Merge pull request #61 from actions/sgoedecke/prompt-file

Support .prompt.yml files
Merge branch 'main' into sgoedecke/prompt-file
2025-07-21 15:14:53 +10:00 · 2025-07-21 15:01:31 +10:00 · 2025-07-21 15:00:56 +10:00 · 2025-07-21 15:00:09 +10:00 · 2025-07-21 14:57:40 +10:00 · 2025-07-21 14:57:19 +10:00
22 changed files with 4251 additions and 113 deletions
--- a/.github/workflows/linter.yml
+++ b/.github/workflows/linter.yml
@@ -38,7 +38,7 @@ jobs:

      - name: Lint Codebase
        id: super-linter
-        uses: super-linter/super-linter/slim@12150456a73e248bdc94d0794898f94e23127c88
+        uses: super-linter/super-linter/slim@5119dcd8011e92182ce8219d9e9efc82f16fddb6
        env:
          DEFAULT_BRANCH: main
          FILTER_REGEX_EXCLUDE: dist/**/*
--- a/.github/workflows/release-new-action-version.yml
+++ b/.github/workflows/release-new-action-version.yml
@@ -0,0 +1,22 @@
+name: Release new action version
+
+on:
+  release:
+    types: [released]
+
+env:
+  TAG_NAME: ${{ github.event.release.tag_name }}
+permissions:
+  contents: write
+
+jobs:
+  update_tag:
+    name:
+      Update the major tag to include the ${{ github.event.release.tag_name }}
+      changes
+    runs-on: ubuntu-latest
+    steps:
+      - name: Update the ${{ env.TAG_NAME }} tag
+        uses: actions/publish-action@v0.2.2
+        with:
+          source-tag: ${{ env.TAG_NAME }}
--- a/.licenses/npm/@types/js-yaml.dep.yml
+++ b/.licenses/npm/@types/js-yaml.dep.yml
@@ -0,0 +1,32 @@
+---
+name: "@types/js-yaml"
+version: 4.0.9
+type: npm
+summary: TypeScript definitions for js-yaml
+homepage: https://github.com/DefinitelyTyped/DefinitelyTyped/tree/master/types/js-yaml
+license: mit
+licenses:
+- sources: LICENSE
+  text: |2
+        MIT License
+
+        Copyright (c) Microsoft Corporation.
+
+        Permission is hereby granted, free of charge, to any person obtaining a copy
+        of this software and associated documentation files (the "Software"), to deal
+        in the Software without restriction, including without limitation the rights
+        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+        copies of the Software, and to permit persons to whom the Software is
+        furnished to do so, subject to the following conditions:
+
+        The above copyright notice and this permission notice shall be included in all
+        copies or substantial portions of the Software.
+
+        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+        SOFTWARE
+notices: []
--- a/.licenses/npm/argparse.dep.yml
+++ b/.licenses/npm/argparse.dep.yml
@@ -0,0 +1,265 @@
+---
+name: argparse
+version: 2.0.1
+type: npm
+summary: CLI arguments parser. Native port of python's argparse.
+homepage:
+license: other
+licenses:
+- sources: LICENSE
+  text: |
+    A. HISTORY OF THE SOFTWARE
+    ==========================
+
+    Python was created in the early 1990s by Guido van Rossum at Stichting
+    Mathematisch Centrum (CWI, see http://www.cwi.nl) in the Netherlands
+    as a successor of a language called ABC.  Guido remains Python's
+    principal author, although it includes many contributions from others.
+
+    In 1995, Guido continued his work on Python at the Corporation for
+    National Research Initiatives (CNRI, see http://www.cnri.reston.va.us)
+    in Reston, Virginia where he released several versions of the
+    software.
+
+    In May 2000, Guido and the Python core development team moved to
+    BeOpen.com to form the BeOpen PythonLabs team.  In October of the same
+    year, the PythonLabs team moved to Digital Creations, which became
+    Zope Corporation.  In 2001, the Python Software Foundation (PSF, see
+    https://www.python.org/psf/) was formed, a non-profit organization
+    created specifically to own Python-related Intellectual Property.
+    Zope Corporation was a sponsoring member of the PSF.
+
+    All Python releases are Open Source (see http://www.opensource.org for
+    the Open Source Definition).  Historically, most, but not all, Python
+    releases have also been GPL-compatible; the table below summarizes
+    the various releases.
+
+        Release         Derived     Year        Owner       GPL-
+                        from                                compatible? (1)
+
+        0.9.0 thru 1.2              1991-1995   CWI         yes
+        1.3 thru 1.5.2  1.2         1995-1999   CNRI        yes
+        1.6             1.5.2       2000        CNRI        no
+        2.0             1.6         2000        BeOpen.com  no
+        1.6.1           1.6         2001        CNRI        yes (2)
+        2.1             2.0+1.6.1   2001        PSF         no
+        2.0.1           2.0+1.6.1   2001        PSF         yes
+        2.1.1           2.1+2.0.1   2001        PSF         yes
+        2.1.2           2.1.1       2002        PSF         yes
+        2.1.3           2.1.2       2002        PSF         yes
+        2.2 and above   2.1.1       2001-now    PSF         yes
+
+    Footnotes:
+
+    (1) GPL-compatible doesn't mean that we're distributing Python under
+        the GPL.  All Python licenses, unlike the GPL, let you distribute
+        a modified version without making your changes open source.  The
+        GPL-compatible licenses make it possible to combine Python with
+        other software that is released under the GPL; the others don't.
+
+    (2) According to Richard Stallman, 1.6.1 is not GPL-compatible,
+        because its license has a choice of law clause.  According to
+        CNRI, however, Stallman's lawyer has told CNRI's lawyer that 1.6.1
+        is "not incompatible" with the GPL.
+
+    Thanks to the many outside volunteers who have worked under Guido's
+    direction to make these releases possible.
+
+
+    B. TERMS AND CONDITIONS FOR ACCESSING OR OTHERWISE USING PYTHON
+    ===============================================================
+
+    PYTHON SOFTWARE FOUNDATION LICENSE VERSION 2
+    --------------------------------------------
+
+    1. This LICENSE AGREEMENT is between the Python Software Foundation
+    ("PSF"), and the Individual or Organization ("Licensee") accessing and
+    otherwise using this software ("Python") in source or binary form and
+    its associated documentation.
+
+    2. Subject to the terms and conditions of this License Agreement, PSF hereby
+    grants Licensee a nonexclusive, royalty-free, world-wide license to reproduce,
+    analyze, test, perform and/or display publicly, prepare derivative works,
+    distribute, and otherwise use Python alone or in any derivative version,
+    provided, however, that PSF's License Agreement and PSF's notice of copyright,
+    i.e., "Copyright (c) 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
+    2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020 Python Software Foundation;
+    All Rights Reserved" are retained in Python alone or in any derivative version
+    prepared by Licensee.
+
+    3. In the event Licensee prepares a derivative work that is based on
+    or incorporates Python or any part thereof, and wants to make
+    the derivative work available to others as provided herein, then
+    Licensee hereby agrees to include in any such work a brief summary of
+    the changes made to Python.
+
+    4. PSF is making Python available to Licensee on an "AS IS"
+    basis.  PSF MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
+    IMPLIED.  BY WAY OF EXAMPLE, BUT NOT LIMITATION, PSF MAKES NO AND
+    DISCLAIMS ANY REPRESENTATION OR WARRANTY OF MERCHANTABILITY OR FITNESS
+    FOR ANY PARTICULAR PURPOSE OR THAT THE USE OF PYTHON WILL NOT
+    INFRINGE ANY THIRD PARTY RIGHTS.
