Add commit job with script-only hooks for test and format validation

Introduces a new bespoke commit job for this repo with three steps:
test, format, and reconcile_and_push. The job validates code quality
before pushing changes by running pytest and ruff checks.

Uses script hooks (not prompt hooks) for the test and format steps,
with guidance on interpreting hook output included in the step
instructions. This simpler approach lets the agent handle results
directly rather than requiring a separate prompt-based evaluator.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Author: nhorton

.claude/commands/add_platform.md

@@ -46,26 +46,29 @@ This job has 4 step(s):
 
 ## Instructions
 
-Determine what the user wants to do and route to the appropriate step.
+This is a **multi-step workflow**. Determine the starting point and run through the steps in sequence.
 
 1. **Analyze user intent** from the text that follows `/add_platform`
 
-2. **Match intent to a step**:
+2. **Identify the starting step** based on intent:
    - research: Capture CLI configuration and hooks system documentation for the new platform
    - add_capabilities: Update job schema and adapters with any new hook events the platform supports
    - implement: Add platform adapter, templates, tests with 100% coverage, and README documentation
    - verify: Set up platform directories and verify deepwork install works correctly
 
-3. **Invoke the matched step** using the Skill tool:
-   ```
-   Skill: <step_command_name>
-   ```
+3. **Run the workflow** starting from the identified step:
+   - Invoke the starting step using the Skill tool
+   - When that step completes, **automatically continue** to the next step in the workflow
+   - Continue until the workflow is complete or the user intervenes
 
-4. **If intent is ambiguous**, ask the user which step they want:
+4. **If intent is ambiguous**, ask the user which step to start from:
    - Present the available steps as numbered options
    - Use AskUserQuestion to let them choose
 
-**Critical**: You MUST invoke the step using the Skill tool. Do not copy/paste the step's instructions. The Skill tool invocation ensures the step's quality validation hooks fire.
+**Critical**:
+- You MUST invoke each step using the Skill tool. Do not copy/paste step instructions.
+- After each step completes, check if there's a next step and invoke it automatically.
+- The workflow continues until all dependent steps are complete.
 
 ## Context Files
 

.claude/commands/commit.md

@@ -0,0 +1,73 @@
+---
+description: Validate, format, and push changes with tests passing
+---
+
+# commit
+
+You are executing the **commit** job. Validate, format, and push changes with tests passing
+
+A pre-commit workflow that ensures code quality before pushing changes.
+
+This job runs through three validation and preparation steps:
+1. Runs the test suite and fixes any failures until all tests pass (max 5 attempts)
+2. Runs ruff formatting and linting, fixing issues until clean (max 5 attempts)
+3. Fetches from remote, rebases if needed, generates a simple commit message,
+   commits changes, and pushes to the remote branch
+
+Each step uses a quality validation loop to ensure it completes successfully
+before moving to the next step. The format step runs as a subagent to
+minimize token usage.
+
+Key behaviors:
+- Rebase strategy when remote has changes (keeps linear history)
+- Simple summary commit messages (no conventional commits format)
+- Maximum 5 fix attempts before stopping
+
+Designed for developers who want a reliable pre-push workflow that catches
+issues early and ensures consistent code quality.
+
+
+## Available Steps
+
+This job has 3 step(s):
+
+### test
+**Run Tests**: Run pytest and fix any failures until all tests pass (max 5 attempts)
+- Command: `uw.commit.test`
+### format
+**Format Code**: Run ruff formatting and linting, fix issues until clean (max 5 attempts, runs as subagent)
+- Command: `uw.commit.format`
+- Requires: test
+### reconcile_and_push
+**Reconcile and Push**: Fetch remote, rebase if needed, commit with simple summary message, and push
+- Command: `uw.commit.reconcile_and_push`
+- Requires: format
+
+## Instructions
+
+This is a **multi-step workflow**. Determine the starting point and run through the steps in sequence.
+
+1. **Analyze user intent** from the text that follows `/commit`
+
+2. **Identify the starting step** based on intent:
+   - test: Run pytest and fix any failures until all tests pass (max 5 attempts)
+   - format: Run ruff formatting and linting, fix issues until clean (max 5 attempts, runs as subagent)
+   - reconcile_and_push: Fetch remote, rebase if needed, commit with simple summary message, and push
+
+3. **Run the workflow** starting from the identified step:
+   - Invoke the starting step using the Skill tool
+   - When that step completes, **automatically continue** to the next step in the workflow
+   - Continue until the workflow is complete or the user intervenes
+
+4. **If intent is ambiguous**, ask the user which step to start from:
+   - Present the available steps as numbered options
+   - Use AskUserQuestion to let them choose
+
+**Critical**:
+- You MUST invoke each step using the Skill tool. Do not copy/paste step instructions.
+- After each step completes, check if there's a next step and invoke it automatically.
+- The workflow continues until all dependent steps are complete.
+
+## Context Files
+
+- Job definition: `.deepwork/jobs/commit/job.yml`

