cursor-ask

Read-only delegate to cursor-agent (Composer) for questions, investigation, design discussion, and adversarial sanity checks. No worktree, no cherry-pick, no Lead diff review — cursor-agent is locked to ask mode and cannot write. Use when user says: ask cursor, cursor sanity check, get a second opinion, adversarial review, design discussion, investigate with cursor, cursor:ask. Do NOT load for: implementation, refactor, file edits, commit/push work, anything requiring write access (use cursor:do or breezing --cursor instead).

Quality

85%

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

Quality

Content

75%Weight 40%Scale 1-5

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 fully executable delegation flow and concrete trust-boundary configuration, and it is well-structured with clear references. The main weakness is repetition of the banner/narration rules across multiple sections, which inflates token cost without adding information.

Suggestions

Collapse the duplicated banner/narration rules: state them once in the Narration Rules section and have Step 0 and Step 1 reference it instead of re-stating the 3-line rule and banner line.

Remove re-explanation of the read-only/no-worktree/no-cherry-pick point in the body since it is already established in the frontmatter description.

Add an explicit post-Step-3 checkpoint (e.g., confirm the 3-5 line summary preserves a host judgment axis) to lift workflow clarity toward 5.

Dimension	Reasoning	Score
Conciseness	The body is dense and operational with no padding of concepts Claude already knows, but the banner/narration rules are restated across the Narration Rules section, Step 0, and Step 1 (the banner line appears three times) and the read-only/no-worktree point repeats the description, which could be tightened.	3 / 5
Actionability	Step 2 provides a fully executable copy-paste bash block (plugin-root resolution plus cursor-companion invocation with error exits), and the Trust Boundary table gives exact config locations and allowlist entries covering the common cases.	5 / 5
Workflow Clarity	A clearly sequenced Step 0-3 flow with explicit validation (valid_root predicate, exit 1/2 on failure) and a justified table explaining why destructive-op safeguards are unneeded in read-only ask mode; held below 5 because Steps 0 and 1 overlap rather than adding new checkpoints and Step 3 lacks an explicit verify-summary checkpoint.	4 / 5
Progressive Disclosure	Well-organized with clear section headers and one-level-deep, clearly signaled references (Related Skills, .claude/rules/cursor-cli-only.md); no bundle files exist, and some inlined content like the long root-resolution script could plausibly live in a referenced file, but its inline placement is justified.	4 / 5
	Total	16 / 20 Passed

Description

95%Weight 40%Scale 1-5

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 strong: it clearly states the read-only delegation purpose, supplies thorough natural trigger phrases, and explicitly disambiguates from sibling write-capable skills. Minor room for improvement only in making the listed actions more concrete.

Dimension	Reasoning	Score
Specificity	Names the read-only delegation domain and several concrete actions ("questions, investigation, design discussion, and adversarial sanity checks") plus explicit write/worktree/cherry-pick exclusions, but the actions are somewhat abstract categories rather than a comprehensive enumerated set.	4 / 5
Completeness	Explicitly answers both what (read-only delegate locked to ask mode, cannot write) and when (concrete "Use when user says:" trigger phrases), with additional negative boundary guidance.	5 / 5
Trigger Term Quality	Comprehensive natural trigger phrases including synonyms and a slash-command form: "ask cursor, cursor sanity check, get a second opinion, adversarial review, design discussion, investigate with cursor, cursor:ask".	5 / 5
Distinctiveness Conflict Risk	Clear niche with distinct triggers and explicit disambiguation against sibling skills ("Do NOT load for... use cursor:do or breezing --cursor instead"), minimizing conflict risk.	5 / 5
	Total	19 / 20 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
referenced_paths_exist	Referenced path issues: 3 missing	Warning

	Total	15 / 16 Passed

Repository: Chachamaru127/claude-code-harness
Path: opencode/skills/cursor-ask/SKILL.md
Commit: 7d8deba

Reviewed: about 15 hours 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.