CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

llm-runtime-architecture

Use when designing how one canonical agent core runs across Codex, Claude Code, Gemini CLI, Cursor, and AGENTS.md-compatible tools.

Quality

80%

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 ./.agents/skills/llm-runtime-architecture/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

86%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 a lean, well-structured instruction skill: an ordered procedure with a concrete output schema, an exact file path, and explicit verification requirements, with no wasted tokens. Its only gap is the absence of a small concrete example (one runtime row or a sample verification command) that would make the output format fully unambiguous.

DimensionReasoningScore

Conciseness

The body is 20 lines of pure instruction — "Keep `AGENTS.md` as the canonical behavior contract", "Keep adapters thin and point them back to the canonical core" — with zero padding and no explanation of concepts Claude already knows. Every token earns its place, matching the 'lean and efficient' anchor; there is nothing to trim and no over-explanation that would push it to 4.

5 / 5

Actionability

Guidance is specific and executable for an instruction-only skill: an exact file path (".agentlas/global-commands.json"), an enumerated per-runtime field list ("entry point, global command, adapter files, available tools, memory access, limitations, and verification command"), and a defined output schema. It falls short of 5 because no example runtime entry, verification command, or sample matrix row is given, leaving minor gaps in what a complete deliverable looks like; it is above 3 because the guidance is concrete, not pseudocode or high-level hints.

4 / 5

Workflow Clarity

The five-step procedure is a clear, ordered sequence, and verification is woven in (step 2 requires naming a "verification command" per runtime and the output includes a `verification` field). This matches 'clear sequence with most checkpoints present; minor validation gaps' — there is no explicit validate-then-proceed checkpoint or feedback loop, keeping it below 5, but the task is a design/documentation workflow (not destructive or batch), so the workflow_clarity-at-3 cap does not apply and verification is more than absent.

4 / 5

Progressive Disclosure

The skill is under 50 lines with two well-organized sections ("## Procedure", "## Output") and no content that belongs in external files; no references/ scripts/ or assets/ bundle exists and the body references none. Per the scoring notes, a sub-50-line skill with no need for external references scores 5 with well-organized sections, which this has.

5 / 5

Total

18

/

20

Passed

Description

65%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, distinctive niche with strong named-runtime trigger terms and an explicit 'Use when' clause, but it buries the 'what' inside the trigger and never states concrete actions or deliverables. It would be substantially stronger by leading with what the skill produces (e.g., a runtime matrix, adapter files, global-commands.json) before the trigger clause.

Suggestions

State the concrete 'what' explicitly, e.g. "Generates a runtime matrix (entry point, global command, adapter files, memory access, limitations, verification) for running one canonical agent core across Codex, Claude Code, Gemini CLI, Cursor, and AGENTS.md-compatible tools" — this also lifts specificity.

Mention concrete artifacts the skill writes, such as ".agentlas/global-commands.json" or thin per-runtime adapters, so the capability list is actionable rather than a single generic verb.

Add natural trigger synonyms such as "cross-platform agent", "portable agent configuration", or "runtime matrix" to improve trigger term coverage.

DimensionReasoningScore

Specificity

The description names the domain concretely ("canonical agent core runs across Codex, Claude Code, Gemini CLI, Cursor") but the only action is the generic verb "designing" — no concrete deliverables like "generates a runtime matrix" or "writes adapter files" are stated. This sits at the 'names domain and 1-2 concrete actions, but not comprehensive' level; it is above level 2 because the domain and tool list are specific, but below level 4 because it lists no concrete actions.

3 / 5

Completeness

The 'when' is explicit ("Use when designing how one canonical agent core runs across...") but the 'what' only exists implicitly inside that trigger clause — the description never separately states what the skill produces. This is above level 2 ("only 'when' is present without 'what'") because the when-clause conveys a fairly specific what, but below level 4 because the 'what' is never made explicit and the trigger could be more specific (e.g., packaging, writing global commands).

3 / 5

Trigger Term Quality

"Codex, Claude Code, Gemini CLI, Cursor, and AGENTS.md-compatible tools" are natural terms a user would say when needing this skill, and "agent core" / "designing" fit the intent. It falls short of level 5's comprehensive synonym coverage — natural variations like "cross-platform agent", "portable agent config", or "runtime matrix" are missing — but is above level 3 because the named runtimes are exactly the phrases users would use.

4 / 5

Distinctiveness Conflict Risk

"one canonical agent core runs across Codex, Claude Code, Gemini CLI, Cursor, and AGENTS.md-compatible tools" carves out a clear niche with distinct named-runtime triggers, making accidental triggering for unrelated skills very unlikely. It matches the 'clear niche with distinct triggers; minimal conflict risk' anchor and does not exhibit the overlap breadth of level 4.

5 / 5

Total

15

/

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

agentlas-ai/Agentlas-OS

Path

.agents/skills/llm-runtime-architecture/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.