.claude/commands/deepwork_jobs.learn.md

@@ -9,10 +9,8 @@ hooks:
 
             ## Quality Criteria
 
-            **AGENT: TAKE ACTION** - Verify the learning process meets ALL quality criteria before completing:
-
-            1. **Conversation Analyzed**: Did you review the conversation for DeepWork job executions?
-            2. **Confusion Identified**: Did you identify points of confusion, errors, or inefficiencies?
+            1. **Conversation Analyzed**: Did the agent review the conversation for DeepWork job executions?
+            2. **Confusion Identified**: Did the agent identify points of confusion, errors, or inefficiencies?
             3. **Instructions Improved**: Were job instructions updated to address identified issues?
             4. **Instructions Concise**: Are instructions free of redundancy and unnecessary verbosity?
             5. **Shared Content Extracted**: Is lengthy/duplicated content extracted into referenced files?
@@ -22,10 +20,6 @@ hooks:
             9. **Generalizable Separated**: Are generalizable improvements in instructions, not AGENTS.md?
             10. **Sync Complete**: Has `deepwork sync` been run if instructions were modified?
 
-            If ANY criterion is not met, continue working to address it.
-            If ALL criteria are satisfied, include `<promise>✓ Quality Criteria Met</promise>` in your response.
-
-
             ## Instructions
 
             Review the conversation and determine if ALL quality criteria above have been satisfied.
@@ -34,8 +28,8 @@ hooks:
             If the agent has included `<promise>✓ Quality Criteria Met</promise>` in their response AND
             all criteria appear to be met, respond with: {"ok": true}
 
-            If criteria are NOT met AND the promise tag is missing, respond with:
-            {"ok": false, "reason": "Continue working. [specific feedback on what's wrong]"}
+            If criteria are NOT met OR the promise tag is missing, respond with:
+            {"ok": false, "reason": "**AGENT: TAKE ACTION** - [which criteria failed and why]"}
 ---
 
 # deepwork_jobs.learn
@@ -386,10 +380,9 @@ Ensure all outputs are:
 This step uses an iterative quality validation loop. After completing your work, stop hook(s) will evaluate whether the outputs meet quality criteria. If criteria are not met, you will be prompted to continue refining.
 
 ### Quality Criteria
-**AGENT: TAKE ACTION** - Verify the learning process meets ALL quality criteria before completing:
 
-1. **Conversation Analyzed**: Did you review the conversation for DeepWork job executions?
-2. **Confusion Identified**: Did you identify points of confusion, errors, or inefficiencies?
+1. **Conversation Analyzed**: Did the agent review the conversation for DeepWork job executions?
+2. **Confusion Identified**: Did the agent identify points of confusion, errors, or inefficiencies?
 3. **Instructions Improved**: Were job instructions updated to address identified issues?
 4. **Instructions Concise**: Are instructions free of redundancy and unnecessary verbosity?
 5. **Shared Content Extracted**: Is lengthy/duplicated content extracted into referenced files?
@@ -399,9 +392,6 @@ This step uses an iterative quality validation loop. After completing your work,
 9. **Generalizable Separated**: Are generalizable improvements in instructions, not AGENTS.md?
 10. **Sync Complete**: Has `deepwork sync` been run if instructions were modified?
 
