diff --git a/cli/aiconfig-editor/src/components/EditorContainer.tsx b/cli/aiconfig-editor/src/components/EditorContainer.tsx
index 98de73afe..af2ce28a8 100644
--- a/cli/aiconfig-editor/src/components/EditorContainer.tsx
+++ b/cli/aiconfig-editor/src/components/EditorContainer.tsx
@@ -47,6 +47,8 @@ export default function EditorContainer({
     [aiconfig]
   );
 
+  // TODO: Implement editor context for callbacks, readonly state, etc.
+
   return (
     <>
       <Container>
diff --git a/cli/aiconfig-editor/src/components/SettingsPropertyRenderer.tsx b/cli/aiconfig-editor/src/components/SettingsPropertyRenderer.tsx
new file mode 100644
index 000000000..55d206c5c
--- /dev/null
+++ b/cli/aiconfig-editor/src/components/SettingsPropertyRenderer.tsx
@@ -0,0 +1,25 @@
+import { Flex } from "@mantine/core";
+import { memo } from "react";
+
+type Props = {
+  propertyName: string;
+  property: { [key: string]: any };
+  isRequired?: boolean;
+  initialValue: any;
+};
+
+export default memo(function SettingsPropertyRenderer({
+  propertyName,
+  property,
+  isRequired = false,
+  initialValue,
+}: Props) {
+  return (
+    <Flex direction="column">
+      <div>{propertyName}</div>
+      <div>{JSON.stringify(property)}</div>
+      <div>isRequired: {JSON.stringify(isRequired)}</div>
+      <div>initialValue: {JSON.stringify(initialValue)}</div>
+    </Flex>
+  );
+});
diff --git a/cli/aiconfig-editor/src/components/prompt/PromptActionBar.tsx b/cli/aiconfig-editor/src/components/prompt/PromptActionBar.tsx
new file mode 100644
index 000000000..7767ab6f1
--- /dev/null
+++ b/cli/aiconfig-editor/src/components/prompt/PromptActionBar.tsx
@@ -0,0 +1,35 @@
+import PromptParametersRenderer from "@/src/components/prompt/PromptParametersRenderer";
+import ModelSettingsRenderer from "@/src/components/prompt/model_settings/ModelSettingsRenderer";
+import PromptMetadataRenderer from "@/src/components/prompt/prompt_metadata/PromptMetadataRenderer";
+import {
+  PromptSchema,
+  checkParametersSupported,
+} from "@/src/utils/promptUtils";
+import { Flex } from "@mantine/core";
+import { Prompt } from "aiconfig";
+import { memo } from "react";
+
+type Props = {
+  prompt: Prompt;
+  promptSchema?: PromptSchema;
+};
+
+export default memo(function PromptActionBar({ prompt, promptSchema }: Props) {
+  // TODO: Handle collapse / expand / drag-to-resize
+  const modelSettingsSchema = promptSchema?.model_settings;
+  const promptMetadataSchema = promptSchema?.prompt_metadata;
+
+  return (
+    <Flex
+      direction="column"
+      justify="space-between"
+      style={{ borderLeft: "1px solid grey" }}
+    >
+      <ModelSettingsRenderer prompt={prompt} schema={modelSettingsSchema} />
+      <PromptMetadataRenderer prompt={prompt} schema={promptMetadataSchema} />
+      {checkParametersSupported(prompt) && (
+        <PromptParametersRenderer prompt={prompt} />
+      )}
+    </Flex>
+  );
+});
diff --git a/cli/aiconfig-editor/src/components/prompt/PromptContainer.tsx b/cli/aiconfig-editor/src/components/prompt/PromptContainer.tsx
index dd03980c3..74b0fefc6 100644
--- a/cli/aiconfig-editor/src/components/prompt/PromptContainer.tsx
+++ b/cli/aiconfig-editor/src/components/prompt/PromptContainer.tsx
@@ -1,5 +1,6 @@
+import PromptActionBar from "@/src/components/prompt/PromptActionBar";
 import PromptInput from "@/src/components/prompt/PromptInput";
-import { getPromptModelName } from "@/src/utils/promptUtils";
+import { getPromptModelName, getPromptSchema } from "@/src/utils/promptUtils";
 import { Flex, Card, Text } from "@mantine/core";
 import { Prompt, PromptInput as AIConfigPromptInput } from "aiconfig";
