Add commit job with quality checks workflow

.deepwork/jobs/commit/job.yml

@@ -49,35 +49,23 @@ steps:
     outputs:
       - code_reviewed  # implicit state: code has been reviewed and issues addressed
     dependencies: []
-    hooks:
-      after_agent:
-        - prompt: |
-            Verify the code review is complete:
-            1. Changed files were identified
-            2. Sub-agent reviewed the code for general issues, DRY opportunities, naming clarity, and test coverage
-            3. All identified issues were addressed or documented as intentional
-            If ALL criteria are met, include `<promise>✓ Quality Criteria Met</promise>`.
+    quality_criteria:
+      - "Changed files were identified"
+      - "Sub-agent reviewed the code for general issues, DRY opportunities, naming clarity, and test coverage"
+      - "All identified issues were addressed or documented as intentional"
 
   - id: test
     name: "Run Tests"
     description: "Pulls latest code and runs tests until all pass. Use after code review passes to verify changes work correctly."
     instructions_file: steps/test.md
-    inputs:
-      - name: test_command
-        description: "Test command to run (optional - will auto-detect if not provided)"
+    inputs: []
     outputs:
       - tests_passing  # implicit state: all tests pass
     dependencies:
       - review
-    hooks:
-      after_agent:
-        - prompt: |
-            Verify the tests are passing:
-            1. Latest code was pulled from the branch
-            2. All tests completed successfully
-            3. No test failures or errors remain
-            4. Test output shows passing status
-            If ALL criteria are met, include `<promise>✓ Quality Criteria Met</promise>`.
+    quality_criteria:
+      - "Latest code was pulled from the branch"
+      - "All tests are passing"
 
   - id: lint
     name: "Lint Code"
@@ -88,14 +76,10 @@ steps:
       - code_formatted  # implicit state: code formatted and linted
     dependencies:
       - test
-    hooks:
-      after_agent:
-        - prompt: |
-            Verify the linting is complete:
-            1. ruff format was run successfully
-            2. ruff check was run successfully (with --fix)
-            3. No remaining lint errors
-            If ALL criteria are met, include `<promise>✓ Quality Criteria Met</promise>`.
+    quality_criteria:
+      - "ruff format was run successfully"
+      - "ruff check was run with --fix flag"
+      - "No remaining lint errors"
 
   - id: commit_and_push
     name: "Commit and Push"
@@ -106,14 +90,9 @@ steps:
       - changes_committed  # implicit state: changes committed and pushed
     dependencies:
       - lint
-    hooks:
-      after_agent:
-        - prompt: |
-            Verify the commit is ready:
-            1. Changed files list was reviewed by the agent
-            2. Files match what was modified during this session (or unexpected changes were investigated)
-            3. CHANGELOG.md was updated with entries in the [Unreleased] section (if changes warrant documentation)
-            4. Version numbers were NOT modified (pyproject.toml version and CHANGELOG version headers must remain unchanged)
-            5. Commit was created with appropriate message
-            6. Changes were pushed to remote
-            If ALL criteria are met, include `<promise>✓ Quality Criteria Met</promise>`.
+    quality_criteria:
+      - "Changed files were verified against expectations"
+      - "CHANGELOG.md was updated with entries in [Unreleased] section (if changes warrant documentation)"
+      - "Version numbers were NOT modified (pyproject.toml version and CHANGELOG version headers unchanged)"
+      - "Commit was created with appropriate message"
+      - "Changes were pushed to remote"

.deepwork/jobs/commit/steps/commit_and_push.md

@@ -78,14 +78,11 @@ Check the list of changed files against what was modified during this session, e
 
 ## Quality Criteria
 
-- Changed files list was reviewed by the agent
-- Files match what was modified during this session (or unexpected changes were investigated and handled)
-- CHANGELOG.md was updated with entries in the `[Unreleased]` section (if changes warrant documentation)
-- Version numbers were NOT modified (in pyproject.toml or CHANGELOG.md version headers)
-- Commit message follows project conventions
-- Commit was created successfully
+- Changed files were verified against expectations
+- CHANGELOG.md was updated with entries in [Unreleased] section (if changes warrant documentation)
+- Version numbers were NOT modified (pyproject.toml version and CHANGELOG version headers unchanged)
+- Commit was created with appropriate message
 - Changes were pushed to remote
-- When all criteria are met, include `<promise>✓ Quality Criteria Met</promise>` in your response
 
 ## Context
 

.deepwork/jobs/commit/steps/lint.md

@@ -63,9 +63,7 @@ Report the results of each command.
 
