executing-tasks-from-any-source（从任何来源执行任务）

Executing Tasks From Any Source

Overview

The /project:do:task workflow supports both GitHub issues and Basic Memory artifacts through intelligent source detection. This skill explains the dual-source execution pattern.

Core principle: Detect input format → Fetch from appropriate source → Execute unified workflow → Update both channels.

When to Use This Skill

• Executing tasks from GitHub issues (traditional issue-based workflow)
• Executing tasks from Basic Memory artifacts (Johnny Decimal format: 5.1.1-task-*)
• Switching between task sources within the same project
• Building tools/agents that need to support both task sources

Quick Reference: Source Detection

Detection logic:

if [[ $ARGUMENTS =~ ^#?[0-9]+$ ]]; then
  # GitHub issue - pure digits with optional #
  ISSUE_NUM=$(echo $ARGUMENTS | tr -d '#')
  SOURCE="github"
elif [[ $ARGUMENTS =~ ^[0-9]+(\.[0-9]+)+.* ]]; then
  # Basic Memory artifact - Johnny Decimal format
  ARTIFACT_ID=$ARGUMENTS
  SOURCE="basicmemory"
else
  echo "Error: Invalid task format. Use GitHub issue (#123) or Johnny Decimal artifact (5.1.1-task-name)"
  exit 1
fi

GitHub Issue Execution Path

Step 1: Fetch Issue Details

# Fetch issue using GitHub CLI
gh issue view $ISSUE_NUM --json title,body,author,labels,assignees,linkedIssues

# Extract components:
# - Title: Issue headline
# - Body: Full description with acceptance criteria
# - Author: Original creator
# - Labels: Topic tags (feature, bug, etc)
# - Assignees: Current assignee
# - LinkedIssues: Related/parent issues

Step 2: Worktree Creation

Pass to using-git-worktrees skill:

feature_identifier: "{ISSUE_NUM}-{slug}"
Example: "123-add-user-authentication"

Step 3: Status Updates (In-Progress)

# Add in-progress label
gh issue edit $ISSUE_NUM --add-label "in-progress"

# Add comment
gh issue comment $ISSUE_NUM -b "Implementation started. Tracking in worktree."

# Remove ready/todo labels
gh issue edit $ISSUE_NUM --remove-label "ready,todo"

Step 4: Status Updates (Completed)

# Update labels
gh issue edit $ISSUE_NUM --remove-label "in-progress"
gh issue edit $ISSUE_NUM --add-label "completed"

# Add completion comment with PR link
gh issue comment $ISSUE_NUM -b "Completed in PR #{PR_NUMBER}. See changes above."

Basic Memory Artifact Execution Path

Artifact Structure

Tasks in Basic Memory use this structure:

---
title: "5.1.1-task-user-authentication"
folder: "5-tasks"
entity_type: "task"
entity_metadata:
  tags: ["authentication", "user-system"]
  status: "pending"
  priority: "high"
  epic_id: "2.1.1-epic-auth-system"
  dependencies: []
---

# Task: Implement User Authentication

## Description
Detailed task description...

## Acceptance Criteria
- [ ] Criterion 1
- [ ] Criterion 2

## Dependencies
- depends_on [[2.1.0-epic-auth-system]]

Step 1: Fetch Artifact Details

# Use basicmemory_read_note with Johnny Decimal lookup
# Tool extracts: title, content, frontmatter (status, epic_id, dependencies)
# from 5-tasks folder matching the artifact ID

# Frontmatter fields parsed:
# - status: "pending" | "in-progress" | "completed"
# - epic_id: Parent epic reference (e.g., "2.1.1-epic-auth")
# - priority: "high" | "medium" | "low"
# - dependencies: List of related artifacts

Step 2: Worktree Creation

Pass to using-git-worktrees skill:

feature_identifier: "{JOHNNY_DECIMAL}-{slug}"
Example: "5.1.1-task-user-authentication"

Step 3: Status Updates (In-Progress)

# Use basicmemory_edit_note to update frontmatter
basicmemory_edit_note(
  identifier: "5.1.1-task-auth",
  operation: "find_replace",
  find_text: "status: pending",
  content: "status: in-progress"
)

# Add implementation note
basicmemory_edit_note(
  identifier: "5.1.1-task-auth",
  operation: "append",
  content: "\n## Implementation Log\n- Started implementation in worktree\n"
)

Step 4: Status Updates (Completed)