-If ANY criterion is not met, continue working to address it.
-If ALL criteria are satisfied, include `<promise>✓ Quality Criteria Met</promise>` in your response.
-
 
 ### Completion Promise
 

.claude/commands/deepwork_jobs.md

@@ -35,25 +35,28 @@ This job has 3 step(s):
 
 ## Instructions
 
-Determine what the user wants to do and route to the appropriate step.
+This is a **multi-step workflow**. Determine the starting point and run through the steps in sequence.
 
 1. **Analyze user intent** from the text that follows `/deepwork_jobs`
 
-2. **Match intent to a step**:
+2. **Identify the starting step** based on intent:
    - define: Create the job.yml specification file by understanding workflow requirements
    - implement: Generate instruction files for each step based on the job.yml specification
    - learn: Reflect on conversation to improve job instructions and capture learnings
 
-3. **Invoke the matched step** using the Skill tool:
-   ```
-   Skill: <step_command_name>
-   ```
+3. **Run the workflow** starting from the identified step:
+   - Invoke the starting step using the Skill tool
+   - When that step completes, **automatically continue** to the next step in the workflow
+   - Continue until the workflow is complete or the user intervenes
 
-4. **If intent is ambiguous**, ask the user which step they want:
+4. **If intent is ambiguous**, ask the user which step to start from:
    - Present the available steps as numbered options
    - Use AskUserQuestion to let them choose
 
-**Critical**: You MUST invoke the step using the Skill tool. Do not copy/paste the step's instructions. The Skill tool invocation ensures the step's quality validation hooks fire.
+**Critical**:
+- You MUST invoke each step using the Skill tool. Do not copy/paste step instructions.
+- After each step completes, check if there's a next step and invoke it automatically.
+- The workflow continues until all dependent steps are complete.
 
 ## Context Files
 

.claude/commands/deepwork_policy.md

@@ -33,23 +33,26 @@ This job has 1 step(s):
 
 ## Instructions
 
-Determine what the user wants to do and route to the appropriate step.
+This is a **multi-step workflow**. Determine the starting point and run through the steps in sequence.
 
 1. **Analyze user intent** from the text that follows `/deepwork_policy`
 
-2. **Match intent to a step**:
+2. **Identify the starting step** based on intent:
    - define: Create or update policy entries in .deepwork.policy.yml
 
-3. **Invoke the matched step** using the Skill tool:
-   ```
-   Skill: <step_command_name>
-   ```
+3. **Run the workflow** starting from the identified step:
+   - Invoke the starting step using the Skill tool
+   - When that step completes, **automatically continue** to the next step in the workflow
+   - Continue until the workflow is complete or the user intervenes
 
-4. **If intent is ambiguous**, ask the user which step they want:
+4. **If intent is ambiguous**, ask the user which step to start from:
    - Present the available steps as numbered options
    - Use AskUserQuestion to let them choose
 
-**Critical**: You MUST invoke the step using the Skill tool. Do not copy/paste the step's instructions. The Skill tool invocation ensures the step's quality validation hooks fire.
+**Critical**:
+- You MUST invoke each step using the Skill tool. Do not copy/paste step instructions.
+- After each step completes, check if there's a next step and invoke it automatically.
+- The workflow continues until all dependent steps are complete.
 
 ## Context Files
 

.claude/commands/update.md

@@ -31,23 +31,26 @@ This job has 1 step(s):
 
 ## Instructions
 
-Determine what the user wants to do and route to the appropriate step.
+This is a **multi-step workflow**. Determine the starting point and run through the steps in sequence.
 