+
+    5. PSF SHALL NOT BE LIABLE TO LICENSEE OR ANY OTHER USERS OF PYTHON
+    FOR ANY INCIDENTAL, SPECIAL, OR CONSEQUENTIAL DAMAGES OR LOSS AS
+    A RESULT OF MODIFYING, DISTRIBUTING, OR OTHERWISE USING PYTHON,
+    OR ANY DERIVATIVE THEREOF, EVEN IF ADVISED OF THE POSSIBILITY THEREOF.
+
+    6. This License Agreement will automatically terminate upon a material
+    breach of its terms and conditions.
+
+    7. Nothing in this License Agreement shall be deemed to create any
+    relationship of agency, partnership, or joint venture between PSF and
+    Licensee.  This License Agreement does not grant permission to use PSF
+    trademarks or trade name in a trademark sense to endorse or promote
+    products or services of Licensee, or any third party.
+
+    8. By copying, installing or otherwise using Python, Licensee
+    agrees to be bound by the terms and conditions of this License
+    Agreement.
+
+
+    BEOPEN.COM LICENSE AGREEMENT FOR PYTHON 2.0
+    -------------------------------------------
+
+    BEOPEN PYTHON OPEN SOURCE LICENSE AGREEMENT VERSION 1
+
+    1. This LICENSE AGREEMENT is between BeOpen.com ("BeOpen"), having an
+    office at 160 Saratoga Avenue, Santa Clara, CA 95051, and the
+    Individual or Organization ("Licensee") accessing and otherwise using
+    this software in source or binary form and its associated
+    documentation ("the Software").
+
+    2. Subject to the terms and conditions of this BeOpen Python License
+    Agreement, BeOpen hereby grants Licensee a non-exclusive,
+    royalty-free, world-wide license to reproduce, analyze, test, perform
+    and/or display publicly, prepare derivative works, distribute, and
+    otherwise use the Software alone or in any derivative version,
+    provided, however, that the BeOpen Python License is retained in the
+    Software, alone or in any derivative version prepared by Licensee.
+
+    3. BeOpen is making the Software available to Licensee on an "AS IS"
+    basis.  BEOPEN MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
+    IMPLIED.  BY WAY OF EXAMPLE, BUT NOT LIMITATION, BEOPEN MAKES NO AND
+    DISCLAIMS ANY REPRESENTATION OR WARRANTY OF MERCHANTABILITY OR FITNESS
+    FOR ANY PARTICULAR PURPOSE OR THAT THE USE OF THE SOFTWARE WILL NOT
+    INFRINGE ANY THIRD PARTY RIGHTS.
+
+    4. BEOPEN SHALL NOT BE LIABLE TO LICENSEE OR ANY OTHER USERS OF THE
+    SOFTWARE FOR ANY INCIDENTAL, SPECIAL, OR CONSEQUENTIAL DAMAGES OR LOSS
+    AS A RESULT OF USING, MODIFYING OR DISTRIBUTING THE SOFTWARE, OR ANY
+    DERIVATIVE THEREOF, EVEN IF ADVISED OF THE POSSIBILITY THEREOF.
+
+    5. This License Agreement will automatically terminate upon a material
+    breach of its terms and conditions.
+
+    6. This License Agreement shall be governed by and interpreted in all
+    respects by the law of the State of California, excluding conflict of
+    law provisions.  Nothing in this License Agreement shall be deemed to
+    create any relationship of agency, partnership, or joint venture
+    between BeOpen and Licensee.  This License Agreement does not grant
+    permission to use BeOpen trademarks or trade names in a trademark
+    sense to endorse or promote products or services of Licensee, or any
+    third party.  As an exception, the "BeOpen Python" logos available at
+    http://www.pythonlabs.com/logos.html may be used according to the
+    permissions granted on that web page.
+
+    7. By copying, installing or otherwise using the software, Licensee
+    agrees to be bound by the terms and conditions of this License
+    Agreement.
+
+
+    CNRI LICENSE AGREEMENT FOR PYTHON 1.6.1
+    ---------------------------------------
+
+    1. This LICENSE AGREEMENT is between the Corporation for National
+    Research Initiatives, having an office at 1895 Preston White Drive,
+    Reston, VA 20191 ("CNRI"), and the Individual or Organization
+    ("Licensee") accessing and otherwise using Python 1.6.1 software in
+    source or binary form and its associated documentation.
+
+    2. Subject to the terms and conditions of this License Agreement, CNRI
+    hereby grants Licensee a nonexclusive, royalty-free, world-wide
+    license to reproduce, analyze, test, perform and/or display publicly,
+    prepare derivative works, distribute, and otherwise use Python 1.6.1
+    alone or in any derivative version, provided, however, that CNRI's
+    License Agreement and CNRI's notice of copyright, i.e., "Copyright (c)
+    1995-2001 Corporation for National Research Initiatives; All Rights
+    Reserved" are retained in Python 1.6.1 alone or in any derivative
+    version prepared by Licensee.  Alternately, in lieu of CNRI's License
+    Agreement, Licensee may substitute the following text (omitting the
+    quotes): "Python 1.6.1 is made available subject to the terms and
+    conditions in CNRI's License Agreement.  This Agreement together with
+    Python 1.6.1 may be located on the Internet using the following
+    unique, persistent identifier (known as a handle): 1895.22/1013.  This
+    Agreement may also be obtained from a proxy server on the Internet
+    using the following URL: http://hdl.handle.net/1895.22/1013".
+
+    3. In the event Licensee prepares a derivative work that is based on
+    or incorporates Python 1.6.1 or any part thereof, and wants to make
+    the derivative work available to others as provided herein, then
+    Licensee hereby agrees to include in any such work a brief summary of
+    the changes made to Python 1.6.1.
+
+    4. CNRI is making Python 1.6.1 available to Licensee on an "AS IS"
+    basis.  CNRI MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
+    IMPLIED.  BY WAY OF EXAMPLE, BUT NOT LIMITATION, CNRI MAKES NO AND
+    DISCLAIMS ANY REPRESENTATION OR WARRANTY OF MERCHANTABILITY OR FITNESS
+    FOR ANY PARTICULAR PURPOSE OR THAT THE USE OF PYTHON 1.6.1 WILL NOT
+    INFRINGE ANY THIRD PARTY RIGHTS.
+
+    5. CNRI SHALL NOT BE LIABLE TO LICENSEE OR ANY OTHER USERS OF PYTHON
+    1.6.1 FOR ANY INCIDENTAL, SPECIAL, OR CONSEQUENTIAL DAMAGES OR LOSS AS
+    A RESULT OF MODIFYING, DISTRIBUTING, OR OTHERWISE USING PYTHON 1.6.1,
+    OR ANY DERIVATIVE THEREOF, EVEN IF ADVISED OF THE POSSIBILITY THEREOF.
+
+    6. This License Agreement will automatically terminate upon a material
+    breach of its terms and conditions.
+
+    7. This License Agreement shall be governed by the federal
+    intellectual property law of the United States, including without
+    limitation the federal copyright law, and, to the extent such
+    U.S. federal law does not apply, by the law of the Commonwealth of
+    Virginia, excluding Virginia's conflict of law provisions.
+    Notwithstanding the foregoing, with regard to derivative works based
+    on Python 1.6.1 that incorporate non-separable material that was
+    previously distributed under the GNU General Public License (GPL), the
+    law of the Commonwealth of Virginia shall govern this License
+    Agreement only as to issues arising under or with respect to
+    Paragraphs 4, 5, and 7 of this License Agreement.  Nothing in this
+    License Agreement shall be deemed to create any relationship of
+    agency, partnership, or joint venture between CNRI and Licensee.  This
+    License Agreement does not grant permission to use CNRI trademarks or
+    trade name in a trademark sense to endorse or promote products or
+    services of Licensee, or any third party.