# Update status to completed
basicmemory_edit_note(
  identifier: "5.1.1-task-auth",
  operation: "find_replace",
  find_text: "status: in-progress",
  content: "status: completed"
)

# Add PR link to artifact
basicmemory_edit_note(
  identifier: "5.1.1-task-auth",
  operation: "append",
  content: "\n- Completed in PR: https://github.com/repo/pulls/123\n"
)

Unified Workflow Steps (Apply to Both Sources)

Once task is fetched from its source, the remaining workflow is identical:

1. Deep Technical Analysis (Step 2)

- Extract user story - Map acceptance criteria - Determine technical scope - Check dependencies - Consider edge cases

1. System Impact Assessment (Step 3)

- Frontend changes needed? - Backend changes needed? - Database migrations? - Testing strategy?

1. Human Validation Check (Step 4)

- Stop and ask if architectural decisions needed - Stop and ask if security/performance critical - Wait for explicit approval

1. Implementation (Steps 6-12)

- Design test strategy - Plan implementation approach - Execute core implementation - Run comprehensive tests - Update documentation - Quality validation

1. Completion (Steps 13-15)

- Update task status (source-specific) - Create pull request linking to task - Monitor code review - After merge: update parent issue/epic

Common Patterns

Pattern 1: Check task dependencies before starting

GitHub:

# Get linked issues that are blockers
gh issue view $ISSUE_NUM --json linkedIssues --jq '.linkedIssues[] | select(.url | contains("blocked"))'

Basic Memory:

# Read the artifact to find depends_on relations
basicmemory_read_note("5.1.1-task-auth")
# Check frontmatter: dependencies: [...] or look for "depends_on [[...]]" in content

Pattern 2: Get parent context

GitHub:

# Find linked parent issue (PRD/Feature)
gh issue view $ISSUE_NUM --json linkedIssues \
  | jq '.linkedIssues[] | select(.title | contains("Feature") or contains("PRD"))'

Basic Memory:

# Fetch parent epic using epic_id from frontmatter
basicmemory_read_note("2.1.1-epic-auth-system")

Pattern 3: Update related items after completion

GitHub:

# Update parent issue with progress
gh issue comment $PARENT_ISSUE_NUM -b "Completed task #{ISSUE_NUM}"

# Link completed task to parent
gh issue edit $ISSUE_NUM --add-label "merged"

Basic Memory:

# Update parent epic
basicmemory_edit_note(
  identifier: "2.1.1-epic-auth",
  operation: "append",
  content: "\n- Completed: [[5.1.1-task-user-authentication]]\n"
)

Error Handling

Invalid Task Format

if [[ ! $ARGUMENTS =~ ^(#?[0-9]+|[0-9]+(\.[0-9]+)+.*)$ ]]; then
  echo "❌ Invalid task format"
  echo "GitHub issue: 123 or #123"
  echo "Basic Memory: 5.1.1-task-name"
  exit 1
fi

GitHub Issue Not Found

if ! gh issue view $ISSUE_NUM 2>/dev/null; then
  echo "❌ GitHub issue #$ISSUE_NUM not found"
  echo "Check the issue number and try again"
  exit 1
fi

Basic Memory Artifact Not Found

if ! basicmemory_read_note($ARTIFACT_ID) 2>/dev/null; then
  echo "❌ Basic Memory artifact $ARTIFACT_ID not found in 5-tasks/"
  echo "Check the artifact ID matches Johnny Decimal format: N.N.N-task-*"
  exit 1
fi

Integration Points

Called by:

• /project:do:task workflow (main entry point)
• Agent task execution systems
• Task queuing systems

Pairs with:

• using-git-worktrees - Creates isolated workspace
• executing-plans - For comprehensive implementation
• subagent-driven-development - For delegated work
• finishing-a-development-branch - For cleanup after completion

Best Practices

1. Always verify task exists before starting workflow
2. Check dependencies before implementing
3. Use consistent worktree naming - helps with resumption
4. Update status in source before creating worktree
5. Document completion in both PR comment and task artifact
6. Keep frontmatter clean - avoid manual formatting in Basic Memory
7. Link related tasks - use relations/dependencies properly

Red Flags

Never:

• Start implementation without fetching task details first
• Ignore dependency warnings
• Proceed with failing tests without asking
• Forget to update task status when switching sources
• Mix GitHub and Basic Memory references inconsistently

Always:

• Detect source format before proceeding
• Fetch complete task context
• Verify dependencies before starting
• Update task status at start and completion
• Link PR to task in both channels