 1. **Analyze user intent** from the text that follows `/update`
 
-2. **Match intent to a step**:
+2. **Identify the starting step** based on intent:
    - job: Edit standard job source files and sync to installed locations
 
-3. **Invoke the matched step** using the Skill tool:
-   ```
-   Skill: <step_command_name>
-   ```
+3. **Run the workflow** starting from the identified step:
+   - Invoke the starting step using the Skill tool
+   - When that step completes, **automatically continue** to the next step in the workflow
+   - Continue until the workflow is complete or the user intervenes
 
-4. **If intent is ambiguous**, ask the user which step they want:
+4. **If intent is ambiguous**, ask the user which step to start from:
    - Present the available steps as numbered options
    - Use AskUserQuestion to let them choose
 
-**Critical**: You MUST invoke the step using the Skill tool. Do not copy/paste the step's instructions. The Skill tool invocation ensures the step's quality validation hooks fire.
+**Critical**:
+- You MUST invoke each step using the Skill tool. Do not copy/paste step instructions.
+- After each step completes, check if there's a next step and invoke it automatically.
+- The workflow continues until all dependent steps are complete.
 
 ## Context Files
 

.claude/commands/uw.add_platform.add_capabilities.md

@@ -9,17 +9,12 @@ hooks:
 
             ## Quality Criteria
 
-            **AGENT: TAKE ACTION** - Verify the capability additions meet ALL criteria:
-            1. Any new hooks from the platform (for slash commands only) are added to src/deepwork/schemas/job_schema.py
-            2. All existing adapters in src/deepwork/adapters.py are updated with the new hook fields
-               (set to None/null if the platform doesn't support that hook)
-            3. Only hooks available on slash command definitions are added (not general CLI hooks)
-            4. job_schema.py remains valid Python with no syntax errors
-            5. adapters.py remains consistent - all adapters have the same hook fields
-            6. If no new hooks are needed, document why in a comment
-
-            If ALL criteria are met, include `<promise>✓ Quality Criteria Met</promise>`.
-
+            1. Are any new hooks from the platform (for slash commands only) added to src/deepwork/schemas/job_schema.py?
+            2. Are all existing adapters in src/deepwork/adapters.py updated with the new hook fields? (set to None/null if the platform doesn't support that hook)
+            3. Are only hooks available on slash command definitions added (not general CLI hooks)?
+            4. Does job_schema.py remain valid Python with no syntax errors?
+            5. Is adapters.py consistent - all adapters have the same hook fields?
+            6. If no new hooks are needed, is there documentation explaining why?
 
             ## Instructions
 
@@ -29,8 +24,8 @@ hooks:
             If the agent has included `<promise>✓ Quality Criteria Met</promise>` in their response AND
             all criteria appear to be met, respond with: {"ok": true}
 
-            If criteria are NOT met AND the promise tag is missing, respond with:
-            {"ok": false, "reason": "Continue working. [specific feedback on what's wrong]"}
+            If criteria are NOT met OR the promise tag is missing, respond with:
+            {"ok": false, "reason": "**AGENT: TAKE ACTION** - [which criteria failed and why]"}
 ---
 
 # add_platform.add_capabilities
@@ -242,16 +237,13 @@ Ensure all outputs are:
 This step uses an iterative quality validation loop. After completing your work, stop hook(s) will evaluate whether the outputs meet quality criteria. If criteria are not met, you will be prompted to continue refining.
 
 ### Quality Criteria
-**AGENT: TAKE ACTION** - Verify the capability additions meet ALL criteria:
-1. Any new hooks from the platform (for slash commands only) are added to src/deepwork/schemas/job_schema.py
-2. All existing adapters in src/deepwork/adapters.py are updated with the new hook fields
-   (set to None/null if the platform doesn't support that hook)
-3. Only hooks available on slash command definitions are added (not general CLI hooks)
-4. job_schema.py remains valid Python with no syntax errors
-5. adapters.py remains consistent - all adapters have the same hook fields
-6. If no new hooks are needed, document why in a comment
-
-If ALL criteria are met, include `<promise>✓ Quality Criteria Met</promise>`.
+
+1. Are any new hooks from the platform (for slash commands only) added to src/deepwork/schemas/job_schema.py?
+2. Are all existing adapters in src/deepwork/adapters.py updated with the new hook fields? (set to None/null if the platform doesn't support that hook)
+3. Are only hooks available on slash command definitions added (not general CLI hooks)?
+4. Does job_schema.py remain valid Python with no syntax errors?
+5. Is adapters.py consistent - all adapters have the same hook fields?
+6. If no new hooks are needed, is there documentation explaining why?
 
 
 ### Completion Promise