inspect-ai 0.3.76__tar.gz → 0.3.78__tar.gz

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/CHANGELOG.md

@@ -1,3 +1,19 @@
+## v0.3.78 (25 March 2025)
+
+- OpenAI: Ensure that assistant messages always have the `msg_` prefix in responses API.
+
+## v0.3.77 (25 March 2025)
+
+- New [think()](https://inspect.aisi.org.uk/tools-standard.html#sec-think) tool that provides models with the ability to include an additional thinking step.
+- OpenAI: Support for the new [Responses API](https://inspect.ai-safety-institute.org.uk/providers.html#responses-api) and [o1-pro](https://platform.openai.com/docs/models/o1-pro) models.
+- OpenAI: Remove base64-encoded audio content from API call JSON in ModelEvent.
+- AzureAI: Support for use of native [OpenAI](https://inspect.ai-safety-institute.org.uk/providers.html#openai-on-azure) and [Mistral](https://inspect.ai-safety-institute.org.uk/providers.html#mistral-on-azure-ai) clients using service qualifiers (e.g. `openai/azure/gpt-4o-mini` or `mistral/azure/Mistral-Large-2411`). 
+- OpenRouter: Handle "error" field in response object and retry for empty responses.
+- Added `--metadata` option to eval for associating metadata with eval runs.
+- Task display: Show reasoning tokens for models that report them.
+- Anthropic: Include reasoning tokens in computation of total tokens
+- Inspect View: Properly wrap tool input for non-code inputs like `think`.
+
 ## v0.3.76 (23 March 2025)
 
 - [bash_session()](https://inspect.ai-safety-institute.org.uk/tools-standard.html#sec-bash-session) tool for creating a stateful bash shell that retains its state across calls from the model.

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: inspect_ai
-Version: 0.3.76
+Version: 0.3.78
 Summary: Framework for large language model evaluations
 Author: UK AI Security Institute
 License: MIT License

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/docs/_tools-standard.md

@@ -11,3 +11,5 @@ Inspect has several standard tools built-in, including:
 -   [Computer](tools-standard.qmd#sec-computer), which provides the model with a desktop computer (viewed through screenshots) that supports mouse and keyboard interaction.
 
 -   [Web Search](tools-standard.qmd#sec-web-search), which uses the Google Search API to execute and summarise web searches.
+
+-   [Think](tools-standard.qmd#sec-think), which provides models the ability to include an additional thinking step as part of getting to its final answer.

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/docs/options.qmd

@@ -172,4 +172,5 @@ Below are sections for the various categories of options supported by `inspect e
 | `--approval` | Config file for tool call approval. |
 | `--env` | Set an environment variable (multiple instances of `--env` are permitted). |
 | `--tags` | Tags to associate with this evaluation run. |
+| `--metadata` | Metadata to associate with this evaluation run (`key=value`)  |
 | `--help` | Display help for command options. |

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/docs/providers.qmd

@@ -29,17 +29,36 @@ The following environment variables are supported by the OpenAI provider
 | `OPENAI_ORG_ID` | OpenAI organization ID (optional) |
 | `OPENAI_PROJECT_ID` | OpenAI project ID (optional) |
 
+### Responses API
+
+By default, Inspect uses the standard OpenAI Chat Completions API. However, if you use features that require the new [Responses API](https://platform.openai.com/docs/api-reference/responses) Inspect will use that instead. 
+
+Currently, the only feature that requires the Responses API is [o1-pro](https://platform.openai.com/docs/models/o1-pro). In the near future, Inspect will add support for OpenAI's built-in computer and web search tools, which will also require the Responses API.
+
+If you want to manually switch to the Responses API you can use the `responses_api` model argument. For example:
+
+```bash
+inspect eval math.py --model openai/gpt-4o -M responses_api=true
+```
+
 ### OpenAI on Azure {#openai-on-azure}
 
