symbolic-regression

Discover governing equations from data using PySR (evolutionary symbolic regression). Physics-constrained search with dimensional analysis, custom operators, and complexity-accuracy tradeoffs. Use when you need an interpretable equation, not a black-box model.

Quality

77%

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 ./backend/cli/skills/physics/symbolic-regression/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 well-structured, highly actionable body dominated by executable code with clear sequences and feedback tables, weakened mainly by progressive disclosure: everything is inlined in a 230-line monolith with no reference files, plus a few minor code-consistency gaps. The skill would score higher with a references/ split and tightened constructor boilerplate.

Suggestions

Move the Key Parameters table and Troubleshooting table into a references/ file (e.g., references/parameters.md) and keep SKILL.md as a lean overview with well-signaled pointers.

Fix workflow 3 by declaring the custom operator in unary_operators (e.g., "inv(x) = 1/x", as correctly done in workflow 4) instead of relying on extra_sympy_mappings for an operator that is never enabled.

Either show a real dimensional-analysis mechanism in workflow 2 or drop the "Dimensional constraints (optional but powerful)" comment and commented-out variable_names line, since PySR dimensional constraint usage is only implied, never demonstrated.

Dimension	Reasoning	Score
Conciseness	The body is code-dominated and avoids explaining concepts Claude already knows (the PySR overview paragraph is skill-specific, not padding). Minor trimming opportunities remain — workflows 1 and 3 duplicate the full constructor boilerplate, and the Key Parameters table partially restates options already shown and commented in the examples — placing it at the 4 anchor rather than 5.	4 / 5
Actionability	All five workflows contain executable, copy-paste-ready Python with realistic parameters, plus an installation command and a troubleshooting table. Not 5 because of small correctness gaps: workflow 3 defines extra_sympy_mappings for "inv" without ever declaring inv in unary_operators, and workflow 2's "Dimensional constraints" are only a commented-out variable_names line with no actual dimensional-analysis mechanism shown.	4 / 5
Workflow Clarity	Workflows are numbered and clearly sequenced, installation is a distinct step, the Tips section includes held-out validation ("Test on held-out data and check dimensional consistency"), and the Troubleshooting table provides symptom-to-fix feedback loops. Not 5 because the workflows themselves lack explicit validation checkpoints (e.g., checking model.convergence_ or score thresholds before accepting an equation), matching the 4 anchor's 'minor validation gaps'.	4 / 5
Progressive Disclosure	No bundle files exist and all ~230 lines live inline in SKILL.md. Section structure is good (headers, tables, numbered workflows), but reference material — the Key Parameters table and Troubleshooting table — and multiple example workflows could be split into references/ files, leaving a leaner overview. This matches the 3 anchor: some structure, but content that should be separate is inline and there are no external references at all.	3 / 5
	Total	15 / 20 Passed

Description

83%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, specific description that names concrete capabilities, an explicit use-when clause, and a distinct niche. Its main weakness is trigger coverage: it misses natural synonyms and user phrasings (e.g., "equation discovery", "fit a formula to data") that would broaden natural invocation.

Suggestions

Add natural user phrasings to the when-clause, e.g., "Use when the user asks for symbolic regression, equation discovery, or wants to fit an interpretable formula to data instead of a black-box model."

Include a couple of synonyms such as "equation discovery" or "formula fitting" so the description matches how users naturally phrase the request.

Dimension	Reasoning	Score
Specificity	The description lists multiple concrete capabilities — "Discover governing equations from data using PySR", "Physics-constrained search with dimensional analysis, custom operators, and complexity-accuracy tradeoffs" — giving comprehensive coverage of what the skill does. It matches the 5 anchor rather than 4 because there are no meaningful gaps in the action inventory for this domain.	5 / 5
Completeness	Both parts are present: a clear what ("Discover governing equations from data using PySR...") and an explicit when ("Use when you need an interpretable equation, not a black-box model"). Not 5 because the when-clause is a single abstract condition rather than concrete trigger phrases users would say, matching the 4 anchor ('when' could be more explicit or specific).	4 / 5
Trigger Term Quality	Good natural-keyword coverage ("governing equations", "symbolic regression", "PySR", "interpretable equation", "black-box model"), but common user phrasings like "equation discovery", "fit a formula to data", or "find an equation from data" are missing. It sits between the 3 and 5 anchors, noticeably above the midpoint.	4 / 5
Distinctiveness Conflict Risk	It carves out a clear niche (PySR/symbolic regression for equation discovery) and explicitly disambiguates from generic ML with "an interpretable equation, not a black-box model". Trigger terms like PySR and symbolic regression are distinct, giving minimal conflict risk, matching the 5 anchor.	5 / 5
	Total	18 / 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: synthetic-sciences/openscience
Path: backend/cli/skills/physics/symbolic-regression/SKILL.md
Commit: 4082a2e

Reviewed: 1 day 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.