+
+    8. By clicking on the "ACCEPT" button where indicated, or by copying,
+    installing or otherwise using Python 1.6.1, Licensee agrees to be
+    bound by the terms and conditions of this License Agreement.
+
+            ACCEPT
+
+
+    CWI LICENSE AGREEMENT FOR PYTHON 0.9.0 THROUGH 1.2
+    --------------------------------------------------
+
+    Copyright (c) 1991 - 1995, Stichting Mathematisch Centrum Amsterdam,
+    The Netherlands.  All rights reserved.
+
+    Permission to use, copy, modify, and distribute this software and its
+    documentation for any purpose and without fee is hereby granted,
+    provided that the above copyright notice appear in all copies and that
+    both that copyright notice and this permission notice appear in
+    supporting documentation, and that the name of Stichting Mathematisch
+    Centrum or CWI not be used in advertising or publicity pertaining to
+    distribution of the software without specific, written prior
+    permission.
+
+    STICHTING MATHEMATISCH CENTRUM DISCLAIMS ALL WARRANTIES WITH REGARD TO
+    THIS SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
+    FITNESS, IN NO EVENT SHALL STICHTING MATHEMATISCH CENTRUM BE LIABLE
+    FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+    WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+    ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT
+    OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+notices: []
--- a/.licenses/npm/js-yaml.dep.yml
+++ b/.licenses/npm/js-yaml.dep.yml
@@ -0,0 +1,32 @@
+---
+name: js-yaml
+version: 4.1.0
+type: npm
+summary: YAML 1.2 parser and serializer
+homepage:
+license: mit
+licenses:
+- sources: LICENSE
+  text: |
+    (The MIT License)
+
+    Copyright (C) 2011-2015 by Vitaly Puzrin
+
+    Permission is hereby granted, free of charge, to any person obtaining a copy
+    of this software and associated documentation files (the "Software"), to deal
+    in the Software without restriction, including without limitation the rights
+    to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+    copies of the Software, and to permit persons to whom the Software is
+    furnished to do so, subject to the following conditions:
+
+    The above copyright notice and this permission notice shall be included in
+    all copies or substantial portions of the Software.
+
+    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+    IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+    FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+    AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+    LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+    OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+    THE SOFTWARE.
+notices: []
--- a/README.md
+++ b/README.md
@@ -36,17 +36,90 @@ jobs:

 ### Using a prompt file

-You can also provide a prompt file instead of an inline prompt:
+You can also provide a prompt file instead of an inline prompt. The action
+supports both plain text files and structured `.prompt.yml` files:

 ```yaml
 steps:
-  - name: Run AI Inference with Prompt File
+  - name: Run AI Inference with Text File
    id: inference
    uses: actions/ai-inference@v1
    with:
      prompt-file: './path/to/prompt.txt'
 ```

+### Using GitHub prompt.yml files
+
+For more advanced use cases, you can use structured `.prompt.yml` files that
+support templating, custom models, and JSON schema responses:
+
+```yaml
+steps:
+  - name: Run AI Inference with Prompt YAML
+    id: inference
+    uses: actions/ai-inference@v1
+    with:
+      prompt-file: './.github/prompts/sample.prompt.yml'
+      input: |
+        var1: hello
+        var2: ${{ steps.some-step.outputs.output }}
+        var3: |
+          Lorem Ipsum
+          Hello World
+```
+
+#### Simple prompt.yml example
+
+```yaml
+messages:
+  - role: system
+    content: Be as concise as possible
+  - role: user
+    content: 'Compare {{a}} and {{b}}, please'
+model: openai/gpt-4o
+```
+
+#### Prompt.yml with JSON schema support
+
+```yaml
+messages:
+  - role: system
+    content:
+      You are a helpful assistant that describes animals using JSON format
+  - role: user
+    content: |-
+      Describe a {{animal}}
+      Use JSON format as specified in the response schema
+model: openai/gpt-4o
+responseFormat: json_schema
+jsonSchema: |-
+  {
+    "name": "describe_animal",
+    "strict": true,
+    "schema": {
+      "type": "object",
+      "properties": {
+        "name": {
+          "type": "string",
+          "description": "The name of the animal"
+        },
+        "habitat": {
+          "type": "string",
+          "description": "The habitat the animal lives in"
+        }
+      },
+      "additionalProperties": false,
+      "required": [
+        "name",
+        "habitat"
+      ]
+    }
+  }
+```
+
+Variables in prompt.yml files are templated using `{{variable}}` format and are
+supplied via the `input` parameter in YAML format.
+
 ### Using a system prompt file

 In addition to the regular prompt, you can provide a system prompt file instead
@@ -90,34 +163,36 @@ repository management, issue tracking, and pull request operations.
 steps:
  - name: AI Inference with GitHub Tools
    id: inference
-    uses: actions/ai-inference@v1
+    uses: actions/ai-inference@v1.2
    with:
      prompt: 'List my open pull requests and create a summary'
      enable-github-mcp: true
+      token: ${{ secrets.USER_PAT }}
 ```

 When MCP is enabled, the AI model will have access to GitHub tools and can
 perform actions like searching issues and PRs.

-**Note:** MCP integration requires your workflow token to have appropriate
-GitHub permissions for the operations the AI will perform.
+**Note:** For now, MCP integration cannot be used with the built-in token. You
+must pass a GitHub PAT into `token:` instead.