-To use OpenAI models on Azure AI, specify an `AZUREAI_OPENAI_API_KEY` along with an `AZUREAI_OPENAI_BASE_URL`. You can then use the normal `openai` provider with the `azure` qualifier, specifying a model name that corresponds to the [Azure Deployment Name](https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/create-resource?pivots=web-portal#deploy-a-model) of your model. For example, if your deployed model name was `gpt4-1106-preview-ythre:`
+The `openai` provider supports OpenAI models deployed on the [Azure AI Foundary](https://ai.azure.com/). To use OpenAI models on Azure AI, specify the following environment variables:
+
+- `AZUREAI_OPENAI_API_KEY`
+- `AZUREAI_OPENAI_BASE_URL` 
+- `AZUREAI_OPENAI_API_VERSION`. 
+
+You can then use the normal `openai` provider with the `azure` qualifier and the name of your model deployment (e.g. `gpt-4o-mini`). For example:
 
 ``` bash
-export AZUREAI_OPENAI_API_KEY=key
+export AZUREAI_OPENAI_API_KEY=your-api-key
 export AZUREAI_OPENAI_BASE_URL=https://your-url-at.azure.com
-inspect eval --model openai/azure/gpt4-1106-preview-ythre
+export AZUREAI_OPENAI_API_VERSION=2025-02-01-preview
+inspect eval math.py --model openai/azure/gpt-4o-mini
 ```
 
-In addition to these variables, you can also set the `OPENAI_API_VERSION` environment variable to specify a specific version of the OpenAI interface on Azure.
+Note that if the `AZUREAI_OPENAI_API_VERSION` is not specified, Inspect will generally default to the latest deployed version, which as of this writing is `2025-02-01-preview`.
 
 ## Anthropic {#anthropic}
 
@@ -194,12 +213,17 @@ The following environment variables are supported by the Mistral provider
 
 ### Mistral on Azure AI {#mistral-on-azure-ai}
 
-To use Mistral models on Azure AI, specify an `AZURE_MISTRAL_API_KEY` along with an `AZUREAI_MISTRAL_BASE_URL`. You can then use the normal `mistral` provider, but you'll need to specify a model name that corresponds to the [Azure Deployment Name](https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/create-resource?pivots=web-portal#deploy-a-model) of your model. For example, if your deployment model name was `mistral-large-ctwi:`
+The `mistral` provider supports Mistral models deployed on the [Azure AI Foundary](https://ai.azure.com/). To use Mistral models on Azure AI, specify the following environment variables:
+
+- `AZURE_MISTRAL_API_KEY`
+- `AZUREAI_MISTRAL_BASE_URL`
+
+You can then use the normal `mistral` provider with the `azure` qualifier and the name of your model deployment (e.g. `Mistral-Large-2411`). For example:
 
 ``` bash
 export AZUREAI_MISTRAL_API_KEY=key
-export AZUREAI_MISTRAL_BASE_URL=https://your-url-at.azure.com
-inspect eval --model mistral/mistral-large-ctwi
+export AZUREAI_MISTRAL_BASE_URL=https://your-url-at.azure.com/models
+inspect eval math.py --model mistral/azure/Mistral-Large-2411
 ```
 
 ## DeepSeek {#deepseek}
@@ -252,13 +276,15 @@ If you are using Anthropic models on Bedrock, you can alternatively use the [Ant
 
 ## Azure AI {#azure-ai}
 
-To use the [Azure AI](https://azure.microsoft.com/en-us/solutions/ai) provider, install the `azure-ai-inference` package, set your credentials and base URL, and specify an [Azure Deployment Name](https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/create-resource?pivots=web-portal#deploy-a-model) as the model name:
+The `azureai` provider supports models deployed on the [Azure AI Foundary](https://ai.azure.com/).
+
+To use the `azureai` provider, install the `azure-ai-inference` package, set your credentials and base URL, and specify the name of the model you have deployed (e.g. `Llama-3.3-70B-Instruct`). For example:
 
 ``` bash
 pip install azure-ai-inference
 export AZUREAI_API_KEY=api-key
-export AZUREAI_BASE_URL=https://your-url-at.azure.com
-$ inspect eval --model azureai/llama-2-70b-chat-wnsnw
+export AZUREAI_BASE_URL=https://your-url-at.azure.com/models
+$ inspect eval math.py --model azureai/Llama-3.3-70B-Instruct
 ```
 
 For the `azureai` provider, custom model args (`-M`) are forwarded to the constructor of the `ChatCompletionsClient` class.

