Research Backlog Task
This skill covers creating backlog tasks when research reveals actionable improvements to semantic-kit.
Context
For research documentation structure and conventions, see research/_meta/README.md.
Prerequisites
Before using this skill:
- •Research findings are documented in research pages
- •Research has been released (version created in
research/CHANGELOG.md) - •You have identified a specific, actionable tool improvement
Decision Tree
code
Should I create a backlog task? ├── Research reveals tool could better match real-world behavior → YES ├── Finding suggests new feature, flag, or output format → YES ├── Documentation suggests tool should handle something differently → YES ├── Research contradicts assumptions built into current tool → YES ├── Finding is purely informational (no tool change) → NO ├── Change is trivial (just mention in conversation) → NO └── Uncertain if actionable → Discuss first, then decide
Creating the Task
Use the Backlog.md MCP tools. Never edit markdown files directly.
Required Fields
code
title: Clear, action-oriented title description: Full context including research findings labels: ["research-backed", ...domain labels] references: [research page paths, research changelog section] acceptanceCriteria: Specific, testable criteria
Task Structure Template
Use this structure for the description field:
markdown
## Research Context **Source:** [[research-page-name]] (research-vX.Y.Z) **Finding:** Summary of what was discovered during research. Include the key insight that suggests a tool change. **Key Citations:** - Citation 1: Brief description of what it supports - Citation 2: Brief description of what it supports ## Proposed Change **Affected command(s):** `command-name` (or "new command") **What should change:** Describe the user-visible change. What will be different? What problem does it solve? ## Implementation Approach **Key files likely involved:** - `src/commands/example.ts` - Brief note on what changes - `src/lib/example.ts` - Brief note on what changes **Approach:** General strategy without full code. What patterns to follow? What libraries to use? **Considerations:** - Edge cases to handle - Backwards compatibility concerns - Performance implications
Standard Labels
Always include research-backed for research-originated tasks.
Domain labels (reverse domain notation):
- •
lens-ai,lens-reader,lens-google,lens-social,lens-screen-reader - •
utility-schema,utility-structure,utility-readability,utility-sitemap - •
validate-html,validate-schema,validate-a11y
Type labels:
- •
feature- New functionality - •
enhancement- Improvement to existing functionality - •
fix- Bug fix - •
refactor- Code improvement without behavior change
References Field
Link to research documentation:
javascript
references: [ "research/topics/content-extraction/sitemaps.md", "research/CHANGELOG.md#research-v050" ]
Acceptance Criteria
Always include these standard items for research-backed tasks:
javascript
acceptanceCriteria: [ // Task-specific criteria first "Command validates XML sitemap structure", "Reports URL count and file size", // ... other specific criteria // Standard research-backed criteria (always include) "Research page updated with toolCoverage entry", "CHANGELOG entry references research page and version" ]
Example Task Creation
javascript
// Using Backlog.md MCP
task_create({
title: "Add sitemap command for validation and inspection",
description: `## Research Context
**Source:** [[sitemaps]] (research-v0.5.0)
**Finding:**
XML sitemaps are critical for search engine crawling but developers often create invalid sitemaps. Research revealed that Google ignores <changefreq> and <priority> — only <lastmod> matters. Common errors include encoding issues, relative URLs, and exceeding size limits.
## Proposed Change
**Affected command(s):** New \`sitemap\` command
**What should change:**
Add a new command that validates and inspects XML sitemaps, reporting URL count, file size, extension usage, and validation errors.
## Implementation Approach
**Key files:**
- \`src/commands/sitemap.ts\` (new)
- \`src/lib/sitemap.ts\` (new)
**Approach:**
Use the sitemap.js library for TypeScript-native parsing. Validate XML structure, check encoding, verify absolute URLs.`,
labels: ["research-backed", "utility-sitemap", "feature"],
priority: "medium",
references: [
"research/topics/seo/sitemaps.md",
"research/CHANGELOG.md#research-v050"
],
acceptanceCriteria: [
"sitemap <url> fetches and parses XML sitemap",
"Validates XML structure and encoding",
"Reports URL count and file size",
"Warns when approaching size limits",
"JSON output includes all validation results",
"Research page updated with toolCoverage entry",
"CHANGELOG entry references research page and version"
]
})
Dependencies
If a task depends on another task being completed first:
javascript
task_create({
// ...
dependencies: ["TASK-016"], // Reference by task ID
// ...
})
After Creating the Task
- •Mention the task ID to the user
- •Summarize what the task covers
- •Note any dependencies or sequencing considerations
Related Skills
- •research-workflow: For performing research before creating tasks
- •finalize-research-task: For developers completing these tasks