 ## Inputs

 Various inputs are defined in [`action.yml`](action.yml) to let you configure
 the action:

-| Name                 | Description                                                                                                                                       | Default                              |
-| -------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------ |
-| `token`              | Token to use for inference. Typically the GITHUB_TOKEN secret                                                                                     | `github.token`                       |
-| `prompt`             | The prompt to send to the model                                                                                                                   | N/A                                  |
-| `prompt-file`        | Path to a file containing the prompt. If both `prompt` and `prompt-file` are provided, `prompt-file` takes precedence                             | `""`                                 |
-| `system-prompt`      | The system prompt to send to the model                                                                                                            | `"You are a helpful assistant"`      |
-| `system-prompt-file` | Path to a file containing the system prompt. If both `system-prompt` and `system-prompt-file` are provided, `system-prompt-file` takes precedence | `""`                                 |
-| `model`              | The model to use for inference. Must be available in the [GitHub Models](https://github.com/marketplace?type=models) catalog                      | `openai/gpt-4.1`                     |
-| `endpoint`           | The endpoint to use for inference. If you're running this as part of an org, you should probably use the org-specific Models endpoint             | `https://models.github.ai/inference` |
-| `max-tokens`         | The max number of tokens to generate                                                                                                              | 200                                  |
-| `enable-github-mcp`  | Enable Model Context Protocol integration with GitHub tools                                                                                       | `false`                              |
+| Name                 | Description                                                                                                                                                   | Default                              |
+| -------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------ |
+| `token`              | Token to use for inference. Typically the GITHUB_TOKEN secret                                                                                                 | `github.token`                       |
+| `prompt`             | The prompt to send to the model                                                                                                                               | N/A                                  |
+| `prompt-file`        | Path to a file containing the prompt (supports .txt and .prompt.yml formats). If both `prompt` and `prompt-file` are provided, `prompt-file` takes precedence | `""`                                 |
+| `input`              | Template variables in YAML format for .prompt.yml files (e.g., `var1: value1` on separate lines)                                                              | `""`                                 |
+| `system-prompt`      | The system prompt to send to the model                                                                                                                        | `"You are a helpful assistant"`      |
+| `system-prompt-file` | Path to a file containing the system prompt. If both `system-prompt` and `system-prompt-file` are provided, `system-prompt-file` takes precedence             | `""`                                 |
+| `model`              | The model to use for inference. Must be available in the [GitHub Models](https://github.com/marketplace?type=models) catalog                                  | `openai/gpt-4o`                      |
+| `endpoint`           | The endpoint to use for inference. If you're running this as part of an org, you should probably use the org-specific Models endpoint                         | `https://models.github.ai/inference` |
+| `max-tokens`         | The max number of tokens to generate                                                                                                                          | 200                                  |
+| `enable-github-mcp`  | Enable Model Context Protocol integration with GitHub tools                                                                                                   | `false`                              |

 ## Outputs

--- a/fixtures/prompts/json-schema.prompt.yml
+++ b/fixtures/prompts/json-schema.prompt.yml
@@ -0,0 +1,41 @@
+messages:
+  - role: system
+    content:
+      You are a helpful assistant that describes animals using JSON format
+  - role: user
+    content: |-
+      Describe a {{animal}}
+      Use JSON format as specified in the response schema
+model: openai/gpt-4o
+responseFormat: json_schema
+jsonSchema: |-
+  {
+    "name": "describe_animal",
+    "strict": true,
+    "schema": {
+      "type": "object",
+      "properties": {
+        "name": {
+          "type": "string",
+          "description": "The name of the animal"
+        },
+        "habitat": {
+          "type": "string",
+          "description": "The habitat the animal lives in"
+        },
+        "characteristics": {
+          "type": "array",
+          "items": {
+            "type": "string"
+          },
+          "description": "Key characteristics of the animal"
+        }
+      },
+      "additionalProperties": false,
+      "required": [
+        "name",
+        "habitat",
+        "characteristics"
+      ]
+    }
+  }
--- a/fixtures/prompts/simple.prompt.yml
+++ b/fixtures/prompts/simple.prompt.yml
@@ -0,0 +1,6 @@
+messages:
+  - role: system
+    content: Be as concise as possible
+  - role: user
+    content: 'Compare {{a}} and {{b}}, please'
+model: openai/gpt-4o
--- a/tests/helpers-inference.test.ts
+++ b/tests/helpers-inference.test.ts
@@ -0,0 +1,163 @@
+import { describe, it, expect } from '@jest/globals'
+import {
+  buildMessages,
+  buildResponseFormat,
+  buildInferenceRequest
+} from '../src/helpers'
+import { PromptConfig } from '../src/prompt'
+
+describe('helpers.ts - inference request building', () => {
+  describe('buildMessages', () => {
+    it('should build messages from prompt config', () => {
+      const promptConfig: PromptConfig = {
+        messages: [
+          { role: 'system', content: 'System message' },
+          { role: 'user', content: 'User message' }
+        ]
+      }
+
+      const result = buildMessages(promptConfig)
+      expect(result).toEqual([
+        { role: 'system', content: 'System message' },
+        { role: 'user', content: 'User message' }
+      ])
+    })
+
+    it('should build messages from legacy format', () => {
+      const result = buildMessages(undefined, 'System prompt', 'User prompt')
+      expect(result).toEqual([
+        { role: 'system', content: 'System prompt' },
+        { role: 'user', content: 'User prompt' }
+      ])
+    })
+
+    it('should use default system prompt when none provided', () => {
+      const result = buildMessages(undefined, undefined, 'User prompt')
+      expect(result).toEqual([
+        { role: 'system', content: 'You are a helpful assistant' },
+        { role: 'user', content: 'User prompt' }
+      ])
+    })
+  })
+
+  describe('buildResponseFormat', () => {
+    it('should build JSON schema response format', () => {
+      const promptConfig: PromptConfig = {
+        messages: [],
+        responseFormat: 'json_schema',
+        jsonSchema: JSON.stringify({
+          name: 'test_schema',
+          schema: { type: 'object' }
+        })
+      }
+
+      const result = buildResponseFormat(promptConfig)
+      expect(result).toEqual({
+        type: 'json_schema',
+        json_schema: {
+          name: 'test_schema',
+          schema: { type: 'object' }
+        }
+      })
+    })
+
+    it('should return undefined for text format', () => {
+      const promptConfig: PromptConfig = {
+        messages: [],
+        responseFormat: 'text'
+      }
+
+      const result = buildResponseFormat(promptConfig)
+      expect(result).toBeUndefined()
+    })
+
+    it('should return undefined when no response format specified', () => {
+      const promptConfig: PromptConfig = {
+        messages: []
+      }
+
+      const result = buildResponseFormat(promptConfig)
+      expect(result).toBeUndefined()
+    })
+
+    it('should throw error for invalid JSON schema', () => {
+      const promptConfig: PromptConfig = {
+        messages: [],
+        responseFormat: 'json_schema',
+        jsonSchema: 'invalid json'
+      }
+
+      expect(() => buildResponseFormat(promptConfig)).toThrow(
+        'Invalid JSON schema'
+      )
+    })
+  })
+
+  describe('buildInferenceRequest', () => {
+    it('should build complete inference request from prompt config', () => {
+      const promptConfig: PromptConfig = {
+        messages: [
+          { role: 'system', content: 'System message' },
+          { role: 'user', content: 'User message' }
+        ],
+        responseFormat: 'json_schema',
+        jsonSchema: JSON.stringify({
+          name: 'test_schema',
+          schema: { type: 'object' }
+        })
+      }
+
+      const result = buildInferenceRequest(
+        promptConfig,
+        undefined,
+        undefined,
+        'gpt-4',
+        100,
+        'https://api.test.com',
+        'test-token'
+      )
+
+      expect(result).toEqual({
+        messages: [
+          { role: 'system', content: 'System message' },
+          { role: 'user', content: 'User message' }
+        ],
+        modelName: 'gpt-4',
+        maxTokens: 100,
+        endpoint: 'https://api.test.com',
+        token: 'test-token',
+        responseFormat: {
+          type: 'json_schema',
+          json_schema: {
+            name: 'test_schema',
+            schema: { type: 'object' }
+          }
+        }
+      })
+    })
+
+    it('should build inference request from legacy format', () => {
+      const result = buildInferenceRequest(
+        undefined,
+        'System prompt',
+        'User prompt',
+        'gpt-4',
+        100,
+        'https://api.test.com',
+        'test-token'
+      )
+
+      expect(result).toEqual({
+        messages: [
+          { role: 'system', content: 'System prompt' },
+          { role: 'user', content: 'User prompt' }
+        ],
+        modelName: 'gpt-4',
+        maxTokens: 100,
+        endpoint: 'https://api.test.com',
+        token: 'test-token',
+        responseFormat: undefined
+      })
+    })
+  })
+})
--- a/tests/inference.test.ts
+++ b/tests/inference.test.ts
@@ -33,8 +33,10 @@ const { simpleInference, mcpInference } = await import('../src/inference.js')

 describe('inference.ts', () => {
  const mockRequest = {
-    systemPrompt: 'You are a test assistant',
-    prompt: 'Hello, AI!',
+    messages: [
+      { role: 'system', content: 'You are a test assistant' },
+      { role: 'user', content: 'Hello, AI!' }
+    ],
    modelName: 'gpt-4',
    maxTokens: 100,
    endpoint: 'https://api.test.com',
--- a/tests/main-prompt-integration.test.ts
+++ b/tests/main-prompt-integration.test.ts
@@ -0,0 +1,182 @@
+import { describe, it, expect, beforeEach, jest } from '@jest/globals'
+import * as core from '../__fixtures__/core.js'
+
+// Create fs mocks
+const mockExistsSync = jest.fn()
+const mockReadFileSync = jest.fn()
+const mockWriteFileSync = jest.fn()
+
+// Create inference mocks
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
+const mockSimpleInference = jest.fn() as jest.MockedFunction<any>
+const mockMcpInference = jest.fn()
+
+// Create MCP mocks
+const mockConnectToGitHubMCP = jest.fn()
+
+// Mock fs module
+jest.unstable_mockModule('fs', () => ({
+  existsSync: mockExistsSync,
+  readFileSync: mockReadFileSync,
+  writeFileSync: mockWriteFileSync
+}))
+
+// Mock the inference functions
+jest.unstable_mockModule('../src/inference.js', () => ({
+  simpleInference: mockSimpleInference,
+  mcpInference: mockMcpInference
+}))
+
+// Mock the MCP connection
+jest.unstable_mockModule('../src/mcp.js', () => ({
+  connectToGitHubMCP: mockConnectToGitHubMCP
+}))
+
+jest.unstable_mockModule('@actions/core', () => core)
+
+// The module being tested should be imported dynamically. This ensures that the
+// mocks are used in place of any actual dependencies.
