CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

ai-engineering

Reviews and guides LLM/AI application engineering — prompts, caching, RAG, agent loops and tools, resilience, model migration, evals, prompt-injection defence, and observability. Invoke with /ai-engineering for "review my prompt", "set up RAG", or "write evals".

Quality

85%

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

77%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 well-engineered thin index: lean, highly actionable process guidance with excellent mode routing and workflow checkpoints. Its critical weakness is that the progressive-disclosure design is not backed by the bundle — all thirteen rule files and all four templates referenced by the routing tables are missing, so the skill cannot actually load the content it defers to.

Suggestions

Ship the referenced bundle files: all 13 rules/*.md and all 4 templates/*.md paths cited in the Area Routing and Required Reading tables are absent from the bundle, so every routing decision leads to a file that cannot be loaded. Either include them or inline the essential rules per area.

Remove the 'Required Reading by Area' table (or collapse it to just the templates and references rows): it re-lists the identical thirteen area-to-file mappings already given in 'Area Routing', spending ~20 lines on pure duplication.

Add one compact inline exemplar per mode (e.g., a single sample review finding with What/Rule/Fix, or a filled-in system-prompt skeleton fragment) so the skill stays actionable even when rule files have not yet been loaded.

DimensionReasoningScore

Conciseness

The body is a lean index that assumes competence — no explanations of what RAG or prompt caching are, one-liner anti-patterns, and date-sensitive material properly delegated to references/recent-changes.md. It is not a 5 because the 'Area Routing' table and the 'Required Reading by Area' table duplicate the same thirteen file mappings (~20 redundant lines).

4 / 5

Actionability

Concrete, executable process guidance throughout: explicit parsing rules ("$0 == \"review\""), a stated-mode output template, a findings format (What/Rule/Fix with path:line and ≤10-line snippets), a one-message question list for design mode, and a template mapping. It falls short of fully copy-paste-ready because all substantive per-area rules are deferred to rule files rather than exemplified inline.

4 / 5

Workflow Clarity

Three modes each get unambiguous detection rules, an explicit checkpoint ('State the detected mode... before continuing'), numbered steps, a consent gate ('Do not edit the file in review mode unless the user asks for fixes'), prioritized findings ('Top 3 fixes'), and a closing Definition of Done checklist — clear sequence with explicit validation, fitting the top anchor. No destructive or batch operations are performed, so the workflow-clarity cap does not apply.

5 / 5

Progressive Disclosure

The written design is exemplary — a self-described 'thin index' with clearly signaled, one-level-deep references and per-area 'load when' guidance — but scored against the actual bundle, 17 of the 19 referenced paths are dangling: none of the rules/*.md or templates/*.md files exist; only references/primary-sources.md and references/recent-changes.md are present. Navigation to the core detailed content is therefore broken, which is more than the 'minor organization gaps' of the 4-anchor, though SKILL.md itself is a proper overview rather than a monolith, keeping it above the 2-anchor.

3 / 5

Total

16

/

20

Passed

Description

91%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 names a coherent domain, enumerates concrete coverage areas, and includes explicit invocation guidance with literal example user phrases. Its only weaknesses are the limited verb variety (review/guide) and an umbrella scope that slightly overlaps narrower AI-dev skills.

DimensionReasoningScore

Specificity

The only differentiated verbs are "Reviews and guides", but the description enumerates nine concrete sub-domains ("prompts, caching, RAG, agent loops and tools, resilience, model migration, evals, prompt-injection defence, and observability"), matching the 'several specific actions; minor gaps' anchor rather than the 5-anchor, whose examples show multiple distinct concrete actions.

4 / 5

Completeness

It explicitly answers both questions: the 'what' ("Reviews and guides LLM/AI application engineering" across nine named areas) and the 'when' with concrete trigger phrases ("Invoke with /ai-engineering for 'review my prompt', 'set up RAG', or 'write evals'"), matching the top anchor. Not 4, because the when-clause is explicit with concrete phrases rather than merely present-but-vague; the missing-'Use when' cap does not apply since equivalent explicit trigger guidance exists.

5 / 5

Trigger Term Quality

It quotes literal user phrasings — "review my prompt", "set up RAG", or "write evals" — which is exactly what a user would naturally say, alongside natural area keywords (RAG, evals, caching, agent loops, prompt injection). This is comprehensive natural-term coverage; the 4-anchor ('a few natural terms missing') would understate the quoted example invocations.

5 / 5

Distinctiveness Conflict Risk

The advisory framing (review/design of one's own LLM application across its lifecycle) plus distinct slash-command triggers give it a mostly distinct niche, but the 13-area umbrella scope ('prompts... observability') creates minor overlap risk with specialized prompt-engineering, RAG, or API-reference skills — fitting the 4-anchor rather than the clear-niche 5-anchor.

4 / 5

Total

18

/

20

Passed

Validation

81%

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

Validation — 13 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_field

'metadata' should map string keys to string values

Warning

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

relative_links

Relative link issues: 34 missing

Warning

Total

13

/

16

Passed

Repository

mthines/agent-skills

Path

skills/quality/ai-engineering/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.