research-deep

Read research outline, launch independent agent for each item for deep research. Disable task output.

Quality

50%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./skills/research-en/research-deep/SKILL.md

Quality

Content

57%Weight 40%Scale 1-3

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The content is well-organized and mostly token-efficient with a concrete prompt template and validation step, but it relies on external files that are not bundled, duplicates the template in the one-shot example, and lacks an explicit validate→fix→retry feedback loop for its batch operation. Tightening the duplication and adding the retry loop would lift the weaker dimensions.

Suggestions

Collapse or shorten the One-shot Example so it does not duplicate the Prompt Template; reference it as 'fill in {item_related_info}, {output_path}, {fields_path}' instead of repeating all requirements.

Add an explicit validation feedback loop in Step 4/5: 'If validation fails, the agent must fix missing fields and re-run validate_json.py until it passes before the task is considered complete.'

Either bundle validate_json.py in ./scripts/ or replace the hardcoded '~/.claude/skills/research/validate_json.py' path with a script-local reference so the command is executable as written.

Dimension	Reasoning	Score
Conciseness	The body is largely efficient and well-structured with a tight prompt template, but the One-shot Example duplicates the Prompt Template almost verbatim and some Parameter Retrieval lines restate what the template already shows, so it could be tightened.	2 / 3
Actionability	Provides a concrete, copy-paste prompt template and a validation command, but references external files that are not bundled here (validate_json.py, fields.yaml, outline.yaml) and leaves steps like 'Find */outline.yaml' and slugify rules as prose rather than executable commands.	2 / 3
Workflow Clarity	Steps are clearly numbered (1–5) with batching and resume logic, and a validation step exists, but there is no validate→fix→retry feedback loop and the batch-approval checkpoint is only implied, so checkpoints are incomplete for a batch operation.	2 / 3
Progressive Disclosure	The body is organized into clear sections (Trigger, Workflow, Agent Config) with a highlighted prompt template, but there are no bundle files in references/scripts/assets, so referenced external artifacts (validate_json.py, fields.yaml) are not actually present to verify navigation depth.	2 / 3
	Total	8 / 12 Passed

Description

42%Weight 40%Scale 1-3

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description states concrete, specific actions but omits any explicit 'Use when...' trigger guidance, capping completeness at 2. It also relies on the skill name and internal jargon rather than natural user phrases, weakening trigger-term quality. Adding a 'Use when...' clause with natural terms would substantially improve it.

Suggestions

Append an explicit trigger clause such as 'Use when the user wants to run a deep, multi-item research pass over a pre-defined outline.yaml, or asks to research many items in parallel.'

Replace internal jargon ('items_per_agent', 'disable task output') in the description with natural user-facing phrasing so the what-statement stays concrete but readable.

Add common natural terms (e.g. 'deep research', 'research multiple topics', 'batch research') to improve trigger coverage and reduce overlap with generic research skills.

Dimension	Reasoning	Score
Specificity	Names concrete actions in the research domain: 'launch independent agent for each item', 'read items list', 'disable task output' — several specific actions tied to a concrete workflow.	3 / 3
Completeness	Answers 'what' (read outline, launch agents per item, disable output) but provides no 'Use when...' clause or equivalent trigger, so when-to-use is entirely missing.	2 / 3
Trigger Term Quality	Contains no natural trigger keywords a user would say; relies on the skill name 'research-deep' and jargon like 'outline' and 'items_per_agent' rather than phrases users actually utter.	1 / 3
Distinctiveness Conflict Risk	The deep-research niche is somewhat specific, but without explicit triggers and given the generic word 'research' it could overlap with other research-oriented skills.	2 / 3
	Total	8 / 12 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning

	Total	15 / 16 Passed

Repository: Weizhena/Deep-Research-skills
Path: skills/research-en/research-deep/SKILL.md
Commit: e5479f8

Reviewed: 11 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.