decision

Use when the quest needs an explicit go, stop, branch, reuse-baseline, write, finalize, reset, or user-decision transition with reasons and evidence.

Quality

72%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./src/skills/decision/SKILL.md

Quality

Content

77%

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

The body is highly actionable with a clear sequenced workflow, explicit validation gates, and concrete tool calls, and it points to four well-signaled reference files; its weaknesses are redundancy (the decision record restated several times) and verbose inline paper/exploration judgment rules that the skill itself says belong in references.

Suggestions

De-duplicate the decision record: state verdict/action/reason/evidence paths/next-direction once and reference it from the workflow and Validation sections instead of repeating it.

Move the publishability stop-loss rule and the exploration-depth gate into a reference file (e.g. research-route-criteria.md) and keep only the trigger condition inline, honoring the stated 'control surface in front' principle.

Collapse the duplicated 'do not repeat the same decision without new evidence' guidance to a single location.

Dimension	Reasoning	Score
Conciseness	The body is domain-specific rather than padded with concepts Claude already knows, but it could be tightened: the decision record (verdict/action/reason/evidence paths/next direction) is restated in the control workflow, the 'Required decision record' section, and the Validation checklist, and 'do not repeat the same decision without new evidence' appears in both AVOID and Interaction discipline, placing it at 2 rather than the lean anchor of 3.	2 / 3
Actionability	For an instruction-only skill the guidance is concrete and executable: a numbered control workflow, a named canonical-actions list, explicit tool calls (e.g. `artifact.submit_paper_outline(mode='select', ...)`, `bash_exec(...)`, `artifact.interact(kind='milestone', reply_mode='threaded', ...)`), and a concrete validation checklist, matching the fully-actionable anchor.	3 / 3
Workflow Clarity	The 5-step control workflow is clearly sequenced with an explicit 'Validation' gate ('Before decision can end, all applicable checks should be true') and feedback loops that route back to `intake-audit` or `idea` when preconditions fail, matching the clear-sequence-with-validation anchor.	3 / 3
Progressive Disclosure	Four real one-level-deep reference files are clearly signaled ('read:' with 'Use them when:' guidance) and all exist, but long inline judgment material — the publishability stop-loss rule, the exploration-depth gate, and the paper routing rules — contradicts the skill's own stated principle of keeping the control surface in front and moving longer route-shaping notes to references, so content that should be separate is inline.	2 / 3
	Total	10 / 12 Passed

Description

67%

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 is specific and distinct, with an explicit 'Use when' trigger clause, but it foregrounds the trigger conditions (when) while leaving the skill's own action (what it does) implied, and its trigger terms are system jargon rather than natural user language.

Suggestions

Lead with an explicit action verb for the 'what' before the 'Use when' clause, e.g. 'Records a route judgment (go, stop, branch, ...) from durable evidence. Use when the quest needs...'.

Add natural-language trigger variations a user might actually say alongside the canonical action names (e.g. 'whether to continue, stop, pivot, or reuse a prior result').

Consider framing at least one trigger in plain terms ('when the next step is unclear or contested') to broaden natural keyword coverage.

Dimension	Reasoning	Score
Specificity	The description lists multiple concrete transition actions — 'go, stop, branch, reuse-baseline, write, finalize, reset, or user-decision transition' — matching the anchor for listing several specific concrete actions rather than vague language.	3 / 3
Completeness	The 'Use when the quest needs...' clause gives an explicit trigger (when), but the performing action (what the skill actually does — make/record the route judgment) is only implied by 'transition with reasons and evidence', not stated as an explicit verb, so it does not reach the both-what-and-when anchor of 3.	2 / 3
Trigger Term Quality	The terms are the system's canonical action vocabulary (e.g. 'reuse-baseline', 'finalize', 'user-decision transition'), which are relevant triggers but technical jargon a user would not naturally say; missing common natural-language variations, so it sits at 2 rather than 3.	2 / 3
Distinctiveness Conflict Risk	Tying triggers to specific quest route transitions ('reuse-baseline', 'finalize', 'reset') gives the skill a clear niche unlikely to fire for unrelated skills, matching the clear-niche anchor.	3 / 3
	Total	10 / 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: ResearAI/DeepScientist
Path: src/skills/decision/SKILL.md
Commit: b366244

Reviewed: 2 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.