research-entry
Research and maintain comparison dataset entries. Use when the user wants to create a new comparison entry, update an existing entry, research a tool/product for a dataset, fill in [TODO] or incomplete fields in an entry, or add a new item to the awesome-comparisons datasets. Supports all datasets (code-editor, terminal, aie-model, code-agent, product-prototyping, other, business-competition).
SKILL.md
| Name | research-entry |
| Description | Research and maintain comparison dataset entries. Use when the user wants to create a new comparison entry, update an existing entry, research a tool/product for a dataset, fill in [TODO] or incomplete fields in an entry, or add a new item to the awesome-comparisons datasets. Supports all datasets (code-editor, terminal, aie-model, code-agent, product-prototyping, other, business-competition). |
name: research-entry description: Research and maintain comparison dataset entries. Use when the user wants to create a new comparison entry, update an existing entry, research a tool/product for a dataset, fill in [TODO] or incomplete fields in an entry, or add a new item to the awesome-comparisons datasets. Supports all datasets (code-editor, terminal, aie-model, code-agent, product-prototyping, other, business-competition).
Research Entry
Research and maintain structured markdown entries across comparison datasets.
Modes
Create — Research a topic, produce a new entry from the dataset's template.
Update — Read an existing entry, research incomplete sections ([TODO] or bare -), fill them in.
Workflow
1. Determine dataset and mode
Read the manifest at configuration/datasets.manifest.json to list available datasets.
Ask the user which dataset if ambiguous. Infer mode from context (new topic = create, existing file = update).
2. Locate paths
- Template:
datasets/{dataset-id}/config/{dataset-id}-comparison-template.md - Data dir:
datasets/{dataset-id}/data/ - Entry file:
datasets/{dataset-id}/data/{entry-name}.md
Entry filenames: lowercase, hyphens for spaces, no special characters (e.g. claude-code.md).
3. Research
Use Perplexity MCP tools for web research:
mcp__perplexity__searchfor broad feature/capability searchesmcp__perplexity__get_documentationfor official docs and API details
Minimum 2-3 reputable sources (official docs, GitHub repos, technical blogs). For each claim, verify across sources before writing.
4. Author or update the entry
Create mode:
- Read the dataset template
- Verify the file does not already exist in the data directory
- Write a new markdown file following the template exactly
- Fill every section with researched content
Update mode:
- Read the existing entry file
- Identify sections with
[TODO], bare-, or empty content - Research those specific areas
- Update only incomplete sections; preserve all existing content
5. Confirm
Present a summary of changes to the user before writing.
Formatting rules
See references/formatting.md for the complete formatting specification covering heading structure, Yes/No fields, label fields, and list conventions.
Constraints
- NEVER modify heading structure (## or ###). Only fill content within sections.
- NEVER overwrite an existing file in create mode unless explicitly told to.
- Use
-for fields that remain unknown after research. - Do not add sections not present in the template.
- Strip HTML comments (
<!-- -->) from output — they are template guidance only.