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/docs/reference/_sidebar.yml

@@ -89,6 +89,8 @@ website:
           href: reference/inspect_ai.tool.qmd#computer
         - text: web_search
           href: reference/inspect_ai.tool.qmd#web_search
+        - text: think
+          href: reference/inspect_ai.tool.qmd#think
         - text: tool_with
           href: reference/inspect_ai.tool.qmd#tool_with
         - text: ToolDef

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/docs/reference/inspect_ai.tool.qmd

@@ -11,6 +11,7 @@ title: "inspect_ai.tool"
 ### web_browser
 ### computer
 ### web_search
+### think
 
 ## Dynamic 
 

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/docs/tools-standard.qmd

@@ -404,4 +404,137 @@ The `web_search()` tool uses [Google Programmable Search Engine](https://program
 
 -   `GOOGLE_CSE_ID` — Google Custom Search Engine ID
 
--   `GOOGLE_CSE_API_KEY` — Google API key used to enable the Search API
+-   `GOOGLE_CSE_API_KEY` — Google API key used to enable the Search API
+
+
+## Think {#sec-think}
+
+The `think()` tool provides models with the ability to include an additional thinking step as part of getting to its final answer. 
+ 
+Note that the `think()` tool is not a substitute for reasoning and extended thinking, but rather an an alternate way of letting models express thinking that is better suited to some tool use scenarios. 
+
+### Usage
+
+You should read the original [think tool article](https://www.anthropic.com/engineering/claude-think-tool) in its entirely to understand where and where not to use the think tool. In summary, good contexts for the think tool include:
+
+1. Tool output analysis. When models need to carefully process the output of previous tool calls before acting and might need to backtrack in its approach;
+2. Policy-heavy environments. When models need to follow detailed guidelines and verify compliance; and
+3. Sequential decision making. When each action builds on previous ones and mistakes are costly (often found in multi-step domains).
+
+Use the `think()` tool alongside other tools like this:
+
+``` python
+from inspect_ai import Task, task
+from inspect_ai.scorer import includes
+from inspect_ai.solver import generate, system_message, use_tools
+from inspect_ai.tool import bash_session, text_editor, think
+
+@task
+def intercode_ctf():
+    return Task(
+        dataset=read_dataset(),
+        solver=[
+            system_message("system.txt"),
+            use_tools([
+                bash_session(timeout=180),
+                text_editor(timeout=180),
+                think()
+            ]),
+            generate(),
+        ],
+        scorer=includes(),
+        sandbox=("docker", "compose.yaml")
+    )
+```
+
+### Tool Description
+
+In the original [think tool article]((https://www.anthropic.com/engineering/claude-think-tool)) (which was based on experimenting with Claude) they found that providing clear instructions on when and how to use the `think()` tool for the particular problem domain it is being used within could sometimes be helpful. For example, here's the prompt they used with SWE-Bench:
+
+```python
+from textwrap import dedent
+
+from inspect_ai import Task, task
+from inspect_ai.scorer import includes
+from inspect_ai.solver import generate, system_message, use_tools
+from inspect_ai.tool import bash_session, text_editor, think
+
+@task
+def swe_bench():
+
+    tools = [
+        bash_session(timeout=180),
+        text_editor(timeout=180),  
+        think(dedent("""
+            Use the think tool to think about something. It will not obtain
+            new information or make any changes to the repository, but just 
+            log the thought. Use it when complex reasoning or brainstorming
+            is needed. For example, if you explore the repo and discover
+            the source of a bug, call this tool to brainstorm several unique
+            ways of fixing the bug, and assess which change(s) are likely to 
+            be simplest and most effective. Alternatively, if you receive
+            some test results, call this tool to brainstorm ways to fix the
+            failing tests.
+        """))
+    ])
+
+    return Task(
+        dataset=read_dataset(),
+        solver=[
+            system_message("system.txt"),
+            use_tools(tools),
+            generate(),
+        ),
+        scorer=includes(),
+        sandbox=("docker", "compose.yaml")
+    )
+```
+
+### System Prompt
+
+In the article they also found that when tool instructions are long and/or complex, including instructions about the `think()` tool in the system prompt can be more effective than placing them in the tool description itself. 
+
+Here's an example of moving the custom `think()` prompt into the system prompt (note that this was *not* done in the article's SWE-Bench experiment, this is merely an example):
+
+```python
+from textwrap import dedent
+
+from inspect_ai import Task, task
+from inspect_ai.scorer import includes
+from inspect_ai.solver import generate, system_message, use_tools
+from inspect_ai.tool import bash_session, text_editor, think
+
+@task
+def swe_bench():
+
+    think_system_message = system_message(dedent("""
+        Use the think tool to think about something. It will not obtain
+        new information or make any changes to the repository, but just 
+        log the thought. Use it when complex reasoning or brainstorming
+        is needed. For example, if you explore the repo and discover
+        the source of a bug, call this tool to brainstorm several unique
+        ways of fixing the bug, and assess which change(s) are likely to 
+        be simplest and most effective. Alternatively, if you receive
+        some test results, call this tool to brainstorm ways to fix the
+        failing tests.
+    """))
+
+    return Task(
+        dataset=read_dataset(),
+        solver=[
+            system_message("system.txt"),
+            think_system_message,
+            use_tools([
+                bash_session(timeout=180),
+                text_editor(timeout=180),  
+                think(),
+            ]),
+            generate(),
+        ],
+        scorer=includes(),
+        sandbox=("docker", "compose.yaml")
+    )
+```
+
+Note that the effectivess of using the system prompt will vary considerably across tasks, tools, and models, so should definitely be the subject of experimentation.
+

