ml-strategy

Machine-learning predictive strategy based on sklearn walk-forward training, feature engineering, and signal generation. Suitable for any OHLCV data.

Quality

74%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./agent/src/skills/ml-strategy/SKILL.md

Quality

Content

71%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.

A strong, fully executable skill body with an exemplary complete pipeline example and explicit validation checkpoints, weakened mainly by redundancy between prose tables and code, and by its monolithic structure with no bundle separation. The workflow is clear and safety-conscious, though it lacks a true error-recovery loop.

Suggestions

Move the complete SignalEngine code to a scripts/ file (e.g. scripts/engine.py) and keep a short quick-start snippet in SKILL.md, reducing inline bulk and duplication.

Consolidate the output contract (no NaN, clipped to [-1,1]) — currently stated in Signal Logic, the docstring, the code comment, and Signal Convention — into a single location.

Trim the Feature Engineering and Parameters tables to names plus one-line meanings, dropping formula columns that duplicate the code.

Dimension	Reasoning	Score
Conciseness	No padding of concepts Claude already knows, but real duplication exists: feature formulas appear in both the code and the Feature Engineering table, parameters in both docstrings and the Parameters table, and the output contract ("no NaN, clipped to [-1,1]") is stated in Signal Logic, the docstring, the code comment, and Signal Convention. Mostly efficient but could be tightened.	3 / 5
Actionability	The SignalEngine example is complete, copy-paste-ready, executable code with division-by-zero guards, NaN handling, docstrings, and a pip install command — matching the fully-executable 5 anchor with common cases covered.	5 / 5
Workflow Clarity	Five explicitly numbered Signal Logic steps with validation as step 1 ("Validate input... skip symbols that fail"), enforced in code (validate_data, NaN checks, single-class skip, WARN on skip). Not 5 because there is no error-recovery feedback loop — failures simply skip-and-warn rather than fix-and-retry.	4 / 5
Progressive Disclosure	Well-sectioned and navigable, but monolithic: the ~200-line complete engine code is inlined in SKILL.md with no bundle files, and the reference tables could live in separate files while the code goes to scripts/. Not 4 because the organization gaps are more than minor for a file of this size.	3 / 5
	Total	15 / 20 Passed

Description

66%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 domain-specific description that names concrete techniques and strong niche keywords, but lacks an explicit "Use when..." trigger clause, leaving the "when" only weakly implied and capping completeness. Adding natural user-facing terms (trading strategy, price prediction) would round out trigger coverage.

Suggestions

Add an explicit trigger clause, e.g. "Use when building ML trading strategies, predicting price direction from OHLCV data, or generating sklearn-based trading signals."

Include natural user synonyms such as "trading strategy", "stock price prediction", or "backtest" so the description matches how users actually phrase the request.

Optionally name the supported models (RandomForest / GradientBoosting / Ridge) to sharpen the "what" toward comprehensive coverage.

Dimension	Reasoning	Score
Specificity	Names the domain plus several concrete techniques — "sklearn walk-forward training, feature engineering, and signal generation" — comparable to the 4-anchor's several specific actions. Not 5 because coverage has gaps (no model types, validation, or prediction-target mention).	4 / 5
Completeness	The "what" is clear (ML predictive strategy with named techniques), but "Suitable for any OHLCV data" only weakly implies "when" — there is no "Use when..." clause or equivalent explicit trigger guidance, which caps completeness at 3 per the judging guidelines.	3 / 5
Trigger Term Quality	Good keyword coverage — "Machine-learning", "sklearn", "walk-forward", "OHLCV", "signal generation" — terms a user in this niche would naturally say. Not 5 because common variations like "trading strategy", "stock price prediction", and "backtest" are missing.	4 / 5
Distinctiveness Conflict Risk	"sklearn walk-forward training" plus "OHLCV data" carves a clear quant-trading ML niche that is mostly distinct from other skills, with only minor overlap risk against sibling trading-strategy skills. Not 5 because no explicit trigger phrases further sharpen the niche.	4 / 5
	Total	15 / 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: HKUDS/Vibe-Trading
Path: agent/src/skills/ml-strategy/SKILL.md
Commit: 251b094

Reviewed: 3 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.