modal-compute

Run explicitly chosen research benchmark or replication jobs on Modal's serverless infrastructure. Use when a Feynman research workflow needs burst remote GPU compute and the Modal CLI is available.

Quality

79%

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 ./skills/modal-compute/SKILL.md

Quality

Content

80%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 tight, highly actionable skill body with excellent progressive disclosure for its size. The main weakness is the absence of any validation/verification checkpoint for what is inherently a batch remote-compute workflow, which caps workflow clarity.

Suggestions

Add an explicit post-run validation step, e.g., after 'modal run' check the returned app/run identifier and confirm raw outputs landed in the research artifact folder before declaring success.

Move the pinned 'torch==2.8.0' / 'python_version' out of the canonical example or into a noted 'reproducibility pin' aside so the primary pattern does not carry time-sensitive version drift.

Add a brief 'Save outputs' step to the script pattern (e.g., writing results to a mounted volume or the artifact folder) so the run-to-output workflow is fully sequenced.

Dimension	Reasoning	Score
Conciseness	The body is lean — a commands table, a flat GPU list, and a single copy-paste script pattern with minimal prose — but the pinned 'torch==2.8.0' and 'python_version="3.11"' are time-sensitive version numbers not placed in a deprecated/old-patterns section, which per the rubric guidelines is a minor over-explanation/fragility cost keeping it just below 5.	4 / 5
Actionability	Provides fully executable, copy-paste-ready guidance: concrete 'modal run'/'modal shell' commands, a complete decorated App script with gpu/timeout/image, and concrete multi-GPU syntax — covering the common run case end to end.	5 / 5
Workflow Clarity	Sections (Setup, Commands, Script pattern, When to use) imply a rough sequence, but this is a batch/remote-compute skill with no explicit validation or verification checkpoint (e.g., confirm run success / inspect saved outputs), and the rubric caps batch-operation skills without validation at 3.	3 / 5
Progressive Disclosure	The body is under 50 lines with no external references needed, and is cleanly organized into well-labeled sections (Setup, Commands, GPU types, Script pattern, When to use), meeting the simple-skill exception for a top progressive-disclosure score.	5 / 5
	Total	17 / 20 Passed

Description

78%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 focused, well-structured description that explicitly covers both capability and trigger conditions with a clear Modal-specific niche. Slight gains available from adding more concrete action variety and replacing the internal 'Feynman research workflow' phrasing with broader natural trigger terms.

Suggestions

Broaden specificity by listing a second distinct action (e.g., '...save raw outputs back to the research artifact folder') so the description names multiple concrete operations rather than one job type.

Replace or augment 'Feynman research workflow' with user-natural phrasing like 'research experiments' or 'replication studies' so the trigger reads as something a user would actually say.

Add a common synonym or file/CLI cue (e.g., 'modal run') to strengthen trigger-term coverage toward comprehensive.

Dimension	Reasoning	Score
Specificity	Names the domain (Modal serverless) and a single concrete action category — 'Run explicitly chosen research benchmark or replication jobs' — but does not enumerate multiple distinct actions, matching the '1-2 concrete actions, not comprehensive' anchor rather than the 'several specific actions' anchor at 4.	3 / 5
Completeness	Explicitly answers both 'what' (run benchmark/replication jobs on Modal serverless) and 'when' via a concrete 'Use when a Feynman research workflow needs burst remote GPU compute and the Modal CLI is available' trigger clause, matching the anchor for clearly answering both with concrete triggers.	5 / 5
Trigger Term Quality	Includes reasonably natural terms ('research benchmark', 'replication jobs', 'burst remote GPU compute', 'Modal CLI') with good coverage, though 'Feynman research workflow' is internal jargon and a few common synonyms are missing, placing it just above the midpoint rather than at comprehensive 5.	4 / 5
Distinctiveness Conflict Risk	The Modal-specific niche plus 'research benchmark/replication' scoping is mostly distinct with minimal overlap, but it could still collide with other remote-GPU/compute-orchestration skills, fitting 'mostly distinct; minor overlap risk' rather than the fully-clear 5.	4 / 5
	Total	16 / 20 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: companion-inc/feynman
Path: skills/modal-compute/SKILL.md
Commit: 95c51d9

Reviewed: about 9 hours 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.