mlflow

Track ML experiments, manage model registry with versioning, deploy models to production, and reproduce experiments with MLflow - framework-agnostic ML lifecycle platform

Quality

66%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

High

Do not use without reviewing

Fix and improve this skill with Tessl

tessl review fix ./backend/cli/skills/ml-training/mlflow/SKILL.md

Quality

Content

57%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 rich with executable, framework-spanning code that makes it highly actionable, but it is verbose and repetitive, lacks validation/feedback workflows for its destructive registry and deployment operations, and buries its (real) reference files instead of handing off detail to them inline.

Suggestions

De-duplicate the autologging and basic-tracking sections (each appears multiple times) and remove marketing stats ('20,000+ organizations', '23k+ stars') to tighten token efficiency.

Add explicit validation/verification steps for destructive and batch operations — e.g., after a stage transition, verify with client.get_latest_versions(); after deployment, curl the /invocations endpoint and check the response before declaring success.

Move the bulk of tracking, model-registry, and deployment detail into the existing references and replace the inlined sections with inline pointers (e.g., 'See references/tracking.md for nested params and step metrics') instead of a single buried 'See Also' list.

Dimension	Reasoning	Score
Conciseness	The body is mostly code-driven and usable, but repeats content (autologging appears ~4 times, basic start_run/log_param logging is shown in Quick Start then again in Core Concepts) and includes marketing fluff ('Users: 20,000+ organizations \| GitHub Stars: 23k+'). It is not a 4 because the repetition and padded stats are clearly trimmable, and not a 2 because the bulk is executable code rather than prose padding.	3 / 5
Actionability	Provides concrete, executable code for tracking, autologging, registry, searching, and deployment across frameworks, covering the common cases. It is not a 5 because placeholders like train_model()/train_pytorch_model() are undefined and some snippets are partial rather than fully copy-paste ready.	4 / 5
Workflow Clarity	Topics are organized but presented as reference catalogs rather than sequenced workflows, and destructive/batch operations (model stage transitions with archive_existing_versions, production deployment) lack validation or feedback loops; per the rubric this caps workflow clarity at 3. It is not a 4 because no explicit validation checkpoints are present.	3 / 5
Progressive Disclosure	Three real reference files exist and are listed, but they are only signaled in a 'See Also' block at the very bottom rather than inline at the relevant sections, and the body duplicates tracking/registry/deployment content that also lives in those references. It is not a 4 because references are buried and inlined content that belongs in separate files is not clearly handed off.	3 / 5
	Total	13 / 20 Passed

Description

75%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 clearly and concretely states what MLflow does and names a distinct platform, but it omits any explicit 'Use when...' trigger guidance, which caps its completeness. Trigger-term coverage is good but lacks synonyms and format extensions.

Suggestions

Add an explicit 'Use when...' clause naming concrete triggers (e.g., 'Use when tracking ML experiments, managing a model registry, versioning models, or deploying models to production').

Include natural synonyms and extensions users might say, such as 'experiment tracking', 'MLOps', 'model serving', and '.pkl'/'HuggingFace' to broaden trigger-term coverage.

Trim the trailing tagline ('framework-agnostic ML lifecycle platform') which adds little trigger value and edges toward marketing fluff.

Dimension	Reasoning	Score
Specificity	Lists four concrete actions ('Track ML experiments', 'manage model registry with versioning', 'deploy models to production', 'reproduce experiments'), matching the comprehensive-coverage anchor. It is not a 4 because the action list is broad and specific rather than having minor gaps.	5 / 5
Completeness	Clearly states what the skill does but provides no 'Use when...' clause or equivalent explicit trigger guidance; per the judging guidelines a missing trigger clause caps completeness at 3. It is not a 4 because 'when' is only weakly implied by the action verbs.	3 / 5
Trigger Term Quality	Includes natural practitioner phrases ('Track ML experiments', 'model registry', 'deploy models', 'reproduce experiments'), but lacks synonyms and file/format extensions like the 5-anchor example. It is not a 3 because keyword coverage is genuinely good, and not a 5 because common variations and synonyms are missing.	4 / 5
Distinctiveness Conflict Risk	Names a specific platform ('MLflow') with a distinct ML-lifecycle niche and clear triggers, giving minimal conflict risk. It is not a 4 because the named-tool scope leaves little overlap with other skills.	5 / 5
	Total	17 / 20 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (706 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/ml-training/mlflow/SKILL.md
Commit: fcd2e11

Reviewed: 2 days ago

Table of Contents

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.