pyvene-interventions

Provides guidance for performing causal interventions on PyTorch models using pyvene's declarative intervention framework. Use when conducting causal tracing, activation patching, interchange intervention training, or testing causal hypotheses about model behavior.

Quality

79%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./skills/mechanistic-interpretability/pyvene/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, actionable skill body with four concrete workflows and clean one-level-deep references. The main weaknesses are mild verbosity (repeated imports, supplementary lists) and a couple of incomplete code examples.

Suggestions

Consolidate the repeated `import pyvene as pv` / `from transformers import ...` boilerplate into a single initial snippet and reference it, or strip it from later workflows to save tokens.

Complete the IIT training-loop example by defining `dataloader` and `criterion` or by adding a note that they are user-supplied, so the code is fully runnable.

Consider moving the Supported Models, External Resources, and Comparison tables into references/ to keep the SKILL.md body a lean overview.

Dimension	Reasoning	Score
Conciseness	Mostly efficient with no basic-concept padding, but repeats import boilerplate across four workflows and includes sections (Supported Models, External Resources, Comparison table) that add length without proportionate instruction; could be tightened.	3 / 5
Actionability	Provides copy-paste-ready executable code across four workflows plus common-issues fixes, but the IIT example references undefined `dataloader` and `criterion`, leaving a minor gap in executability.	4 / 5
Workflow Clarity	Workflows are clearly sequenced with numbered/commented steps and a checklist for causal tracing; validation checkpoints are mostly implicit rather than explicit, but these are read-only analysis workflows rather than destructive or state-mutating batch operations.	4 / 5
Progressive Disclosure	The body is an overview with a clearly signaled Reference Documentation table linking one level deep to the real references/README.md, api.md, and tutorials.md files, keeping detailed API and tutorial material appropriately split out.	5 / 5
	Total	16 / 20 Passed

Description

87%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, third-person description that clearly states both capability and trigger conditions with concrete domain-specific terms. Minor gaps in action/keyword coverage keep specificity and trigger quality just below full marks.

Dimension	Reasoning	Score
Specificity	Names the domain and lists several concrete actions ("causal tracing", "activation patching", "interchange intervention training", "testing causal hypotheses") but omits related capabilities like model steering and saving/sharing interventions, leaving minor coverage gaps.	4 / 5
Completeness	Explicitly answers both "what" ("Provides guidance for performing causal interventions on PyTorch models using pyvene's declarative intervention framework") and "when" ("Use when conducting causal tracing, activation patching, interchange intervention training, or testing causal hypotheses") with concrete trigger phrases.	5 / 5
Trigger Term Quality	Includes natural interpretability terms users would say ("causal tracing", "activation patching", "interchange intervention training", "causal hypotheses"), but lacks common synonyms like DAS, ROME, circuit analysis, or model steering.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (pyvene declarative causal interventions) with distinctive, specialized triggers that are unlikely to fire for unrelated skills.	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: OpenLAIR/dr-claw
Path: skills/mechanistic-interpretability/pyvene/SKILL.md
Commit: 4b35756

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.