kata-execute-quick-task（kata 执行快速任务）

Quick mode is the same system with a shorter path:

• Spawns kata-planner (quick mode) + kata-executor(s)
• Skips kata-phase-researcher, kata-plan-checker, kata-verifier
• Quick tasks live in .planning/quick/ separate from planned phases
• Updates STATE.md "Quick Tasks Completed" table (NOT ROADMAP.md)

Use when: You know exactly what to do and the task is small enough to not need research or verification.

<execution_context> Orchestration is inline - no separate workflow file. Quick mode is deliberately simpler than full Kata. </execution_context>

Read model profile for agent spawning:

MODEL_PROFILE=$(node scripts/kata-lib.cjs read-config "model_profile" "balanced")

Default to "balanced" if not set.

Model lookup table:

Store resolved models for use in Task calls below.

Step 1: Pre-flight validation

Check that an active Kata project exists:

if [ ! -f .planning/ROADMAP.md ]; then
  echo "Quick mode requires an active project with ROADMAP.md."
  echo "Run /kata-new-project first."
  exit 1
fi

If validation fails, stop immediately with the error message.

Quick tasks can run mid-phase - validation only checks ROADMAP.md exists, not phase status.

Step 1.5: Parse issue argument (optional)

Check for issue file path argument:

ISSUE_FILE=""
ISSUE_NUMBER=""
ISSUE_TITLE=""
ISSUE_PROBLEM=""

# Check for --issue flag
if echo "$ARGUMENTS" | grep -q -- "--issue"; then
  ISSUE_FILE=$(echo "$ARGUMENTS" | grep -oE '\-\-issue [^ ]+' | cut -d' ' -f2)

  if [ -f "$ISSUE_FILE" ]; then
    # Extract issue metadata
    ISSUE_TITLE=$(grep "^title:" "$ISSUE_FILE" | cut -d':' -f2- | xargs)
    PROVENANCE=$(grep "^provenance:" "$ISSUE_FILE" | cut -d' ' -f2)

    if echo "$PROVENANCE" | grep -q "^github:"; then
      ISSUE_NUMBER=$(echo "$PROVENANCE" | grep -oE '#[0-9]+' | tr -d '#')
    fi

    # Extract problem section for context
    ISSUE_PROBLEM=$(sed -n '/^## Problem/,/^## /p' "$ISSUE_FILE" | tail -n +2 | head -n -1)
  fi
fi

If ISSUE_FILE provided but file not found: error and exit. If ISSUE_FILE provided and valid: Use issue title as description (skip Step 2 prompt).

Step 2: Get task description

If $ISSUE_FILE is set (issue-driven quick task):

DESCRIPTION="$ISSUE_TITLE"
echo "Using issue title: $DESCRIPTION"

Skip the AskUserQuestion prompt — use the issue title as the description.

If no issue context (standard quick task):

Prompt user interactively for the task description:

AskUserQuestion(
  header: "Quick Task",
  question: "What do you want to do?",
  followUp: null
)

Store response as $DESCRIPTION.

If empty, re-prompt: "Please provide a task description."

Generate slug from description (both paths):

Generate slug from description:

slug=$(echo "$DESCRIPTION" | tr '[:upper:]' '[:lower:]' | sed 's/[^a-z0-9]/-/g' | sed 's/--*/-/g' | sed 's/^-//;s/-$//' | cut -c1-40)

Step 3: Calculate next quick task number

Ensure .planning/quick/ directory exists and find the next sequential number:

# Ensure .planning/quick/ exists
mkdir -p .planning/quick

# Find highest existing number and increment
last=$(ls -1d .planning/quick/[0-9][0-9][0-9]-* 2>/dev/null | sort -r | head -1 | xargs -I{} basename {} | grep -oE '^[0-9]+')

if [ -z "$last" ]; then
  next_num="001"
else
  next_num=$(printf "%03d" $((10#$last + 1)))
fi

Step 4: Create quick task directory

Create the directory for this quick task:

QUICK_DIR=".planning/quick/${next_num}-${slug}"
mkdir -p "$QUICK_DIR"

Report to user:

Creating quick task ${next_num}: ${DESCRIPTION}
Directory: ${QUICK_DIR}

Store $QUICK_DIR for use in orchestration.

Step 4.5: Read context files

Read files before spawning agents using the Read tool. The @ syntax does not work across Task() boundaries - content must be inlined.

Read these files:

• .planning/STATE.md (required) — store as STATE_CONTENT
• skills/kata-plan-phase/references/planner-instructions.md (cross-skill reference) — store as planner_instructions_content
• skills/kata-execute-phase/references/executor-instructions.md (cross-skill reference) — store as executor_instructions_content

Store content for use in Task prompts below.

Step 5: Spawn planner (quick mode)

Spawn kata-planner with quick mode context:

Task(
  prompt="<agent-instructions>\n{planner_instructions_content}\n</agent-instructions>\n\n" +
"<planning_context>

**Mode:** quick
**Directory:** ${QUICK_DIR}
**Description:** ${DESCRIPTION}

**Project State:**
${STATE_CONTENT}

**Issue Context (if from issue):**
${ISSUE_FILE:+Issue File: $ISSUE_FILE}
${ISSUE_NUMBER:+GitHub Issue: #$ISSUE_NUMBER}
${ISSUE_PROBLEM:+
Problem:
$ISSUE_PROBLEM}

</planning_context>

<constraints>
- Create a SINGLE plan with 1-3 focused tasks
- Quick tasks should be atomic and self-contained
- No research phase, no checker phase
- Target ~30% context usage (simple, focused)
</constraints>