 import { memo, useCallback } from "react";
@@ -17,19 +18,28 @@ export default memo(function PromptContainer({
   onChangePromptInput,
   defaultConfigModelName,
 }: Props) {
-  // TODO: Show prompt name & metadata inside of settings editor later
+  // TODO: Move this to context
   const onChangeInput = useCallback(
     (newInput: AIConfigPromptInput) => onChangePromptInput(index, newInput),
     [index, onChangePromptInput]
   );
+
+  const promptSchema = getPromptSchema(prompt, defaultConfigModelName);
+
   return (
     <div style={{ marginTop: 16 }}>
-      <Card>
-        <Flex justify="space-between" m="sm">
-          <Text weight="bold">{`{{${prompt.name}}}}`}</Text>
-          <Text>{getPromptModelName(prompt, defaultConfigModelName)}</Text>
+      <Card withBorder>
+        <Flex justify="space-between">
+          <Flex direction="column">
+            <Flex justify="space-between" m="sm">
+              <Text weight="bold">{`{{${prompt.name}}}}`}</Text>
+              <Text>{getPromptModelName(prompt, defaultConfigModelName)}</Text>
+            </Flex>
+            <PromptInput input={prompt.input} onChangeInput={onChangeInput} />
+            {/* <PromptOutput /> */}
+          </Flex>
+          <PromptActionBar prompt={prompt} promptSchema={promptSchema} />
         </Flex>
-        <PromptInput input={prompt.input} onChangeInput={onChangeInput} />
       </Card>
     </div>
   );
diff --git a/cli/aiconfig-editor/src/components/prompt/PromptParametersRenderer.tsx b/cli/aiconfig-editor/src/components/prompt/PromptParametersRenderer.tsx
new file mode 100644
index 000000000..6ee32fd79
--- /dev/null
+++ b/cli/aiconfig-editor/src/components/prompt/PromptParametersRenderer.tsx
@@ -0,0 +1,10 @@
+import { Prompt } from "aiconfig";
+import { memo } from "react";
+
+type Props = {
+  prompt: Prompt;
+};
+
+export default memo(function PromptParametersRenderer({ prompt }: Props) {
+  return null; // TODO: Implement
+});
diff --git a/cli/aiconfig-editor/src/components/prompt/model_settings/ModelSettingsConfigRenderer.tsx b/cli/aiconfig-editor/src/components/prompt/model_settings/ModelSettingsConfigRenderer.tsx
new file mode 100644
index 000000000..260fc4763
--- /dev/null
+++ b/cli/aiconfig-editor/src/components/prompt/model_settings/ModelSettingsConfigRenderer.tsx
@@ -0,0 +1,10 @@
+import { JSONObject } from "aiconfig/dist/common";
+import { memo } from "react";
+
+type Props = {
+  settings: JSONObject;
+};
+
+export default memo(function ModelSettingsConfigRenderer({ settings }: Props) {
+  return <div>{JSON.stringify(settings)}</div>;
+});
diff --git a/cli/aiconfig-editor/src/components/prompt/model_settings/ModelSettingsRenderer.tsx b/cli/aiconfig-editor/src/components/prompt/model_settings/ModelSettingsRenderer.tsx
new file mode 100644
index 000000000..637a32540
--- /dev/null
+++ b/cli/aiconfig-editor/src/components/prompt/model_settings/ModelSettingsRenderer.tsx
@@ -0,0 +1,42 @@
+import ModelSettingsConfigRenderer from "@/src/components/prompt/model_settings/ModelSettingsConfigRenderer";
+import ModelSettingsSchemaRenderer from "@/src/components/prompt/model_settings/ModelSettingsSchemaRenderer";
+import { ModelSettingsSchema } from "@/src/utils/promptUtils";
+import { Flex, Text } from "@mantine/core";
+import { Prompt } from "aiconfig";
+import { memo } from "react";
+
+type Props = {
+  prompt: Prompt;
+  schema?: ModelSettingsSchema;
+};
+
+// Don't default to config-level model settings since that could be confusing
+// to have them shown at the prompt level in the editor but not in the config
+function getModelSettings(prompt: Prompt) {
+  if (typeof prompt.metadata?.model !== "string") {