 - ruff format was run successfully
 - ruff check was run with --fix flag
-- No remaining lint errors (or all are documented and intentional)
-- Sub-agent was used to conserve context
-- When all criteria are met, include `<promise>✓ Quality Criteria Met</promise>` in your response
+- No remaining lint errors
 
 ## Context
 

.deepwork/jobs/commit/steps/review.md

@@ -23,32 +23,11 @@ Use a sub-agent to review the staged/changed code and identify issues that shoul
 
    Use the Task tool with these parameters:
    - `subagent_type`: "general-purpose"
-   - `prompt`: Include the list of changed files and the review criteria below
-
-   The sub-agent should review each changed file for:
-
-   **General Issues**
-   - Logic errors or potential bugs
-   - Error handling gaps
-   - Security concerns
-   - Performance issues
-
-   **DRY Opportunities**
-   - Duplicated code that should be extracted into functions
-   - Repeated patterns that could be abstracted
-   - Copy-pasted logic with minor variations
-
-   **Naming Clarity**
-   - Variables, functions, and classes should have clear, descriptive names
-   - Names should reflect purpose and intent
-   - Avoid abbreviations that aren't universally understood
-   - Consistent naming conventions throughout
-
-   **Test Coverage**
-   - New functions or classes should have corresponding tests
-   - New code paths should be tested
-   - Edge cases should be covered
-   - If tests are missing, note what should be tested
+   - `prompt`: Instruct the sub-agent to:
+     - Read the code review standards from `doc/code_review_standards.md`
+     - Read each of the changed files
+     - Review each file against the standards
+     - Report issues found with file, line number, severity, and suggested fix
 
 3. **Review sub-agent findings**
    - Examine each issue identified
@@ -65,40 +44,11 @@ Use a sub-agent to review the staged/changed code and identify issues that shoul
    - If you made substantial changes, consider running another review pass
    - Ensure fixes didn't introduce new issues
 
-## Example Sub-Agent Prompt
-
-```
-Review the following changed files for code quality issues:
-
-Files to review:
-- src/module.py
-- src/utils.py
-- tests/test_module.py
-
-For each file, check for:
-
-1. **General issues**: Logic errors, bugs, error handling gaps, security concerns
-2. **DRY opportunities**: Duplicated code, repeated patterns that should be extracted
-3. **Naming clarity**: Are variable/function/class names clear and descriptive?
-4. **Test coverage**: Does new functionality have corresponding tests?
-
-Read each file and provide a structured report of issues found, organized by category.
-For each issue, include:
-- File and line number
-- Description of the issue
-- Suggested fix
-
-If no issues are found in a category, state that explicitly.
-```
-
 ## Quality Criteria
 
 - Changed files were identified
-- Sub-agent reviewed all changed files
-- Issues were categorized (general, DRY, naming, tests)
+- Sub-agent read the code review standards and reviewed all changed files
 - All identified issues were addressed or documented as intentional
-- Sub-agent was used to conserve context
-- When all criteria are met, include `<promise>✓ Quality Criteria Met</promise>` in your response
 
 ## Context
 

.deepwork/jobs/commit/steps/test.md

@@ -44,9 +44,7 @@ Execute the test suite for the project and iteratively fix any failures until al
 ## Quality Criteria
 
 - Latest code was pulled from the branch
-- Test command was correctly identified or used from input
-- All tests are now passing
-- When all criteria are met, include `<promise>✓ Quality Criteria Met</promise>` in your response
+- All tests are passing
 
 ## Context
 

doc/code_review_standards.md

@@ -0,0 +1,67 @@
+# Code Review Standards
+
+This document defines the standards used during code review in the commit workflow.
+
+## Review Categories
+
+### General Issues
+
+Check for:
+- Logic errors or potential bugs
+- Error handling gaps
+- Security concerns (injection, authentication, authorization)
+- Performance issues (inefficient algorithms, unnecessary computation)
+- Resource leaks (unclosed files, connections)
+
+### DRY (Don't Repeat Yourself)
+
+Look for:
+- Duplicated code that should be extracted into functions
+- Repeated patterns that could be abstracted
+- Copy-pasted logic with minor variations
+- Similar code blocks that differ only in variable names
+
+### Naming Clarity
+
+Ensure:
+- Variables, functions, and classes have clear, descriptive names
+- Names reflect purpose and intent
+- Abbreviations are avoided unless universally understood
+- Naming conventions are consistent throughout the codebase
+
+### Test Coverage
+
+Verify:
+- New functions or classes have corresponding tests
+- New code paths are tested
+- Edge cases are covered
+- Error conditions are tested
+- If tests are missing, note what should be tested
+
+### Test Quality
+
+Ensure tests add value and are not duplicative:
+- Each test should verify a distinct behavior or scenario
+- Tests should not duplicate what other tests already cover
+- Test names should clearly describe what they're testing
+- Tests should be meaningful, not just checking trivial cases
+- Avoid testing implementation details; focus on behavior
+- If multiple tests appear redundant, suggest consolidation
+
+## Severity Levels
+
+When reporting issues, categorize by severity:
+
+- **Critical**: Must fix before commit (bugs, security issues)
+- **High**: Should fix before commit (logic errors, missing error handling)
+- **Medium**: Recommended to fix (DRY violations, unclear naming)
+- **Low**: Nice to have (style improvements, minor optimizations)
+
+## Review Output Format
+
+For each issue found, provide:
+1. File and line number
+2. Severity level
+3. Category (General/DRY/Naming/Tests)
+4. Description of the issue
+5. Suggested fix or improvement

