deep-interview

Socratic deep interview with mathematical ambiguity gating before explicit execution approval

Quality

60%

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 ./skills/deep-interview/SKILL.md

Quality

Content

66%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 exceptionally actionable and its multi-phase workflow is clearly sequenced with strong validation gates and feedback loops. However, it is severely bloated by internal repetition of the same rules (threshold resolution, prompt-budget discipline, the approval pipeline) and keeps everything in one monolithic file with a reference to a non-existent `docs/company-context-interface.md`. Splitting reference material into bundle files and deduplicating repeated policy statements would recover significant token budget at no loss of clarity.

Suggestions

State each policy exactly once and reference it (e.g., keep Phase 0 as the sole definition of threshold resolution; replace the repeats in step 3.5, the announcement, Tool_Usage, and the checklist with pointers) — this addresses the redundancy that drives conciseness to 2.

Split bulk material into bundle files: move the spec markdown template, the scoring prompt, the Advanced configuration/integration sections, and the challenge-agent tables into `references/` files linked from the body, instead of inlining everything in SKILL.md.

Fix or remove the dangling reference to `docs/company-context-interface.md` (line 405) — the file is not bundled, so the pointer currently dead-ends.

Dimension	Reasoning	Score
Conciseness	The 815-line body restates the same constraints many times: the Phase 0 threshold-resolution rule appears in Phase 0, step 3.5, the announcement, Tool_Usage, the spec template, and the Final Checklist (~6 times); the oversized-context/prompt-budget rule is repeated in Execution_Policy, step 3.6, Step 2a, Phase 4, Phase 5, and the checklist; the 3-stage approval pipeline is spelled out three times (Phase 5 diagram, the Advanced pipeline block, and the Approval-Gated Pipeline section); and the Advanced section duplicates Phase 3's challenge-agent table. This matches anchor 2 ('several unnecessary... padded sections') — the duplication is not explanation of concepts Claude already knows, so it is not a 1, but the pervasive redundancy is well beyond 'could be tightened'.	2 / 5
Actionability	The guidance is fully concrete and executable: exact state JSON schemas, a copy-paste scoring prompt with model/temperature specified, exact ambiguity formulas with dimension weights, an exact spec-file markdown template, exact file paths (`.omc/specs/deep-interview-{slug}.md`, `.omc/state/`), exact required first-line output text, and Good/Bad examples covering common cases. This matches anchor 5 (fully executable, copy-paste ready, specific examples covering common cases); it is not a 4 because no key execution details are missing.	5 / 5
Workflow Clarity	Phases 0-5 are explicitly sequenced with blocking prerequisites (Phase 0 before Phase 1, Round 0 before scoring, step 3.5 re-verification), a validation gate every round (ambiguity score vs. resolved threshold, weakest-dimension targeting rotation), feedback loops for error recovery (Phase 0 re-entry on missing threshold, escalation conditions, ontology-stall reframe, hard cap at 20 rounds), and a comprehensive final checklist. This matches anchor 5: clear sequence, explicit validation, error-recovery loops, and a checklist.	5 / 5
Progressive Disclosure	There are no bundle files (references/, scripts/, assets/ do not exist), so the entire 815-line body is a single monolithic file in which content that clearly belongs in separate files is inlined — the spec template, the scoring prompt, the Advanced configuration/integration material, and the challenge-agent tables. It also carries a dangling reference to `docs/company-context-interface.md` (line 405), which is not bundled. This fits anchor 2 ('content that clearly belongs in separate files is inlined'); internal headers give it navigability, so it is above 1, but the monolithic size plus the broken reference keep it below 3.	2 / 5
	Total	14 / 20 Passed

Description

53%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 has a clear, mostly specific 'what' but completely lacks a 'when' clause, which caps its usefulness for skill triggering. Trigger keywords exist ('Socratic', 'deep interview') but common user phrasings around vague ideas and requirements gathering are missing. Appending a 'Use when the user has a vague idea or says "interview me", "ask me everything", or wants validated requirements before execution' clause would lift the two lowest dimensions.

Suggestions

Add an explicit 'when' clause to the description, e.g. 'Use when the user has a vague idea or says "interview me", "don't assume", or wants mathematically-validated requirements before execution' — this addresses the missing trigger guidance that caps completeness at 3.

Include natural trigger synonyms users would actually say ('vague idea', 'requirements gathering', 'clarify what I want', 'not sure what I want') to broaden trigger_term_quality beyond 'Socratic'/'deep interview'.

Enumerate 2-3 concrete capabilities in the description itself (asks one targeted question per round, scores clarity across weighted dimensions, writes a spec and gates execution on explicit approval) to raise specificity from 3 to 4.

Dimension	Reasoning	Score
Specificity	The description names the domain and 1-2 concrete mechanisms ("Socratic deep interview", "mathematical ambiguity gating", "explicit execution approval"), but the actions are stated abstractly rather than as a comprehensive list of what the skill actually does (asking one question at a time, scoring clarity dimensions, crystallizing a spec, bridging to execution). It sits between anchor 3 (domain + 1-2 concrete actions, not comprehensive) and anchor 4; it does not reach 4 because 'gating' and 'approval' are process properties, not enumerated capabilities.	3 / 5
Completeness	The 'what' is present and reasonably clear (Socratic interview with ambiguity gating and approval), but there is no 'when' / 'Use when' clause or equivalent trigger guidance anywhere in the description — the body contains a detailed <Use_When> section that the description fails to surface. Per the judging guideline, a missing 'Use when' clause caps completeness at 3; it cannot score 4 because 'when' is entirely absent, not merely implicit.	3 / 5
Trigger Term Quality	'deep interview' and 'Socratic' are natural phrases a user would say (matching the body's own trigger list: "deep interview", "interview me", "socratic"), but the description omits common variations and synonyms users actually use, such as 'requirements gathering', 'clarify my idea', 'vague idea', 'ask me everything'. This matches anchor 3: some relevant keywords, missing common variations.	3 / 5
Distinctiveness Conflict Risk	'Socratic deep interview' carves a fairly distinct niche (requirements clarification) that is unlikely to trigger for document-processing or coding-execution skills, though 'interview' and the absence of trigger guidance leave minor overlap risk with planning/brainstorming skills like omc-plan. This fits anchor 4 (mostly distinct, minor overlap risk) better than 5, which requires explicit distinct trigger phrases.	4 / 5
	Total	13 / 20 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (815 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: Yeachan-Heo/oh-my-claudecode
Path: skills/deep-interview/SKILL.md
Commit: dc7ba1d

Reviewed: 3 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.