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/src/inspect_ai/_cli/eval.py

@@ -115,6 +115,13 @@ def eval_options(func: Callable[..., Any]) -> Callable[..., click.Context]:
         help="Tags to associate with this evaluation run.",
         envvar="INSPECT_EVAL_TAGS",
     )
+    @click.option(
+        "--metadata",
+        multiple=True,
+        type=str,
+        help="Metadata to associate with this evaluation run (more than one --metadata argument can be specified).",
+        envvar="INSPECT_EVAL_METADATA",
+    )
     @click.option(
         "--trace",
         type=bool,
@@ -449,6 +456,7 @@ def eval_command(
     s: tuple[str] | None,
     solver_config: str | None,
     tags: str | None,
+    metadata: tuple[str] | None,
     trace: bool | None,
     approval: str | None,
     sandbox: str | None,
@@ -525,6 +533,7 @@ def eval_command(
         s=s,
         solver_config=solver_config,
         tags=tags,
+        metadata=metadata,
         trace=trace,
         approval=approval,
         sandbox=sandbox,
@@ -616,6 +625,7 @@ def eval_set_command(
     s: tuple[str] | None,
     solver_config: str | None,
     tags: str | None,
+    metadata: tuple[str] | None,
     sandbox: str | None,
     no_sandbox_cleanup: bool | None,
     epochs: int | None,
@@ -695,6 +705,7 @@ def eval_set_command(
         s=s,
         solver_config=solver_config,
         tags=tags,
+        metadata=metadata,
         trace=trace,
         approval=approval,
         sandbox=sandbox,
@@ -749,6 +760,7 @@ def eval_exec(
     s: tuple[str] | None,
     solver_config: str | None,
     tags: str | None,
+    metadata: tuple[str] | None,
     trace: bool | None,
     approval: str | None,
     sandbox: str | None,
@@ -790,6 +802,9 @@ def eval_exec(
     # parse tags
     eval_tags = parse_comma_separated(tags)
 
+    # parse metadata
+    eval_metadata = parse_cli_args(metadata)
+
     # resolve epochs
     eval_epochs = (
         Epochs(epochs, create_reducers(parse_comma_separated(epochs_reducer)))
@@ -825,6 +840,7 @@ def eval_exec(
             task_args=task_args,
             solver=SolverSpec(solver, solver_args) if solver else None,
             tags=eval_tags,
+            metadata=eval_metadata,
             trace=trace,
             approval=approval,
             sandbox=parse_sandbox(sandbox),

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/src/inspect_ai/_display/core/results.py

@@ -131,9 +131,14 @@ def task_stats(stats: EvalStats) -> RenderableType:
         else:
             input_tokens = f"[bold]I: [/bold]{usage.input_tokens:,}"
 
+        if usage.reasoning_tokens is not None:
+            reasoning_tokens = f", [bold]R: [/bold]{usage.reasoning_tokens:,}"
+        else:
+            reasoning_tokens = ""
+
         table.add_row(
             Text(model, style="bold"),
-            f"  {usage.total_tokens:,} tokens [{input_tokens}, [bold]O: [/bold]{usage.output_tokens:,}]",
+            f"  {usage.total_tokens:,} tokens [{input_tokens}, [bold]O: [/bold]{usage.output_tokens:,}{reasoning_tokens}]",
             style=theme.light,
         )
 

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/src/inspect_ai/_eval/eval.py

