Tessl • The management layer for AI agent skills

An exceptionally actionable, well-sequenced porting workflow with strong validation checkpoints and repo-specific detail that assumes Claude's competence. The main structural weakness is that SKILL.md is a ~200-line monolith: the integration checklist, package layout, and command-class spec would work better as one-level-deep reference files.

Suggestions

Move the "Integration surfaces checklist" tables and the "Command class" spec into separate reference files (e.g., references/integration-checklist.md, references/command-class.md) linked from a brief overview section in SKILL.md.

Consider extracting the package layout tree into its own reference file so SKILL.md stays a navigable overview.

Tighten the longest bullets (e.g., the --low-ram img2img row and cross-model touch points) by moving rationale into a reference and keeping the rule statement inline.

Dimension	Reasoning	Score
Conciseness	The body is dense with repo-specific knowledge Claude would not know (ModelConfig wiring, WeightDefinition conventions, PR references) and assumes expertise throughout, with almost no padding or explanation of known concepts. It falls short of the lean 5 anchor due to occasional wordiness, e.g., the long --low-ram/img2img bullet and some multi-clause checklist rows.	4 / 5
Actionability	Guidance is fully concrete and executable: exact commands ("MFLUX_PRESERVE_TEST_OUTPUT=1 uv run <test command>", "just lint", "just test-fast", "just ci-extract"), full file paths, a package-layout tree, and an explicit validate/load/generate command-class spec with a named copy source (src/mflux/models/qwen/cli/qwen_image_generate.py). Specific examples cover the common port cases.	5 / 5
Workflow Clarity	A 7-step sequenced workflow (scope → port → port order → deterministic validation → refactor → pre-merge polish → finalize) with explicit validation checkpoints: deterministic output tests that "lock" correctness, visual inspection gates, re-running tests during refactors to avoid regressions, milestone commits, and pre-merge re-runs of lint/test-fast/slow golden tests. Error-recovery feedback loops are present (e.g., OOM ⇒ re-check shapes, diffusers sanity check via mflux-debugging when outputs disagree).	5 / 5
Progressive Disclosure	The body is well-sectioned with headers and tables, but no bundle files exist and everything is inlined in SKILL.md: the integration-surfaces checklists, package layout spec, and command-class pattern are substantial reference material that clearly belongs in separate files. This matches the 3 anchor (content that should be separate is inline) rather than 2, since structure and navigation within the file are good.	3 / 5
	Total	17 / 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 concise, specific description that clearly states what the skill does in third person with a well-defined niche. Its main weakness is the absence of any "Use when..." trigger clause, which both caps completeness and leaves trigger guidance implicit.

Suggestions

Add an explicit trigger clause, e.g., "Use when porting a diffusers/HF model to mflux or MLX, or when adding a new model family to the mflux repo."

Include natural synonyms users might say, such as "diffusers", "Apple Silicon", or "translate model to MLX", to broaden trigger term coverage.

Optionally enumerate one or two more concrete capabilities (e.g., deterministic output tests, weight mapping) to lift specificity toward comprehensive coverage.

Dimension	Reasoning	Score
Specificity	"Port ML models into mflux/MLX", "correctness-first validation", and "refactor toward mflux style" name the domain plus three concrete actions. It sits above the 3 anchor (only 1-2 actions) but below 5 because coverage omits porting specifics like weight mapping, CLI wiring, and testing.	4 / 5
Completeness	The "what" is clear (port ML models into mflux/MLX, validate, refactor), but there is no "Use when..." clause or equivalent explicit trigger guidance, capping completeness at 3 per the rubric guideline. It is not a 2 because the "what" is concrete and specific rather than vague.	3 / 5
Trigger Term Quality	"Port", "ML models", "mflux", and "MLX" are the natural terms a user would say for this need, giving good keyword coverage. A few natural synonyms are missing (e.g., "diffusers", "Apple Silicon"), keeping it below the comprehensive 5 anchor.	4 / 5
Distinctiveness Conflict Risk	"mflux/MLX" pins a clear niche with distinct triggers; no other generic skill (model training, general ML) would plausibly claim this description. Minimal conflict risk.	5 / 5
	Total	16 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
referenced_paths_exist	Referenced path issues: 1 missing	Warning

	Total	15 / 16 Passed

Repository: mflux-community/mflux
Path: .cursor/skills/mflux-model-porting/SKILL.md
Revision: a3b21fc

Reviewed: 4 days ago

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.

mflux-model-porting

Table of Contents

Is this your skill?