+const { run } = await import('../src/main.js')
+
+describe('main.ts - prompt.yml integration', () => {
+  beforeEach(() => {
+    jest.clearAllMocks()
+
+    // Mock environment variables
+    process.env['GITHUB_TOKEN'] = 'test-token'
+
+    // Mock core.getInput to return appropriate values
+    core.getInput.mockImplementation((name: string) => {
+      switch (name) {
+        case 'model':
+          return 'openai/gpt-4o'
+        case 'max-tokens':
+          return '200'
+        case 'endpoint':
+          return 'https://models.github.ai/inference'
+        case 'enable-github-mcp':
+          return 'false'
+        default:
+          return ''
+      }
+    })
+
+    // Mock core.getBooleanInput
+    const mockGetBooleanInput = core.getBooleanInput as jest.Mock
+    mockGetBooleanInput.mockReturnValue(false)
+
+    // Mock fs.readFileSync for prompt file
+    mockReadFileSync.mockReturnValue(`
+messages:
+  - role: system
+    content: Be as concise as possible
+  - role: user
+    content: 'Compare {{a}} and {{b}}, please'
+model: openai/gpt-4o
+    `)
+
+    // Mock fs.writeFileSync
+    mockWriteFileSync.mockImplementation(() => {})
+
+    // Mock simpleInference
+    mockSimpleInference.mockResolvedValue('Mocked AI response')
+  })
+
+  it('should handle prompt YAML files with template variables', async () => {
+    mockExistsSync.mockReturnValue(true)
+    core.getInput.mockImplementation((name: string) => {
+      switch (name) {
+        case 'prompt-file':
+          return 'test.prompt.yml'
+        case 'input':
+          return 'a: cats\nb: dogs'
+        case 'model':
+          return 'openai/gpt-4o'
+        case 'max-tokens':
+          return '200'
+        case 'endpoint':
+          return 'https://models.github.ai/inference'
+        case 'enable-github-mcp':
+          return 'false'
+        default:
+          return ''
+      }
+    })
+
+    await run()
+
+    // Verify simpleInference was called with the correct message structure
+    expect(mockSimpleInference).toHaveBeenCalledWith(
+      expect.objectContaining({
+        messages: [
+          {
+            role: 'system',
+            content: 'Be as concise as possible'
+          },
+          {
+            role: 'user',
+            content: 'Compare cats and dogs, please'
+          }
+        ],
+        modelName: 'openai/gpt-4o',
+        maxTokens: 200,
+        endpoint: 'https://models.github.ai/inference',
+        token: 'test-token'
+      })
+    )
+
+    // Verify outputs were set
+    expect(core.setOutput).toHaveBeenCalledWith(
+      'response',
+      'Mocked AI response'
+    )
+    expect(core.setOutput).toHaveBeenCalledWith(
+      'response-file',
+      expect.any(String)
+    )
+  })
+
+  it('should fall back to legacy format when not using prompt YAML', async () => {
+    mockExistsSync.mockReturnValue(false)
+    core.getInput.mockImplementation((name: string) => {
+      switch (name) {
+        case 'prompt':
+          return 'Hello, world!'
+        case 'system-prompt':
+          return 'You are helpful'
+        case 'model':
+          return 'openai/gpt-4o'
+        case 'max-tokens':
+          return '200'
+        case 'endpoint':
+          return 'https://models.github.ai/inference'
+        case 'enable-github-mcp':
+          return 'false'
+        default:
+          return ''
+      }
+    })
+
+    await run()
+
+    // Verify simpleInference was called with converted message format
+    expect(mockSimpleInference).toHaveBeenCalledWith(
+      expect.objectContaining({
+        messages: [
+          {
+            role: 'system',
+            content: 'You are helpful'
+          },
+          {
+            role: 'user',
+            content: 'Hello, world!'
+          }
+        ],
+        modelName: 'openai/gpt-4o',
+        maxTokens: 200,
+        endpoint: 'https://models.github.ai/inference',
+        token: 'test-token'
+      })
+    )
+  })
+})
--- a/tests/main.test.ts
+++ b/tests/main.test.ts
@@ -161,12 +161,15 @@ describe('main.ts', () => {
    await run()

    expect(mockSimpleInference).toHaveBeenCalledWith({
-      systemPrompt: 'You are a test assistant.',
-      prompt: 'Hello, AI!',
+      messages: [
+        { role: 'system', content: 'You are a test assistant.' },
+        { role: 'user', content: 'Hello, AI!' }
+      ],
      modelName: 'gpt-4',
      maxTokens: 100,
      endpoint: 'https://api.test.com',
-      token: 'fake-token'
+      token: 'fake-token',
+      responseFormat: undefined
    })
    expect(mockConnectToGitHubMCP).not.toHaveBeenCalled()
    expect(mockMcpInference).not.toHaveBeenCalled()
@@ -193,8 +196,10 @@ describe('main.ts', () => {
    expect(mockConnectToGitHubMCP).toHaveBeenCalledWith('fake-token')
    expect(mockMcpInference).toHaveBeenCalledWith(
      expect.objectContaining({
-        systemPrompt: 'You are a test assistant.',
-        prompt: 'Hello, AI!',
+        messages: [
+          { role: 'system', content: 'You are a test assistant.' },
+          { role: 'user', content: 'Hello, AI!' }
+        ],
        token: 'fake-token'
      }),
      mockMcpClient
@@ -243,12 +248,15 @@ describe('main.ts', () => {
    await run()

    expect(mockSimpleInference).toHaveBeenCalledWith({
-      systemPrompt: systemPromptContent,
-      prompt: promptContent,
+      messages: [
+        { role: 'system', content: systemPromptContent },
+        { role: 'user', content: promptContent }
+      ],
      modelName: 'gpt-4',
      maxTokens: 100,
      endpoint: 'https://api.test.com',
-      token: 'fake-token'
+      token: 'fake-token',
+      responseFormat: undefined
    })
    verifyStandardResponse()
  })
--- a/tests/prompt.test.ts
+++ b/tests/prompt.test.ts
@@ -0,0 +1,133 @@
+import { describe, it, expect } from '@jest/globals'
+import * as path from 'path'
+import { fileURLToPath } from 'url'
+import {
+  parseTemplateVariables,
+  replaceTemplateVariables,
+  loadPromptFile,
+  isPromptYamlFile
+} from '../src/prompt'
+
+const __filename = fileURLToPath(import.meta.url)
+const __dirname = path.dirname(__filename)
+
+describe('prompt.ts', () => {
+  describe('parseTemplateVariables', () => {
+    it('should parse simple YAML variables', () => {
+      const input = `
+a: hello
+b: world
+      `
+      const result = parseTemplateVariables(input)
+      expect(result).toEqual({ a: 'hello', b: 'world' })
+    })
+
+    it('should parse multiline variables', () => {
+      const input = `
+var1: hello
+var2: |
+  This is a
+  multiline string
+      `
+      const result = parseTemplateVariables(input)
+      expect(result.var1).toBe('hello')
+      expect(result.var2).toContain('This is a')
+      expect(result.var2).toContain('multiline string')
+    })
+
+    it('should return empty object for empty input', () => {
+      const result = parseTemplateVariables('')
+      expect(result).toEqual({})
+    })
+
+    it('should throw error for invalid YAML', () => {
+      const input = 'invalid: yaml: content:'
+      expect(() => parseTemplateVariables(input)).toThrow()
+    })
+  })
+
+  describe('replaceTemplateVariables', () => {
+    it('should replace simple variables', () => {
+      const text = 'Hello {{name}}, welcome to {{place}}!'
