator assesses whether an AI assistant's response remains c -The Harmfulness evaluator detects potentially harmful content in an AI assistant's response. This trace-level evaluator identifies harmful information including insults, hate speech, violence, inappropriate sexual content, and stereotyping. +The Harmfulness evaluator detects potentially harmful content in an AI assistant’s response. This trace-level evaluator identifies harmful information including insults, hate speech, violence, inappropriate sexual content, and stereotyping. @@ -319 +319 @@ The Harmfulness evaluator detects potentially harmful content in an AI assistant -The Helpfulness evaluator assesses how effectively an AI assistant's response helps users progress toward their goals. This trace-level evaluator focuses on the user's perspective and whether the response moves them closer to achieving their objectives. +The Helpfulness evaluator assesses how effectively an AI assistant’s response helps users progress toward their goals. This trace-level evaluator focuses on the user’s perspective and whether the response moves them closer to achieving their objectives. @@ -384 +384 @@ The Helpfulness evaluator assesses how effectively an AI assistant's response he -The Instruction following evaluator assesses whether an AI assistant's response adheres to all explicit instructions provided in the user's input. This trace-level evaluator focuses on compliance with specific directives regardless of overall response quality. +The Instruction following evaluator assesses whether an AI assistant’s response adheres to all explicit instructions provided in the user’s input. This trace-level evaluator focuses on compliance with specific directives regardless of overall response quality. @@ -437 +437 @@ The Instruction following evaluator assesses whether an AI assistant's response -The Refusal evaluator detects when an AI assistant declines to address or fulfill a user's request. This trace-level evaluator identifies various forms of refusal, from direct declines to indirect avoidance of the requested task. +The Refusal evaluator detects when an AI assistant declines to address or fulfill a user’s request. This trace-level evaluator identifies various forms of refusal, from direct declines to indirect avoidance of the requested task. @@ -466 +466 @@ The Refusal evaluator detects when an AI assistant declines to address or fulfil -The Response relevance evaluator assesses how well an AI assistant's response addresses the specific question or request. This trace-level evaluator measures the focus and relevance of the response to the given input. +The Response relevance evaluator assesses how well an AI assistant’s response addresses the specific question or request. This trace-level evaluator measures the focus and relevance of the response to the given input. @@ -504 +504 @@ The Response relevance evaluator assesses how well an AI assistant's response ad -The Stereotyping evaluator detects bias and stereotypical content in an AI assistant's response. This trace-level evaluator identifies responses that contain prejudicial assumptions or generalizations about specific groups of people. +The Stereotyping evaluator detects bias and stereotypical content in an AI assistant’s response. This trace-level evaluator identifies responses that contain prejudicial assumptions or generalizations about specific groups of people.