library/jobs/README.md

@@ -37,6 +37,10 @@ The job definition file contains:
 - `version`: Semantic version (e.g., "1.0.0")
 - `summary`: Brief description (under 200 characters)
 - `description`: Detailed explanation of the job's purpose
+- `workflows`: Named sequences of steps (optional)
+  - `name`: Workflow identifier
+  - `summary`: What the workflow accomplishes
+  - `steps`: Ordered list of step IDs to execute
 - `steps`: Array of step definitions with:
   - `id`: Step identifier
   - `name`: Human-readable step name
@@ -45,6 +49,7 @@ The job definition file contains:
   - `inputs`: What the step requires
   - `outputs`: What the step produces
   - `dependencies`: Other steps that must complete first
+  - `quality_criteria`: Measurable criteria for step completion
 
 ### steps/
 

library/jobs/commit/block_bash_with_instructions.sh

@@ -0,0 +1,74 @@
+#!/bin/bash
+# block_bash_with_instructions.sh - Blocks specific bash commands and provides alternative instructions
+#
+# This hook intercepts Bash tool use calls and blocks commands that match
+# specific patterns, providing alternative instructions to the agent.
+#
+# Usage: Registered as a PreToolUse hook in .claude/settings.json
+#
+# Input (stdin): JSON from Claude Code hook system containing tool_name and tool_input
+# Output (stderr): Error message if blocked (Claude Code reads stderr for exit code 2)
+# Exit codes:
+#   0 - Success (allow action)
+#   2 - Blocking error (prevent action with message)
+
+set -e
+
+# =============================================================================
+# BLOCKED COMMANDS CONFIGURATION
+# =============================================================================
+# Format: Each entry is a regex pattern followed by a delimiter (|||) and instructions
+# The regex is matched against the full bash command
+# Add new blocked commands here:
+
+BLOCKED_COMMANDS=(
+    '^[[:space:]]*git[[:space:]]+commit|||All commits must be done via the `/commit` skill. Do not use git commit directly. Instead, run `/commit` to start the commit workflow which includes code review, testing, and linting before committing.'
+)
+
+# =============================================================================
+# HOOK LOGIC - DO NOT MODIFY BELOW UNLESS NECESSARY
+# =============================================================================
+
+# Read stdin into variable
+HOOK_INPUT=""
+if [ ! -t 0 ]; then
+    HOOK_INPUT=$(cat)
+fi
+
+# Exit early if no input
+if [ -z "${HOOK_INPUT}" ]; then
+    exit 0
+fi
+
+# Extract tool_name from input
+TOOL_NAME=$(echo "${HOOK_INPUT}" | jq -r '.tool_name // empty' 2>/dev/null)
+
+# Only process Bash tool calls
+if [ "${TOOL_NAME}" != "Bash" ]; then
+    exit 0
+fi
+
+# Extract the command from tool_input
+COMMAND=$(echo "${HOOK_INPUT}" | jq -r '.tool_input.command // empty' 2>/dev/null)
+
+# Exit if no command
+if [ -z "${COMMAND}" ]; then
+    exit 0
+fi
+
+# Check each blocked pattern
+for entry in "${BLOCKED_COMMANDS[@]}"; do
+    # Split entry by delimiter
+    pattern="${entry%%|||*}"
+    instructions="${entry##*|||}"
+
+    # Check if command matches pattern (using extended regex)
+    if echo "${COMMAND}" | grep -qE "${pattern}"; then
+        # Output error message to stderr (Claude Code reads stderr for exit code 2)
+        echo "${instructions}" >&2
+        exit 2
+    fi
+done
+
+# Command is allowed
+exit 0