+      const variables = { name: 'John', place: 'GitHub' }
+      const result = replaceTemplateVariables(text, variables)
+      expect(result).toBe('Hello John, welcome to GitHub!')
+    })
+
+    it('should leave unreplaced variables as is', () => {
+      const text = 'Hello {{name}}, welcome to {{unknown}}!'
+      const variables = { name: 'John' }
+      const result = replaceTemplateVariables(text, variables)
+      expect(result).toBe('Hello John, welcome to {{unknown}}!')
+    })
+
+    it('should handle no variables', () => {
+      const text = 'No variables here'
+      const variables = {}
+      const result = replaceTemplateVariables(text, variables)
+      expect(result).toBe('No variables here')
+    })
+  })
+
+  describe('isPromptYamlFile', () => {
+    it('should detect .prompt.yml files', () => {
+      expect(isPromptYamlFile('test.prompt.yml')).toBe(true)
+      expect(isPromptYamlFile('path/to/test.prompt.yml')).toBe(true)
+    })
+
+    it('should detect .prompt.yaml files', () => {
+      expect(isPromptYamlFile('test.prompt.yaml')).toBe(true)
+      expect(isPromptYamlFile('path/to/test.prompt.yaml')).toBe(true)
+    })
+
+    it('should reject other file types', () => {
+      expect(isPromptYamlFile('test.txt')).toBe(false)
+      expect(isPromptYamlFile('test.yml')).toBe(false)
+      expect(isPromptYamlFile('test.yaml')).toBe(false)
+      expect(isPromptYamlFile('test.prompt')).toBe(false)
+    })
+  })
+
+  describe('loadPromptFile', () => {
+    it('should load simple prompt file', () => {
+      const filePath = path.join(
+        __dirname,
+        '../__fixtures__/prompts/simple.prompt.yml'
+      )
+      const variables = { a: 'cats', b: 'dogs' }
+      const result = loadPromptFile(filePath, variables)
+
+      expect(result.messages).toHaveLength(2)
+      expect(result.messages[0]).toEqual({
+        role: 'system',
+        content: 'Be as concise as possible'
+      })
+      expect(result.messages[1]).toEqual({
+        role: 'user',
+        content: 'Compare cats and dogs, please'
+      })
+      expect(result.model).toBe('openai/gpt-4o')
+    })
+
+    it('should load JSON schema prompt file', () => {
+      const filePath = path.join(
+        __dirname,
+        '../__fixtures__/prompts/json-schema.prompt.yml'
+      )
+      const variables = { animal: 'dog' }
+      const result = loadPromptFile(filePath, variables)
+
+      expect(result.messages).toHaveLength(2)
+      expect(result.messages[1].content).toContain('Describe a dog')
+      expect(result.responseFormat).toBe('json_schema')
+      expect(result.jsonSchema).toBeDefined()
+      expect(result.jsonSchema).toContain('describe_animal')
+    })
+
+    it('should throw error for non-existent file', () => {
+      expect(() => loadPromptFile('non-existent.prompt.yml')).toThrow(
+        'Prompt file not found'
+      )
+    })
+  })
+})
--- a/action.yml
+++ b/action.yml
@@ -14,7 +14,13 @@ inputs:
    required: false
    default: ''
  prompt-file:
-    description: Path to a file containing the prompt
+    description:
+      Path to a file containing the prompt (supports .txt and .prompt.yml
+      formats)
+    required: false
+    default: ''
+  input:
+    description: Template variables in YAML format for .prompt.yml files
    required: false
    default: ''
  model:
--- a/dist/index.js
+++ b/dist/index.js
--- a/dist/index.js.map
+++ b/dist/index.js.map
--- a/package-lock.json
+++ b/package-lock.json
@@ -12,6 +12,8 @@
        "@actions/core": "^1.11.1",
        "@modelcontextprotocol/sdk": "^1.15.1",
        "@rollup/plugin-json": "^6.1.0",
+        "@types/js-yaml": "^4.0.9",
+        "js-yaml": "^4.1.0",
        "pkce-challenge": "^5.0.0"
      },
      "devDependencies": {
@@ -4407,6 +4409,12 @@
        "pretty-format": "^29.0.0"
      }
    },
+    "node_modules/@types/js-yaml": {
+      "version": "4.0.9",
+      "resolved": "https://registry.npmjs.org/@types/js-yaml/-/js-yaml-4.0.9.tgz",
+      "integrity": "sha512-k4MGaQl5TGo/iipqb2UDG2UwjXziSWkh0uysQelTlJpX1qGlpUZYm8PnO4DxG1qBomtJUdYJ6qR6xdIah10JLg==",
+      "license": "MIT"
+    },
    "node_modules/@types/json-schema": {
      "version": "7.0.15",
      "resolved": "https://registry.npmjs.org/@types/json-schema/-/json-schema-7.0.15.tgz",
@@ -5173,7 +5181,6 @@
      "version": "2.0.1",
      "resolved": "https://registry.npmjs.org/argparse/-/argparse-2.0.1.tgz",
      "integrity": "sha512-8+9WqebbFzpX9OR+Wa6O29asIogeRMzcGtAINdpMHHyAg10f05aSFVBbcEqGf/PXw1EjAZ+q2/bEBg3DvurK3Q==",
-      "dev": true,
      "license": "Python-2.0"
    },
    "node_modules/array-buffer-byte-length": {
@@ -10254,7 +10261,6 @@
      "version": "4.1.0",
      "resolved": "https://registry.npmjs.org/js-yaml/-/js-yaml-4.1.0.tgz",
      "integrity": "sha512-wpxZs9NoxZaJESJGIZTyDEaYpl0FKSA+FB9aJiyemKhMwkxQg63h4T1KJgUGHpTqPDNRcmmYLugrRjJlBtWvRA==",
-      "dev": true,
      "license": "MIT",
      "dependencies": {
        "argparse": "^2.0.1"
--- a/package.json
+++ b/package.json
@@ -40,6 +40,8 @@
    "@actions/core": "^1.11.1",
    "@modelcontextprotocol/sdk": "^1.15.1",
    "@rollup/plugin-json": "^6.1.0",
+    "@types/js-yaml": "^4.0.9",
+    "js-yaml": "^4.1.0",
    "pkce-challenge": "^5.0.0"
  },
  "devDependencies": {
--- a/src/helpers.ts
+++ b/src/helpers.ts
@@ -1,6 +1,8 @@
 import * as core from '@actions/core'
 import { GetChatCompletionsDefaultResponse } from '@azure-rest/ai-inference'
 import * as fs from 'fs'
+import { PromptConfig } from './prompt.js'
+import { InferenceRequest } from './inference.js'

 /**
 * Helper function to load content from a file or use fallback input
@@ -64,3 +66,79 @@ export function handleUnexpectedResponse(
        : JSON.stringify(response.body))
  )
 }
+
+/**
+ * Build messages array from either prompt config or legacy format
+ */
+export function buildMessages(
+  promptConfig?: PromptConfig,
+  systemPrompt?: string,
+  prompt?: string
+): Array<{ role: string; content: string }> {
+  if (promptConfig?.messages && promptConfig.messages.length > 0) {
+    // Use new message format
+    return promptConfig.messages.map((msg) => ({
+      role: msg.role,
+      content: msg.content
+    }))
+  } else {
+    // Use legacy format
+    return [
+      {
+        role: 'system',
+        content: systemPrompt || 'You are a helpful assistant'
+      },
+      { role: 'user', content: prompt || '' }
+    ]
+  }
+}
+
+/**
