grasp

Verify understanding after AI work through intent-scented entry points. Type: (ResultUngrasped, User, VERIFY, Result) → VerifiedUnderstanding. Alias: Katalepsis(κατάληψις).

Quality

17%

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Passed

No known issues

Optimize this skill with Tessl

npx tessl skill review --optimize ./katalepsis/skills/grasp/SKILL.md

Quality

Discovery

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

This description is nearly impenetrable, relying on academic jargon, Greek terms, and a type-signature notation that provides no practical guidance for skill selection. It fails to describe concrete actions, lacks natural trigger terms, and provides no explicit 'when to use' guidance. The description would be essentially unusable in a multi-skill selection context.

Suggestions

Replace abstract jargon with plain language describing concrete actions, e.g., 'Checks whether the user understands AI-generated results by asking clarifying questions and providing simplified explanations.'

Add an explicit 'Use when...' clause with natural trigger terms, e.g., 'Use when the user seems confused by AI output, asks what a result means, or requests verification of AI-generated work.'

Remove the type signature notation and Greek alias, which add no value for skill selection, and instead describe the input/output in plain terms.

Dimension	Reasoning	Score
Specificity	The description uses highly abstract, jargon-heavy language like 'intent-scented entry points', 'ResultUngrasped', and 'Katalepsis(κατάληψις)'. No concrete actions are listed — it does not describe what specific steps or outputs are produced.	1 / 3
Completeness	The 'what' is extremely vague ('verify understanding after AI work') and there is no explicit 'when' clause or trigger guidance. The type signature notation does not serve as a usable trigger for skill selection.	1 / 3
Trigger Term Quality	The keywords used ('Katalepsis', 'κατάληψις', 'ResultUngrasped', 'intent-scented entry points') are not terms any user would naturally say. There are no natural language trigger terms like 'explain', 'check understanding', or 'verify results'.	1 / 3
Distinctiveness Conflict Risk	The description is so niche and jargon-laden that it is unlikely to conflict with other skills, but this is due to obscurity rather than clear, well-defined boundaries. It could still overlap with any skill related to verification or comprehension checking.	2 / 3
	Total	5 / 12 Passed

Implementation

27%

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

This skill is an extraordinarily dense formal specification of a comprehension verification protocol. While it demonstrates deep theoretical rigor and covers edge cases exhaustively, it is far too verbose and abstract for practical use as a SKILL.md — the formal type theory notation, Greek-letter state variables, and academic framing dramatically inflate token cost without proportional actionability gains. The content would benefit enormously from being split into a concise overview SKILL.md with detailed specifications in separate reference files.

Suggestions

Reduce the SKILL.md to a concise overview (under 100 lines) covering the core flow: detect ungrasped result → present entry points → Socratic verification loop → confirm understanding. Move the formal type system, state machine, and detailed sub-step specifications into separate reference files.

Replace the abstract formal notation block with 1-2 concrete worked examples showing a complete Katalepsis interaction (e.g., user says 'what did you change?' after a code refactor, and the skill walks through actual entry points, actual probes, and actual evaluation).

Remove explanations of concepts Claude already knows (Socratic questioning, comprehension verification, what a misconception is) and trust Claude to apply these concepts from brief directives.

Create bundle files for the detailed taxonomies (Gap Taxonomy, Entry Point Taxonomy, Artifact Basis Taxonomy) and the formal type/state specifications, referencing them from the main SKILL.md with clear one-level-deep links.

Dimension	Reasoning	Score
Conciseness	Extremely verbose and dense — the formal specification block alone is hundreds of lines of type theory, state machines, and Greek-letter variables that Claude does not need spelled out in this detail. Concepts like 'Socratic questioning' and 'comprehension verification' are explained at enormous length when Claude already understands them. The token cost is very high relative to the actionable content.	1 / 3
Actionability	The skill does provide concrete phase-by-phase steps, TaskCreate/TaskUpdate call patterns, and structured question templates with option formats. However, much of the guidance is buried in abstract formal notation rather than executable examples, and the actual verification questions are template placeholders rather than concrete worked examples showing real probes for real AI outputs.	2 / 3
Workflow Clarity	The multi-phase workflow (Phase 0→1→2→3) is clearly sequenced with explicit transitions and loop conditions. However, the sheer complexity and density of sub-steps (3, 3b, 3c, 3d, 3e), branching conditions, and formal state management makes it very difficult to follow in practice. Validation checkpoints exist (coverage checks, convergence evidence) but are obscured by the formal machinery.	2 / 3
Progressive Disclosure	The entire specification is a monolithic wall of text with no bundle files or external references to break up the content. The formal specification block, type definitions, phase details, taxonomies, and rules are all inlined in a single massive document. There is no separation of overview from detailed reference material.	1 / 3
	Total	6 / 12 Passed

Validation

100%

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 — 11 / 11 Passed

Validation for skill structure

No warnings or errors.

Repository: jongwony/epistemic-protocols
Commit: a1acc87

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.