@@ -68,6 +68,7 @@ def eval(
     sandbox_cleanup: bool | None = None,
     solver: Solver | list[Solver] | SolverSpec | None = None,
     tags: list[str] | None = None,
+    metadata: dict[str, Any] | None = None,
     trace: bool | None = None,
     display: DisplayType | None = None,
     approval: str | list[ApprovalPolicy] | None = None,
@@ -116,6 +117,7 @@ def eval(
         solver: Alternative solver for task(s).
             Optional (uses task solver by default).
         tags: Tags to associate with this evaluation run.
+        metadata: Metadata to associate with this evaluation run.
         trace: Trace message interactions with evaluated model to terminal.
         display: Task display type (defaults to 'full').
         approval: Tool use approval policies.
@@ -186,6 +188,7 @@ def eval(
                 sandbox_cleanup=sandbox_cleanup,
                 solver=solver,
                 tags=tags,
+                metadata=metadata,
                 approval=approval,
                 log_level=log_level,
                 log_level_transcript=log_level_transcript,
@@ -235,6 +238,7 @@ async def eval_async(
     sandbox_cleanup: bool | None = None,
     solver: Solver | list[Solver] | SolverSpec | None = None,
     tags: list[str] | None = None,
+    metadata: dict[str, Any] | None = None,
     approval: str | list[ApprovalPolicy] | ApprovalPolicyConfig | None = None,
     log_level: str | None = None,
     log_level_transcript: str | None = None,
@@ -274,7 +278,8 @@ async def eval_async(
         sandbox: Sandbox environment type (or optionally a str or tuple with a shorthand spec)
         sandbox_cleanup: Cleanup sandbox environments after task completes (defaults to True)
         solver: Alternative solver for task(s).  Optional (uses task solver by default).
-        tags (list[str] | None): Tags to associate with this evaluation run.
+        tags: Tags to associate with this evaluation run.
+        metadata: Metadata to associate with this evaluation run.
         approval: Tool use approval policies.
           Either a path to an approval policy config file or a list of approval policies.
           Defaults to no approval policy.
@@ -449,6 +454,7 @@ async def eval_async(
                         epochs_reducer=epochs_reducer,
                         solver=solver,
                         tags=tags,
+                        metadata=metadata,
                         score=score,
                         debug_errors=debug_errors is True,
                         **kwargs,
@@ -473,6 +479,7 @@ async def eval_async(
                 epochs_reducer=epochs_reducer,
                 solver=solver,
                 tags=tags,
+                metadata=metadata,
                 score=score,
                 **kwargs,
             )

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/src/inspect_ai/_eval/evalset.py

@@ -68,6 +68,7 @@ def eval_set(
     sandbox_cleanup: bool | None = None,
     solver: Solver | list[Solver] | SolverSpec | None = None,
     tags: list[str] | None = None,
+    metadata: dict[str, Any] | None = None,
     trace: bool | None = None,
     display: DisplayType | None = None,
     approval: str | list[ApprovalPolicy] | None = None,
@@ -127,6 +128,7 @@ def eval_set(
         solver: Alternative solver(s) for
             evaluating task(s). ptional (uses task solver by default).
         tags: Tags to associate with this evaluation run.
+        metadata: Metadata to associate with this evaluation run.
         trace: Trace message interactions with evaluated model to terminal.
         display: Task display type (defaults to 'full').
         approval: Tool use approval policies.
@@ -193,6 +195,7 @@ def eval_set(
             sandbox_cleanup=sandbox_cleanup,
             solver=solver,
             tags=tags,
+            metadata=metadata,
             trace=trace,
             display=display,
             approval=approval,

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/src/inspect_ai/_eval/run.py

@@ -2,7 +2,7 @@ import functools
 import logging
 import os
 import sys
-from typing import Awaitable, Callable, Set, cast
+from typing import Any, Awaitable, Callable, Set, cast
 
 from inspect_ai._eval.task.task import Task
 from inspect_ai._util.trace import trace_action
@@ -68,6 +68,7 @@ async def eval_run(
     epochs_reducer: list[ScoreReducer] | None = None,
     solver: Solver | SolverSpec | None = None,
     tags: list[str] | None = None,
+    metadata: dict[str, Any] | None = None,
     debug_errors: bool = False,
     score: bool = True,
     **kwargs: Unpack[GenerateConfigArgs],
@@ -205,7 +206,7 @@ async def eval_run(
                     task_args=resolved_task.task_args,
                     model_args=resolved_task.model.model_args,
                     eval_config=task_eval_config,
-                    metadata=task.metadata,
+                    metadata=((metadata or {}) | (task.metadata or {})) or None,
                     recorder=recorder,
                 )
                 await logger.init()

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/src/inspect_ai/_util/content.py

@@ -12,6 +12,9 @@ class ContentText(BaseModel):
     text: str
     """Text content."""
 