+ * Build response format object for API from prompt config
+ */
+export function buildResponseFormat(
+  promptConfig?: PromptConfig
+): { type: 'json_schema'; json_schema: unknown } | undefined {
+  if (
+    promptConfig?.responseFormat === 'json_schema' &&
+    promptConfig.jsonSchema
+  ) {
+    try {
+      const schema = JSON.parse(promptConfig.jsonSchema)
+      return {
+        type: 'json_schema',
+        json_schema: schema
+      }
+    } catch (error) {
+      throw new Error(
+        `Invalid JSON schema: ${error instanceof Error ? error.message : 'Unknown error'}`
+      )
+    }
+  }
+  return undefined
+}
+
+/**
+ * Build complete InferenceRequest from prompt config and inputs
+ */
+export function buildInferenceRequest(
+  promptConfig: PromptConfig | undefined,
+  systemPrompt: string | undefined,
+  prompt: string | undefined,
+  modelName: string,
+  maxTokens: number,
+  endpoint: string,
+  token: string
+): InferenceRequest {
+  const messages = buildMessages(promptConfig, systemPrompt, prompt)
+  const responseFormat = buildResponseFormat(promptConfig)
+
+  return {
+    messages,
+    modelName,
+    maxTokens,
+    endpoint,
+    token,
+    responseFormat
+  }
+}
--- a/src/inference.ts
+++ b/src/inference.ts
@@ -1,16 +1,30 @@
 import * as core from '@actions/core'
 import ModelClient, { isUnexpected } from '@azure-rest/ai-inference'
 import { AzureKeyCredential } from '@azure/core-auth'
-import { GitHubMCPClient, executeToolCalls } from './mcp.js'
+import { GitHubMCPClient, executeToolCalls, MCPTool, ToolCall } from './mcp.js'
 import { handleUnexpectedResponse } from './helpers.js'

+interface ChatMessage {
+  role: string
+  content: string | null
+  tool_calls?: ToolCall[]
+}
+
+interface ChatCompletionsRequestBody {
+  messages: ChatMessage[]
+  max_tokens: number
+  model: string
+  response_format?: { type: 'json_schema'; json_schema: unknown }
+  tools?: MCPTool[]
+}
+
 export interface InferenceRequest {
-  systemPrompt: string
-  prompt: string
+  messages: Array<{ role: string; content: string }>
  modelName: string
  maxTokens: number
  endpoint: string
  token: string
+  responseFormat?: { type: 'json_schema'; json_schema: unknown } // Processed response format for the API
 }

 export interface InferenceResponse {
@@ -41,18 +55,17 @@ export async function simpleInference(
    }
  )

-  const requestBody = {
-    messages: [
-      {
-        role: 'system',
-        content: request.systemPrompt
-      },
-      { role: 'user', content: request.prompt }
-    ],
+  const requestBody: ChatCompletionsRequestBody = {
+    messages: request.messages,
    max_tokens: request.maxTokens,
    model: request.modelName
  }

+  // Add response format if specified
+  if (request.responseFormat) {
+    requestBody.response_format = request.responseFormat
+  }
+
  const response = await client.path('/chat/completions').post({
    body: requestBody
  })
@@ -84,14 +97,8 @@ export async function mcpInference(
    }
  )

-  // Start with the initial conversation
-  const messages = [
-    {
-      role: 'system',
-      content: request.systemPrompt
-    },
-    { role: 'user', content: request.prompt }
-  ]
+  // Start with the pre-processed messages
+  const messages: ChatMessage[] = [...request.messages]

  let iterationCount = 0
  const maxIterations = 5 // Prevent infinite loops
@@ -100,13 +107,18 @@ export async function mcpInference(
    iterationCount++
    core.info(`MCP inference iteration ${iterationCount}`)

-    const requestBody = {
+    const requestBody: ChatCompletionsRequestBody = {
      messages: messages,
      max_tokens: request.maxTokens,
      model: request.modelName,
      tools: githubMcpClient.tools
    }

+    // Add response format if specified (only on first iteration to avoid conflicts)
+    if (iterationCount === 1 && request.responseFormat) {
+      requestBody.response_format = request.responseFormat
+    }
+
    const response = await client.path('/chat/completions').post({
      body: requestBody
    })
--- a/src/main.ts
+++ b/src/main.ts
@@ -3,8 +3,14 @@ import * as fs from 'fs'
 import * as os from 'os'
 import * as path from 'path'
 import { connectToGitHubMCP } from './mcp.js'
-import { simpleInference, mcpInference, InferenceRequest } from './inference.js'
-import { loadContentFromFileOrInput } from './helpers.js'
+import { simpleInference, mcpInference } from './inference.js'
+import { loadContentFromFileOrInput, buildInferenceRequest } from './helpers.js'
+import {
+  loadPromptFile,
+  parseTemplateVariables,
+  isPromptYamlFile,
+  PromptConfig
+} from './prompt.js'

 const RESPONSE_FILE = 'modelResponse.txt'

@@ -15,16 +21,37 @@ const RESPONSE_FILE = 'modelResponse.txt'
 */
 export async function run(): Promise<void> {
  try {
-    const prompt = loadContentFromFileOrInput('prompt-file', 'prompt')
+    const promptFilePath = core.getInput('prompt-file')
+    const inputVariables = core.getInput('input')

-    const systemPrompt = loadContentFromFileOrInput(
-      'system-prompt-file',
-      'system-prompt',
-      'You are a helpful assistant'
-    )
+    let promptConfig: PromptConfig | undefined = undefined
+    let systemPrompt: string | undefined = undefined
+    let prompt: string | undefined = undefined

-    const modelName: string = core.getInput('model')
-    const maxTokens: number = parseInt(core.getInput('max-tokens'), 10)
+    // Check if we're using a prompt YAML file
+    if (promptFilePath && isPromptYamlFile(promptFilePath)) {
+      core.info('Using prompt YAML file format')
+
+      // Parse template variables
+      const templateVariables = parseTemplateVariables(inputVariables)
+
+      // Load and process prompt file
+      promptConfig = loadPromptFile(promptFilePath, templateVariables)
+    } else {
+      // Use legacy format
+      core.info('Using legacy prompt format')
+
+      prompt = loadContentFromFileOrInput('prompt-file', 'prompt')
+      systemPrompt = loadContentFromFileOrInput(
+        'system-prompt-file',
+        'system-prompt',
+        'You are a helpful assistant'
+      )
+    }
+
+    // Get common parameters
+    const modelName = promptConfig?.model || core.getInput('model')
+    const maxTokens = parseInt(core.getInput('max-tokens'), 10)

    const token = process.env['GITHUB_TOKEN'] || core.getInput('token')
    if (token === undefined) {
@@ -32,21 +59,24 @@ export async function run(): Promise<void> {
    }

    const endpoint = core.getInput('endpoint')
