CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

preview

Use when the user is about to commit to one of several directions they cannot judge from descriptions — a choice deferred to a principle, the options reworked instead of chosen, or 'I'd have to see it'. The user can then say what each means for their purpose, on what ground, and which one they take.

Quality

61%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./proplasma/skills/preview/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

35%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.

An ambitious formal protocol whose body buries a workable workflow (relay spec → build placeholder probes → present contrast map → person closes → harvest → verified discard) under extreme abstraction and fourfold redundancy, with a ~450-line Lean contract inlined in the SKILL.md. The one reference file is well used, but the skill offers almost no concrete, executable material — no example probe, map, or round — leaving the reader to reverse-engineer the practice from the specification.

Suggestions

Add one worked example — a sample vignette probe's narration text, a small rendered contrast map with per-cell footing, and one composed round — so the protocol becomes executable guidance instead of a specification to interpret (addresses actionability).

State the workflow exactly once as a short numbered sequence in plain language, and delete the repeated restatements across FLOW, MORPHISM, PHASE TRANSITIONS, LOOP, and the specRelay/present entries; trim the `.present` entry to its decision rules (addresses conciseness).

Move the full Lean contract to a bundle file such as `references/contract.md`, loaded on demand the way `round-composition.md` already is, and keep SKILL.md as a lean overview with the core principle and step sequence (addresses progressive_disclosure and conciseness).

DimensionReasoningScore

Conciseness

The ~500-line body restates the same workflow at least four times (FLOW, MORPHISM, PHASE TRANSITIONS/LOOP, and TOOL GROUNDING's specRelay/present entries), and the `.present` entry alone is a several-hundred-word wall of nested clauses. It does not pad with concepts Claude already knows, which keeps it above anchor 1, but the redundancy and density are far past 'could be tightened' (anchor 3) — 'noticeably verbose; several unnecessary... padded sections' fits.

2 / 5

Actionability

For an instruction-only skill the guidance is overwhelmingly abstract: 'Expose direction unknowns through divergent-discard instantiation before commitment' and a Lean specification whose 'axioms' are declared model judgments. There are no concrete examples of a probe narration, a rendered contrast map, or a composed round, so Claude receives high-level structure without the specific steps or material to execute — matching 'minimal concrete guidance; high-level hints but missing the specific steps'. It is above anchor 1 only because the operation list and ordering constraints give some real structure.

2 / 5

Workflow Clarity

A sequence does exist (gather → relay_spec → instantiate → present → fuse → constitute → harvest → cleanup → assemble) with real checkpoints — cleanupVerify verifies absence, the harvest must be read before discard, and unclear closings are handled explicitly. But the sequence is scattered across four restatements in heavily obfuscated language, and there is no concrete procedure for what a practitioner actually does in a round, so 'steps listed... checkpoints' is present while clarity is compromised between anchors 3 and 4.

3 / 5

Progressive Disclosure

The single reference, `references/round-composition.md` (a real file), is genuinely well-signaled with explicit occasion-bound conditions for when to read it, and is one level deep. However, the body is dominated by a ~450-line inline Lean block that is the kind of contract detail a SKILL.md overview should point to rather than inline, so the structure is only 'some' — above the minimal-structure anchor (2) thanks to the reference, below 'most content appropriately placed' (4).

3 / 5

Total

10

/

20

Passed

Description

70%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.

A well-constructed trigger description: explicit 'Use when' clause, natural trigger phrasing ('I'd have to see it'), and a distinct niche. Its main weakness is that the capability side is stated as user outcomes rather than the skill's concrete actions, leaving the reader to infer what the skill actually does. Third-person voice is used consistently.

DimensionReasoningScore

Specificity

The description names the domain (pre-commit direction choice) and a couple of outcome-level actions ("say what each means for their purpose, on what ground, and which one they take"), but these are the user's actions rather than concrete actions the skill performs; nothing like 'builds placeholder probes' or 'renders a contrast map' is stated. It sits at 'names domain and 1-2 concrete actions' — above the generic-minimal anchor (2) but below 'lists several specific actions' (4).

3 / 5

Completeness

Both parts are present: an explicit 'Use when the user is about to commit to one of several directions they cannot judge from descriptions' with concrete trigger cases, and a 'what' clause ('The user can then say what each means... and which one they take'). The 'what' is framed as user outcome rather than what the skill actually does (materialize placeholder probes, show a contrast map), which keeps it below the fully explicit anchor (5) but above anchor 3, where 'when' would be weak or missing.

4 / 5

Trigger Term Quality

Includes a genuinely natural user phrase — "'I'd have to see it'" — plus recognizable situational triggers like "a choice deferred to a principle" and "the options reworked instead of chosen". Good coverage, but common variations (e.g., 'help me decide', 'compare options', 'which approach') are missing, so it falls short of the comprehensive-synonym anchor (5) while clearly above 'some relevant keywords' (3).

4 / 5

Distinctiveness Conflict Risk

The trigger conditions are a clear niche — direction choices that 'cannot be judged from descriptions', with three distinctive situational markers — so conflict risk with other skills is low. Minor overlap remains with generic decision-help or comparison skills, which keeps it below the minimal-conflict anchor (5).

4 / 5

Total

15

/

20

Passed

Validation

93%

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

CriteriaDescriptionResult

skill_md_line_count

SKILL.md is long (556 lines); consider splitting into references/ and linking

Warning

Total

15

/

16

Passed

Repository

jongwony/epistemic-protocols

Path

proplasma/skills/preview/SKILL.md

Reviewed

Table of Contents

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.