<output>
Write plan to: ${QUICK_DIR}/${next_num}-PLAN.md
Return: ## PLANNING COMPLETE with plan path
</output>
",
  subagent_type="general-purpose",
  model="{planner_model}",
  description="Quick plan: ${DESCRIPTION}"
)

After planner returns:

1. Verify plan exists at ${QUICK_DIR}/${next_num}-PLAN.md
2. Extract plan count (typically 1 for quick tasks)
3. Report: "Plan created: ${QUICK_DIR}/${next_num}-PLAN.md"

If plan not found, error: "Planner failed to create ${next_num}-PLAN.md"

Step 6: Spawn executor

Read the plan content before spawning using the Read tool:

• ${QUICK_DIR}/${next_num}-PLAN.md

Spawn kata-executor with inlined plan (use the STATE_CONTENT from step 4.5):

Task(
  prompt="<agent-instructions>\n{executor_instructions_content}\n</agent-instructions>\n\n" +
"Execute quick task ${next_num}.

<plan>
${PLAN_CONTENT}
</plan>

<project_state>
${STATE_CONTENT}
</project_state>

<constraints>
- Execute all tasks in the plan
- Commit each task atomically
- Create summary at: ${QUICK_DIR}/${next_num}-SUMMARY.md
- Do NOT update ROADMAP.md (quick tasks are separate from planned phases)
</constraints>
",
  subagent_type="general-purpose",
  model="{executor_model}",
  description="Execute: ${DESCRIPTION}"
)

After executor returns:

1. Verify summary exists at ${QUICK_DIR}/${next_num}-SUMMARY.md
2. Extract commit hash from executor output
3. Report completion status

If summary not found, error: "Executor failed to create ${next_num}-SUMMARY.md"

Note: For quick tasks producing multiple plans (rare), spawn executors in parallel waves per phase-execute patterns.

Step 7: Update STATE.md

Update STATE.md with quick task completion record.

7a. Check if "Quick Tasks Completed" section exists:

Read STATE.md and check for ### Quick Tasks Completed section.

7b. If section doesn't exist, create it:

Insert after ### Blockers/Concerns section:

### Quick Tasks Completed

| #   | Description | Date | Commit | Directory |
| --- | ----------- | ---- | ------ | --------- |

7c. Append new row to table:

| ${next_num} | ${DESCRIPTION} | $(date +%Y-%m-%d) | ${commit_hash} | [${next_num}-${slug}](./quick/${next_num}-${slug}/) |

7d. Update "Last activity" line:

Find and update the line:

Last activity: $(date +%Y-%m-%d) - Completed quick task ${next_num}: ${DESCRIPTION}

Use Edit tool to make these changes atomically

Step 7.5: Create PR with issue closure (if issue-driven)

Check pr_workflow config:

PR_WORKFLOW=$(node scripts/kata-lib.cjs read-config "pr_workflow" "false")

If $ISSUE_NUMBER is set (issue-driven quick task):

If PR_WORKFLOW=true:

# Create branch for this quick task
BRANCH="fix/quick-${next_num}-${slug}"
git checkout -b "$BRANCH"

# Push branch
git push -u origin "$BRANCH"

# Build PR body
cat > /tmp/quick-pr-body.md << PR_EOF
## Summary

Completes issue #${ISSUE_NUMBER}: ${ISSUE_TITLE}

## Changes

Quick task ${next_num}: ${DESCRIPTION}

See: ${QUICK_DIR}/${next_num}-SUMMARY.md

Closes #${ISSUE_NUMBER}
PR_EOF

# Create PR
gh pr create \
  --title "fix: ${ISSUE_TITLE}" \
  --body-file /tmp/quick-pr-body.md

echo "PR created with Closes #${ISSUE_NUMBER}"

If PR_WORKFLOW=false:

# Close issue directly (no PR workflow)
gh issue close "$ISSUE_NUMBER" --comment "Completed via quick task ${next_num}"
echo "Issue #${ISSUE_NUMBER} closed directly (pr_workflow=false)"

If no $ISSUE_NUMBER: Skip PR creation (standard quick task, not issue-driven).

Step 8: Final commit and completion

Stage and commit quick task artifacts:

# Stage quick task artifacts
git add ${QUICK_DIR}/${next_num}-PLAN.md
git add ${QUICK_DIR}/${next_num}-SUMMARY.md
git add .planning/STATE.md

# Commit with quick task format
git commit -m "$(cat <<'EOF'
docs(quick-${next_num}): ${DESCRIPTION}

Quick task completed.

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

Get final commit hash:

commit_hash=$(git rev-parse --short HEAD)

Display completion output:

---

Kata > QUICK TASK COMPLETE

Quick Task ${next_num}: ${DESCRIPTION}

Summary: ${QUICK_DIR}/${next_num}-SUMMARY.md
Commit: ${commit_hash}

---

Ready for next task: /kata-execute-quick-task

<success_criteria>

• ROADMAP.md validation passes
• User provides task description (or uses issue title if --issue flag)
• Slug generated (lowercase, hyphens, max 40 chars)
• Next number calculated (001, 002, 003...)
• Directory created at .planning/quick/NNN-slug/
• ${next_num}-PLAN.md created by planner
• ${next_num}-SUMMARY.md created by executor
• STATE.md updated with quick task row
• Artifacts committed
• Issue context parsed from --issue flag (if provided)
• PR created with Closes #X (if issue-driven and pr_workflow=true)
• Issue closed directly (if issue-driven and pr_workflow=false) </success_criteria>