-    const enableMcp = core.getBooleanInput('enable-github-mcp') || false

-    const inferenceRequest: InferenceRequest = {
+    // Build the inference request with pre-processed messages and response format
+    const inferenceRequest = buildInferenceRequest(
+      promptConfig,
      systemPrompt,
      prompt,
      modelName,
      maxTokens,
      endpoint,
      token
-    }
+    )
+
+    const enableMcp = core.getBooleanInput('enable-github-mcp') || false

    let modelResponse: string | null = null

    if (enableMcp) {
-      const mcpClient = await connectToGitHubMCP(token)
+      const mcpClient = await connectToGitHubMCP(inferenceRequest.token)

      if (mcpClient) {
        modelResponse = await mcpInference(inferenceRequest, mcpClient)
--- a/src/prompt.ts
+++ b/src/prompt.ts
@@ -0,0 +1,111 @@
+import * as core from '@actions/core'
+import * as fs from 'fs'
+import * as yaml from 'js-yaml'
+
+export interface PromptMessage {
+  role: 'system' | 'user' | 'assistant'
+  content: string
+}
+
+export interface PromptConfig {
+  messages: PromptMessage[]
+  model?: string
+  responseFormat?: 'text' | 'json_schema'
+  jsonSchema?: string
+}
+
+export interface TemplateVariables {
+  [key: string]: string
+}
+
+/**
+ * Parse template variables from YAML input string
+ */
+export function parseTemplateVariables(input: string): TemplateVariables {
+  if (!input.trim()) {
+    return {}
+  }
+
+  try {
+    const parsed = yaml.load(input) as TemplateVariables
+    if (typeof parsed !== 'object' || parsed === null) {
+      throw new Error('Template variables must be a YAML object')
+    }
+    return parsed
+  } catch (error) {
+    throw new Error(
+      `Failed to parse template variables: ${error instanceof Error ? error.message : 'Unknown error'}`
+    )
+  }
+}
+
+/**
+ * Replace template variables in text using {{variable}} syntax
+ */
+export function replaceTemplateVariables(
+  text: string,
+  variables: TemplateVariables
+): string {
+  return text.replace(/\{\{([\w.-]+)\}\}/g, (match, variableName) => {
+    if (variableName in variables) {
+      return variables[variableName]
+    }
+    core.warning(
+      `Template variable '${variableName}' not found in input variables`
+    )
+    return match // Return the original placeholder if variable not found
+  })
+}
+
+/**
+ * Load and parse a prompt YAML file with template variable substitution
+ */
+export function loadPromptFile(
+  filePath: string,
+  templateVariables: TemplateVariables = {}
+): PromptConfig {
+  if (!fs.existsSync(filePath)) {
+    throw new Error(`Prompt file not found: ${filePath}`)
+  }
+
+  const fileContent = fs.readFileSync(filePath, 'utf-8')
+
+  // Apply template variable substitution
+  const processedContent = replaceTemplateVariables(
+    fileContent,
+    templateVariables
+  )
+
+  try {
+    const config = yaml.load(processedContent) as PromptConfig
+
+    if (!config.messages || !Array.isArray(config.messages)) {
+      throw new Error('Prompt file must contain a "messages" array')
+    }
+
+    // Validate messages
+    for (const message of config.messages) {
+      if (!message.role || !message.content) {
+        throw new Error(
+          'Each message must have "role" and "content" properties'
+        )
+      }
+      if (!['system', 'user', 'assistant'].includes(message.role)) {
+        throw new Error(`Invalid message role: ${message.role}`)
+      }
+    }
+
+    return config
+  } catch (error) {
+    throw new Error(
+      `Failed to parse prompt file: ${error instanceof Error ? error.message : 'Unknown error'}`
+    )
+  }
+}
+
+/**
+ * Check if a file is a prompt YAML file based on extension
+ */
+export function isPromptYamlFile(filePath: string): boolean {
+  return filePath.endsWith('.prompt.yml') || filePath.endsWith('.prompt.yaml')
+}
Author	SHA1	Message	Date
Sean Goedecke	9693b137b6	Merge pull request #61 from actions/sgoedecke/prompt-file Support .prompt.yml files	2025-07-21 15:14:53 +10:00
Sean Goedecke	d0b2f23c43	Merge branch 'main' into sgoedecke/prompt-file	2025-07-21 15:01:31 +10:00
Sean Goedecke	0df96479bc	Merge pull request #60 from actions/sgoedecke/update-readme Update readme to say MCP needs a PAT	2025-07-21 15:00:56 +10:00
Sean Goedecke	446f075e3b	Merge branch 'main' into sgoedecke/update-readme	2025-07-21 15:00:09 +10:00
Sean Goedecke	ce58b26ac7	Merge pull request #59 from actions/sgoedecke-patch-1 Add GitHub Actions workflow for releasing new version	2025-07-21 14:57:40 +10:00
Sean Goedecke	1cf96b0212	Merge branch 'main' into sgoedecke/update-readme	2025-07-21 14:57:19 +10:00
Sean Goedecke	f79e4e11cb	regenerate dist	2025-07-21 04:56:41 +00:00
Sean Goedecke	72102e50bf	Update src/prompt.ts Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>	2025-07-21 14:55:49 +10:00
Sean Goedecke	2bc30a525a	regenerate dist	2025-07-21 04:35:30 +00:00
Sean Goedecke	8f64ac1284	Fixup types and tests	2025-07-21 04:31:06 +00:00
Sean Goedecke	1f89e942aa	licensed cache	2025-07-21 04:25:01 +00:00
Sean Goedecke	77a7cbe11b	Merge branch 'main' into sgoedecke-patch-1	2025-07-21 14:18:05 +10:00
Sean Goedecke	fcc8550115	lint	2025-07-21 04:01:22 +00:00
Sean Goedecke	29b5f08d0f	Update README.md Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>	2025-07-21 13:50:38 +10:00
Sean Goedecke	6f7bd88d1d	lint	2025-07-21 03:49:57 +00:00
Yuzuki	8ae5306787	Merge pull request #62 from actions/dependabot/github_actions/super-linter/super-linter-8.0.0 Bump super-linter/super-linter from 7.4.0 to 8.0.0	2025-07-21 13:13:50 +10:00
dependabot[bot]	9445295106	Bump super-linter/super-linter from 7.4.0 to 8.0.0 Bumps [super-linter/super-linter](https://github.com/super-linter/super-linter) from 7.4.0 to 8.0.0. - [Release notes](https://github.com/super-linter/super-linter/releases) - [Changelog](https://github.com/super-linter/super-linter/blob/main/CHANGELOG.md) - [Commits](`12150456a7...5119dcd801`) --- updated-dependencies: - dependency-name: super-linter/super-linter dependency-version: 8.0.0 dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com>	2025-07-21 02:52:40 +00:00
Sean Goedecke	e385879671	Merge branch 'main' into sgoedecke/prompt-file	2025-07-21 00:21:07 +00:00
Sean Goedecke	1780121e3b	Support .prompt.yml files	2025-07-21 00:11:26 +00:00
Sean Goedecke	b002da2928	Add GitHub Actions workflow for releasing new version	2025-07-21 09:41:47 +10:00
Sean Goedecke	ba509f9275	update readme	2025-07-16 07:51:13 +00:00