Tweaks to safe outputs feature

.github/workflows/test-claude-add-issue-labels.md

@@ -0,0 +1,18 @@
+---
+on:
+  issues:
+    types: [opened]
+  reaction: eyes
+
+engine: 
+  id: claude
+
+permissions:
+  issues: read
+
+safe-outputs:
+  add-issue-labels:
+---
+
+Add the issue labels "quack" and "dog" to the issue.
+

.github/workflows/test-claude-create-issue.md

@@ -0,0 +1,19 @@
+---
+on:
+  workflow_dispatch:
+
+engine: 
+  id: claude
+
+permissions:
+  issues: read
+
+safe-outputs:
+  create-issue:
+    title-prefix: "[claude-test] "
+    labels: [claude, automation, haiku]
+---
+
+Create an issue with title "Hello" and body "World"
+
+Add a haiku about GitHub Actions and AI to the issue body.

.github/workflows/test-claude-create-pull-request.md

@@ -0,0 +1,18 @@
+---
+on:
+  workflow_dispatch:
+
+engine: 
+  id: claude
+
+safe-outputs:
+  create-pull-request:
+    title-prefix: "[claude-test] "
+    labels: [claude, automation, bot]
+---
+
+Add a file "TEST.md" with content "Hello, World!"
+
+Create a pull request with title "Hello" and body "World"
+
+Add a haiku about GitHub Actions and AI to the PR body.

.github/workflows/test-claude.md

@@ -11,22 +11,13 @@ on:
 engine: 
   id: claude
   model: claude-3-5-sonnet-20241022
-  permissions:
-    network:
-      allowed: []
 timeout_minutes: 10
 permissions:
   pull-requests: write
   actions: read
   contents: read
-output:
-  labels:
-    allowed: ["bug", "feature"]
-  issue:
-    title-prefix: "[claude-test] "
-    labels: [claude, automation, haiku]
-  issue_comment: {}
-  pull-request:
+safe-outputs:
+  create-pull-request:
     title-prefix: "[claude-test] "
     labels: [claude, automation, bot]
 tools:
@@ -126,14 +117,6 @@ Your comment should include:
 
 **Current Context**: You have access to the current pull request content via: "${{ needs.task.outputs.text }}"
 
-### Action Output: Create a Haiku
-
-**IMPORTANT**: After completing your PR analysis and posting your comment, please create a haiku about the changes you analyzed and write it to the action output. The haiku should capture the essence of the pull request in a creative and poetic way.
-
-Write your haiku to the file "${{ env.GITHUB_AW_OUTPUT }}" (use the `Write` tool). This will make it available as a workflow output that other jobs can access.
-
-Make your haiku relevant to the specific changes you analyzed in this PR. Be creative and thoughtful in your poetic interpretation of the code changes.
-
 ### Additional Task: Random Quote Generation
 
 **IMPORTANT**: After creating your haiku, please generate a random inspirational quote about software development, coding, or technology and append it to a new file called "quote.md". 

.github/workflows/test-proxy.md

@@ -4,8 +4,8 @@ on:
     branches: [ "main" ]
   workflow_dispatch:
 
-output:
-  issue_comment: {}
+safe-outputs:
+  add-issue-comment:
 
 tools:
   fetch:
@@ -21,7 +21,8 @@ tools:
 
   github:
     allowed:
-      - "get_issue"
+      - "create_issue"
+      - "create_comment"
 
 engine: claude
 runs-on: ubuntu-latest
@@ -46,10 +47,10 @@ Test the MCP network permissions feature to validate that domain restrictions ar
 
 If there are any failures, security issues, or unexpected behaviors:
 
-- Write a detailed report to ${{ env.GITHUB_AW_OUTPUT }} documenting:
-- Which domains were successfully accessed vs blocked
-- Error messages received for blocked domains  
-- Any security observations or recommendations
-- Specific failure details that need attention
+- Write a detailed report documenting:
+  - Which domains were successfully accessed vs blocked
+  - Error messages received for blocked domains  
+  - Any security observations or recommendations
+  - Specific failure details that need attention
 
-The test results will be automatically posted as a comment on PR #${{ github.event.pull_request.number }}.
+Post the test results as an issue comment on PR #${{ github.event.pull_request.number }}.

docs/frontmatter.md

@@ -21,7 +21,7 @@ The YAML frontmatter supports standard GitHub Actions properties plus additional
 - `engine`: AI engine configuration (claude/codex) with optional max-turns setting and network permissions
 - `tools`: Available tools and MCP servers for the AI engine  
 - `cache`: Cache configuration for workflow dependencies
-- `output`: [Safe Output Processing](safe-outputs.md) for automatic issue creation and comment posting.
+- `safe-outputs`: [Safe Output Processing](safe-outputs.md) for automatic issue creation and comment posting.
 
 ## Trigger Events (`on:`)
 
@@ -281,9 +281,9 @@ engine:
            - "*.safe-domain.org"
    ```
 
-## Output Configuration (`output:`)
+## Safe Outputs Configuration (`safe-outputs:`)
 
-See [Safe Output Processing](safe-outputs.md) for automatic issue creation and comment posting.
+See [Safe Outputs Processing](safe-outputs.md) for automatic issue creation, comment posting and other safe outputs.
 
 ## Run Configuration (`run-name:`, `runs-on:`, `timeout_minutes:`)