+    return prompt.metadata?.model?.settings;
+  }
+}
+
+export default memo(function ModelSettingsRenderer({ prompt, schema }: Props) {
+  const modelSettings = getModelSettings(prompt);
+
+  let settingsComponent;
+
+  if (schema) {
+    settingsComponent = (
+      <ModelSettingsSchemaRenderer settings={modelSettings} schema={schema} />
+    );
+  } else if (modelSettings) {
+    settingsComponent = (
+      <ModelSettingsConfigRenderer settings={modelSettings} />
+    );
+  }
+
+  return (
+    <Flex direction="column">
+      <Text>Model Settings</Text>
+      {settingsComponent}
+    </Flex>
+  );
+});
diff --git a/cli/aiconfig-editor/src/components/prompt/model_settings/ModelSettingsSchemaRenderer.tsx b/cli/aiconfig-editor/src/components/prompt/model_settings/ModelSettingsSchemaRenderer.tsx
new file mode 100644
index 000000000..78b6f9625
--- /dev/null
+++ b/cli/aiconfig-editor/src/components/prompt/model_settings/ModelSettingsSchemaRenderer.tsx
@@ -0,0 +1,32 @@
+import SettingsPropertyRenderer from "@/src/components/SettingsPropertyRenderer";
+import { useSchemaState } from "@/src/hooks/useSchemaState";
+import { ModelSettingsSchema } from "@/src/utils/promptUtils";
+import { Flex } from "@mantine/core";
+import { JSONObject } from "aiconfig/dist/common";
+import { memo } from "react";
+
+type Props = {
+  schema: ModelSettingsSchema;
+  settings?: JSONObject;
+};
+
+export default memo(function ModelSettingsSchemaRenderer({
+  schema,
+  settings,
+}: Props) {
+  const { schemaState } = useSchemaState(schema, settings);
+
+  return (
+    <Flex direction="column">
+      {Object.entries(schema.properties).map(([key, value]) => (
+        <SettingsPropertyRenderer
+          propertyName={key}
+          key={key}
+          property={value}
+          isRequired={schema.required?.includes(key)}
+          initialValue={schemaState[key].value}
+        />
+      ))}
+    </Flex>
+  );
+});
diff --git a/cli/aiconfig-editor/src/components/prompt/prompt_metadata/PromptMetadataRenderer.tsx b/cli/aiconfig-editor/src/components/prompt/prompt_metadata/PromptMetadataRenderer.tsx
new file mode 100644
index 000000000..6af552ff2
--- /dev/null
+++ b/cli/aiconfig-editor/src/components/prompt/prompt_metadata/PromptMetadataRenderer.tsx
@@ -0,0 +1,24 @@
+import { PromptMetadataSchema } from "@/src/utils/promptUtils";
+import { Prompt } from "aiconfig";
+import { memo } from "react";
+
+type Props = {
+  prompt: Prompt;
+  schema?: PromptMetadataSchema;
+};
+
+function ModelMetadataConfigRenderer({ prompt }: Props) {
+  return null; // TODO: Implement
+}
+
+function ModelMetadataSchemaRenderer({ prompt, schema }: Props) {
+  return null; // TODO: Implement
+}
+
+export default memo(function ModelSettingsRenderer({ prompt, schema }: Props) {
+  return schema ? (
+    <ModelMetadataSchemaRenderer prompt={prompt} schema={schema} />
+  ) : (
+    <ModelMetadataConfigRenderer prompt={prompt} />
+  );
+});
diff --git a/cli/aiconfig-editor/src/hooks/useSchemaState.ts b/cli/aiconfig-editor/src/hooks/useSchemaState.ts
new file mode 100644
index 000000000..53eb7ab4b
--- /dev/null
+++ b/cli/aiconfig-editor/src/hooks/useSchemaState.ts
@@ -0,0 +1,46 @@
+import {
+  ModelSettingsSchema,
+  PromptMetadataSchema,
+} from "@/src/utils/promptUtils";
+import { useCallback, useRef, useState } from "react";
+
+// Local state to maintain all the possible properties from a schema, as well as 'dirty' state to track
+// which properties have been changed. This is used to determine which properties to propagate to the config.
+// Otherwise, the config would be bloated with unnecessary settings just by loading it in the editor.
