Fix test-dispatcher workflow prompt to explicitly call dispatch_workflow tool

[Copilot]

The test-dispatcher workflow configured safe-outputs.dispatch-workflow but never generated safe outputs because the prompt described capability instead of directing action.

Changes

.github/workflows/test-dispatcher.md

• Changed passive description to imperative task statement
• Added explicit tool invocation instructions with JSON example
• Added constraint to prevent file-writing workarounds

Before/After

Before:

This workflow demonstrates the dispatch-workflow safe output capability.
The agent can trigger the test-workflow using the test_workflow tool.

After:

**Your task**: Call the `dispatch_workflow` tool to trigger the `test-workflow` workflow.

**Important**: You MUST use the safe output tool - do NOT write to files or attempt other methods.

## Instructions
1. **Call the safe output tool**: Use `dispatch_workflow` to trigger the test-workflow
2. **Workflow name**: Specify `test-workflow` as the workflow to dispatch
3. **Inputs (optional)**: You can provide test parameters if needed, but they are optional

## Example
```json
{
  "type": "dispatch_workflow",
  "workflow_name": "test-workflow",
  "inputs": {
    "test_param": "example value"
  }
}

This resolves "No Safe Outputs Generated" errors in runs 21547216704, 21547252198, 21547524795, 21547522694, 21547527335, 21547532343, 21547522773.

<!-- START COPILOT ORIGINAL PROMPT -->



<details>

<summary>Original prompt</summary>


----

*This section details on the original issue you should resolve*

<issue_title>Fix test-dispatcher workflow to use safe outputs properly</issue_title>
<issue_description># Q Workflow Optimization Report

## Issue Context

This PR addresses issue githubnext/gh-aw#12971, where the test-dispatcher workflow was repeatedly failing with "No Safe Outputs Generated" errors. The workflow ran successfully but didn't produce any safe outputs, causing downstream processing to fail.

### Failed Run IDs Analyzed
- Run 21547216704 (original issue)
- Run 21547252198
- Run 21547524795
- Run 21547522694
- Run 21547527335
- Run 21547532343
- Run 21547522773

All runs exhibited the same pattern: agent job succeeded but produced no safe outputs.

## Root Cause

The test-dispatcher workflow had a `safe-outputs` configuration with `dispatch-workflow` capability, but **the prompt didn't instruct the agent to actually use it**. The original prompt was:

```markdown
This workflow demonstrates the dispatch-workflow safe output capability.
The agent can trigger the test-workflow using the test_workflow tool.

This was too passive - it described what the agent could do, but didn't tell it what it should do.

Changes Made

.github/workflows/test-dispatcher.md

• Added explicit instructions to call the dispatch_workflow safe output tool
• Added clear task description with specific action required
• Emphasized safe output usage with warning against file writing
• Structured prompt with sections for better clarity

The new prompt explicitly instructs:

**Your task**: Call the `dispatch_workflow` tool to trigger the `test-workflow` workflow.

**Important**: You MUST use the safe output tool - do NOT write to files or attempt other methods.

Expected Improvements

• ✅ Eliminates "No Safe Outputs Generated" errors by providing clear instructions
• ✅ Improves workflow reliability with explicit task definition
• ✅ Better agent guidance with structured prompt format
• ✅ Prevents alternative approaches that don't use safe outputs

Validation

Modified workflow file validated:

• ✅ Frontmatter structure intact
• ✅ Safe output configuration preserved
• ✅ Prompt structure improved with clear task definition
• ✅ No lock files included (will be compiled automatically)

Related

Closes githubnext/gh-aw#12971

AI generated by Q

---

Note

This was originally intended as a pull request, but the git push operation failed.

Workflow Run: View run details and download patch artifact

The patch file is available in the agent-artifacts artifact in the workflow run linked above.

To apply the patch locally:

# Download the artifact from the workflow run https://github.com/githubnext/gh-aw/actions/runs/21547617819
# (Use GitHub MCP tools if gh CLI is not available)
gh run download 21547617819 -n agent-artifacts

# The patch file will be at agent-artifacts/tmp/gh-aw/aw.patch after download
# Apply the patch
git am agent-artifacts/tmp/gh-aw/aw.patch

Show patch (39 lines)

From e3d13165e248fa3f3fd7f9e16929b9c4ced0d04b Mon Sep 17 00:00:00 2001
From: "github-actions[bot]" <github-actions[bot]@users.noreply.github.com>
Date: Sat, 31 Jan 2026 16:50:38 +0000
Subject: [PATCH] fix(test-dispatcher): add explicit safe output instructions

The test-dispatcher workflow was failing because the prompt didn't
instruct the agent to use the dispatch_workflow safe output tool.

Changes:
- Added clear instructions to call dispatch_workflow tool
- Emphasized using safe outputs instead of file writing
- Structured prompt with task and requirements

Fixes githubnext/gh-aw#12971
---
 .github/workflows/test-dispatcher.md | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/test-dispatcher.md b/.github/workflows/test-dispatcher.md
index 824b8a7..6fc9349 100644
--- a/.github/workflows/test-dispatcher.md
+++ b/.github/workflows/test-dispatcher.md
@@ -14,4 +14,11 @@ safe-outputs:
 # Test Dispatcher Workflow
 
 This workflow demonstrates the dispatch-workflow safe output capability.
-The agent can trigger the test-workflow using the test_workflow tool.
+
+## Instructions
+
+You have access to the `dispatch_workflow` safe output tool that can trigger workflows.
+
+**Your task**: Call the `dispatch_workflow` tool to trigger the `test-workflow` workflow.
+
+**Important**: You MUST use the safe output tool - do NOT write to files or attempt other methods.
-- 
2.52.0

Comments on the Issue (you are @copilot in this section)

• Fixes Fix test-dispatcher workflow to use safe outputs properly #12978

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.