+    refusal: bool | None = Field(default=None)
+    """Was this a refusal message?"""
+
 
 class ContentReasoning(BaseModel):
     """Reasoning content.

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/src/inspect_ai/_view/www/dist/assets/index.js

@@ -21577,7 +21577,7 @@ var require_assets = __commonJS({
           className: clsx(
             "source-code",
             "sourceCode",
-            `language-${highlightLanguage}`,
+            highlightLanguage ? `language-${highlightLanguage}` : void 0,
             styles$10.outputCode
           ),
           children: formattedContent
@@ -21613,6 +21613,22 @@ var require_assets = __commonJS({
       }
       const collapse = Array.isArray(output2) ? output2.every((item2) => !isContentImage(item2)) : !isContentImage(output2);
       const normalizedContent = reactExports.useMemo(() => normalizeContent$1(output2), [output2]);
+      const hasContent = normalizedContent.find((c2) => {
+        if (c2.type === "tool") {
+          for (const t2 of c2.content) {
+            if (t2.type === "text") {
+              if (t2.text) {
+                return true;
+              }
+            } else {
+              return true;
+            }
+          }
+          return false;
+        } else {
+          return true;
+        }
+      });
       const contents2 = mode !== "compact" ? input2 : input2 || functionCall;
       return /* @__PURE__ */ jsxRuntimeExports.jsxs("div", { children: [
         mode !== "compact" && (!view || view.title) ? /* @__PURE__ */ jsxRuntimeExports.jsx(ToolTitle, { title: (view == null ? void 0 : view.title) || functionCall }) : "",
@@ -21625,7 +21641,7 @@ var require_assets = __commonJS({
               toolCallView: view
             }
           ),
-          /* @__PURE__ */ jsxRuntimeExports.jsx(ExpandablePanel, { collapse, border: true, lines: 15, children: /* @__PURE__ */ jsxRuntimeExports.jsx(MessageContent, { contents: normalizedContent }) })
+          hasContent ? /* @__PURE__ */ jsxRuntimeExports.jsx(ExpandablePanel, { collapse, border: true, lines: 15, children: /* @__PURE__ */ jsxRuntimeExports.jsx(MessageContent, { contents: normalizedContent }) }) : void 0
         ] }) })
       ] });
     };

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/src/inspect_ai/_view/www/src/samples/chat/tools/ToolCallView.tsx

@@ -83,8 +83,24 @@ export const ToolCallView: FC<ToolCallViewProps> = ({
     : !isContentImage(output);
   const normalizedContent = useMemo(() => normalizeContent(output), [output]);
 
-  const contents = mode !== "compact" ? input : input || functionCall;
+  const hasContent = normalizedContent.find((c) => {
+    if (c.type === "tool") {
+      for (const t of c.content) {
+        if (t.type === "text") {
+          if (t.text) {
+            return true;
+          }
+        } else {
+          return true;
+        }
+      }
+      return false;
+    } else {
+      return true;
+    }
+  });
 
+  const contents = mode !== "compact" ? input : input || functionCall;
   return (
     <div>
       {mode !== "compact" && (!view || view.title) ? (
@@ -99,9 +115,11 @@ export const ToolCallView: FC<ToolCallViewProps> = ({
             contents={contents}
             toolCallView={view}
           />
-          <ExpandablePanel collapse={collapse} border={true} lines={15}>
-            <MessageContent contents={normalizedContent} />
-          </ExpandablePanel>
+          {hasContent ? (
+            <ExpandablePanel collapse={collapse} border={true} lines={15}>
+              <MessageContent contents={normalizedContent} />
+            </ExpandablePanel>
+          ) : undefined}
         </div>
       </div>
     </div>

{inspect_ai-0.3.76 → inspect_ai-0.3.78}/src/inspect_ai/_view/www/src/samples/chat/tools/ToolInput.tsx

@@ -64,7 +64,7 @@ export const ToolInput: FC<ToolInputProps> = memo((props) => {
         className={clsx(
           "source-code",
           "sourceCode",
-          `language-${highlightLanguage}`,
+          highlightLanguage ? `language-${highlightLanguage}` : undefined,
           styles.outputCode,
         )}
       >