+export function useSchemaState(
+  schema: ModelSettingsSchema | PromptMetadataSchema,
+  initialData?: Record<string, unknown>
+) {
+  const [schemaState, setSchemaState] = useState<
+    Record<string, { value: unknown; dirty: boolean }>
+  >(
+    Object.keys(schema.properties).reduce((acc, key) => {
+      acc[key] = { value: initialData?.[key] ?? null, dirty: false };
+      return acc;
+    }, {} as Record<string, { value: unknown; dirty: boolean }>)
+  );
+
+  const stateRef = useRef(schemaState);
+  stateRef.current = schemaState;
+
+  // Get the state of the schema as a concrete object, only including properties that have been changed
+  const getConcreteState = useCallback(
+    () =>
+      Object.keys(stateRef.current).reduce((acc, key) => {
+        if (stateRef.current[key].dirty) {
+          acc[key] = stateRef.current[key].value;
+        }
+        return acc;
+      }, {} as Record<string, unknown>),
+    []
+  );
+
+  const setSchemaValue = useCallback((key: string, value: unknown) => {
+    setSchemaState((currentState) => ({
+      ...currentState,
+      [key]: { value, dirty: true },
+    }));
+  }, []);
+
+  return { schemaState, getConcreteState, setSchemaValue };
+}
diff --git a/cli/aiconfig-editor/src/shared/prompt_schemas/OpenAIChatModelParserPromptSchema.ts b/cli/aiconfig-editor/src/shared/prompt_schemas/OpenAIChatModelParserPromptSchema.ts
new file mode 100644
index 000000000..2113842fd
--- /dev/null
+++ b/cli/aiconfig-editor/src/shared/prompt_schemas/OpenAIChatModelParserPromptSchema.ts
@@ -0,0 +1,106 @@
+import { PromptSchema } from "@/src/utils/promptUtils";
+
+export const OpenAIChatModelParserPromptSchema: PromptSchema = {
+  input: {
+    type: "string",
+  },
+  model_settings: {
+    properties: {
+      system_prompt: {
+        type: "string",
+      },
+      frequency_penalty: {
+        type: "number",
+        minimum: -2.0,
+        maximum: 2.0,
+      },
+      function_call: {
+        type: "union",
+        types: [
+          {
+            type: "string",
+            enum: ["none", "auto"],
+          },
+          {
+            type: "object",
+            required: ["name"],
+            properties: {
+              name: {
+                type: "string",
+              },
+            },
+          },
+        ],
+      },
+      functions: {
+        type: "array",
+        items: {
+          type: "object",
+          required: ["name", "parameters"],
+          parameters: {
+            name: {
+              type: "string",
+            },
+            parameters: {
+              type: "object", // TODO: Figure this out -- it's a JSON schema object
+            },
+            description: {
+              type: "string",
+            },
+          },
+        },
+      },
+      logit_bias: {
+        type: "map",
+        keys: {
+          type: "string",
+        },
+        items: {
+          type: "integer",
+          minimum: -100,
+          maximum: 100,
+        },
+      },
+      max_tokens: {
+        type: "number",
+      },
+      n: {
+        type: "number",
+      },
+      presence_penalty: {
+        type: "number",
+        minimum: -2.0,
+        maximum: 2.0,
+      },
+      stop: {
+        type: "array",
+        items: {
+          type: "string",
+        },
+      },
+      stream: {
+        type: "boolean",
+      },
+      temperature: {
+        type: "number",
+        minimum: 0.0,
+        maximum: 2.0,
+      },
+      top_p: {
+        type: "number",
+        minimum: 0.0,
+        maximum: 1.0,
+      },
+      user: {
+        type: "string",
+      },
+    },
+  },
+  prompt_metadata: {
+    properties: {
+      remember_chat_context: {
+        type: "boolean",
+      },
+    },
+  },
+};
diff --git a/cli/aiconfig-editor/src/utils/promptUtils.ts b/cli/aiconfig-editor/src/utils/promptUtils.ts
index 74991dcc2..504f66ac6 100644
--- a/cli/aiconfig-editor/src/utils/promptUtils.ts
+++ b/cli/aiconfig-editor/src/utils/promptUtils.ts
@@ -1,4 +1,5 @@
-import { AIConfig, Prompt } from "aiconfig";
+import { Prompt } from "aiconfig";
+import { OpenAIChatModelParserPromptSchema } from "../shared/prompt_schemas/OpenAIChatModelParserPromptSchema";
 
 /**
  * Get the name of the model for the specified prompt. The name will either be specified in the prompt's
@@ -23,3 +24,69 @@ export function getPromptModelName(
   // Model must be specified as default if not specified for the Prompt
   return defaultConfigModelName!;
 }
+
+// TODO: Schemas should be statically defined with the model parsers and loaded alongside config, keyed on registered model names
+// Notes here: We could technically just refine the full JSON schema for the prompt since we currently allow model parsers to have full control, including:
+//   - defining metadata outside of model metadata --> e.g. remember_chat_context for openai parser
+//   - placing responsibility for proper handling/serialization of core config properties on the model parser (parameters, tags)
+// That would allow us to do some powerful things like:
+//   - share JSON schema for model parser between languages
+//   - validate serialization against it
+// Long-term, maybe we just have model parser core class handle serialization of parameters/tags and concrete implementations just
+// handle serialization of non-core metadata and model inference settings?
