CtrlK
BlogDocsLog inGet started
Tessl Logo

ultracave

Caveman at maximum compression: fragments, one word when one word is enough, each fact once. Invoke only with /ultracave or /caveman ultra. Stays on until "stop caveman" or "normal mode".

76

Quality

95%

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

SKILL.md
Quality
Evals
Security

Quality

Content

100%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.

Exemplary instruction-only skill content: terse yet complete, with concrete examples, an explicit safety floor, enumerated exception conditions, and a pre-send validation checklist. The structure is clean and appropriately sized for a single-purpose behavioral mode with no bundle files needed.

DimensionReasoningScore

Conciseness

The body is the embodiment of its own advice — every line is a rule, an example, or an edge case, with zero padding and no explanation of concepts Claude already knows. Even the apparent redundancy of 'Ultracave is caveman with the grammar stripped' earns its place by distinguishing this mode from plain caveman mode, matching the 'every token earns its place' anchor.

5 / 5

Actionability

Concrete Bad/Good example pairs (e.g., 'Bad: The component re-renders because… / Good: Inline object prop, new ref, re-render.'), an explicit never-cut floor listing exact categories ('code, commands, paths, API names, error strings (verbatim). not/never/no/only/except. Numbers and units.'), and a specified /caveman status behavior give fully executable guidance. As an instruction-only skill, the absence of code is not penalized — the guidance is specific and directly actionable.

5 / 5

Workflow Clarity

This is a simple single-purpose skill and the single action is unambiguous: compress every response, with clear persistence rules, an explicit never-cut floor, enumerated break conditions ('security warning. Irreversible action, confirm first. Any fragment with two readings.'), and a pre-send validation check. The pre-send checklist serves as the validation checkpoint, satisfying the simple-skill exception fully.

5 / 5

Progressive Disclosure

At roughly 54 lines with no external references needed, the body is organized into clear, well-ordered sections (Persistence, Floor, Rules, When to break the rules, Pre-send check) with no content that belongs in a separate file. Per the rubric's simple-skill guidance, this earns full credit on structure alone.

5 / 5

Total

20

/

20

Passed

Description

87%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 strong description: it states what the skill does in concrete terms, gives explicit invocation-only triggers with exact commands, and defines persistence boundaries. Its only weakness is slight terseness in capability coverage — it omits preserved-content details and natural-language synonyms like 'terse' that users might use.

DimensionReasoningScore

Specificity

The description lists several concrete style behaviors — 'fragments', 'one word when one word is enough', 'each fact once' — which goes beyond the 1-2 actions of anchor 3 but lacks the comprehensive action coverage of anchor 5. Minor gaps remain (e.g., it does not mention what content is preserved, such as code or error strings).

4 / 5

Completeness

It clearly answers both questions: the 'what' is the compressed fragment style ('Caveman at maximum compression: fragments, one word when one word is enough, each fact once') and the 'when' is explicit with concrete trigger phrases ('Invoke only with /ultracave or /caveman ultra'), including session persistence scope ('Stays on until "stop caveman" or "normal mode"').

5 / 5

Trigger Term Quality

It names the exact invocation commands users would naturally type ('/ultracave', '/caveman ultra', 'stop caveman', 'normal mode'), giving good keyword coverage. A few natural phrasings a user might say are missing, such as 'terse', 'shorter responses', or 'compress your answers', so it falls just short of anchor 5's comprehensive synonym coverage.

4 / 5

Distinctiveness Conflict Risk

'Invoke only with' restricts triggering to two distinct slash commands, carving out a clear niche with minimal overlap risk against generic writing-style or formatting skills. The explicit exit commands further reduce accidental triggering.

5 / 5

Total

18

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository
JuliusBrussee/caveman
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.