+// In any case, below we are defining 3 keys in the schema:
+//   - input: the prompt input, either string or more refined object including data / attachment types
+//   - model_settings: supported settings for the model, will be serialized either in top-level model metadata or prompt.metadata.model.settings
+//   - prompt_metadata: additional metadata for the prompt, EXCLUDES parameters, tags, model
+// A couple reasons for doing this vs refinining full JSON schema:
+//   - we may want to set constraints, like range of acceptable values, which aren't part of the prompt representation in JSON schema
+//   - we can focus on the important 3 params defined above, instead of requiring re-definining the full JSON schema
+// Should we define a JSON schema for PromptSchema type so we can safely serialize/deserialize them?
+export const PROMPT_SCHEMAS: Record<string, PromptSchema> = {
+  "gpt-3.5-turbo": OpenAIChatModelParserPromptSchema,
+  "gpt-4": OpenAIChatModelParserPromptSchema,
+};
+
+export type PromptInputSchema = {
+  type: "string" | "object";
+};
+
+export type ModelSettingsSchema = {
+  properties: Record<string, { [key: string]: any }>;
+  required?: string[];
+};
+
+export type PromptMetadataSchema = {
+  properties: Record<string, { [key: string]: any }>;
+  required?: string[];
+};
+
+export type PromptSchema = {
+  input: PromptInputSchema;
+  model_settings?: ModelSettingsSchema;
+  prompt_metadata?: PromptMetadataSchema;
+};
+
+export function getPromptSchema(
+  prompt: Prompt,
+  defaultConfigModelName?: string
+): PromptSchema | undefined {
+  const modelName = getPromptModelName(prompt, defaultConfigModelName);
+  return PROMPT_SCHEMAS[modelName];
+}
+
+function isTextInputModality(prompt: Prompt): boolean {
+  const schema = getPromptSchema(prompt);
+  if (schema) {
+    return schema.input.type === "string"; // TODO: Handle object case
+  }
+
+  return (
+    // TODO: Handle case where data is object with values associated with mimetype?
+    typeof prompt.input === "string" || typeof prompt.input.data === "string"
+  );
+}
+
+export function checkParametersSupported(prompt: Prompt): boolean {
+  return prompt.metadata?.parameters != null || isTextInputModality(prompt);
+}
diff --git a/cli/aiconfig-editor/travel.aiconfig.json b/cli/aiconfig-editor/travel.aiconfig.json
index eb48e36ff..9924f316e 100644
--- a/cli/aiconfig-editor/travel.aiconfig.json
+++ b/cli/aiconfig-editor/travel.aiconfig.json
@@ -31,6 +31,21 @@
           "order_by": "geographic location"
         }
       }
+    },
+    {
+      "name": "single_day_itinerary",
+      "input": "Refine this itinerary to fit within a single day: {{gen_itinerary.output}}.",
+      "metadata": {
+        "model": {
+          "name": "gpt-4",
+          "settings": {
+            "max_tokens": 1000
+          }
+        },
+        "parameters": {
+          "order_by": "geographic location"
+        }
+      }
     }
   ]
 }
\ No newline at end of file