tensorboard

Visualize training metrics, debug models with histograms, compare experiments, visualize model graphs, and profile performance with TensorBoard - Google's ML visualization toolkit

Quality

70%

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/ml-training/tensorboard/SKILL.md

Quality

Content

75%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 content is highly actionable with comprehensive executable examples for both major frameworks and well-organized sections pointing to real reference files. Its main weakness is length/conciseness: it repeats scalar-logging patterns and inlines large training-loop examples that partly duplicate earlier sections and could be trimmed or offloaded to the reference files.

Suggestions

Consolidate the repeated scalar-logging examples and trim the full PyTorch/TF integration training loops, offloading the long-form versions to references/integrations.md to reduce token cost.

Add a brief validation checkpoint in the profiling and logging workflows (e.g. confirm the writer flushed and TensorBoard serves the runs directory before declaring success).

Move the Advanced Features detail (embeddings, hparams, PR curves, text) into the existing reference files, keeping SKILL.md as a tighter overview with one representative snippet each.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient with runnable code blocks, but it is long (~470 lines) and re-explains core concepts Claude already knows (e.g. restating SummaryWriter usage across multiple near-duplicate scalar-logging examples) and pads the integration section with a full training loop that overlaps earlier examples, so it could be tightened.	3 / 5
Actionability	Nearly every section provides copy-paste-ready, executable code for both PyTorch and TensorFlow covering the common cases (scalars, images, histograms, graphs, embeddings, hparams, profiling), with concrete launch commands, fully actionable.	5 / 5
Workflow Clarity	Multi-step processes (training loop, profiling, comparing experiments) are clearly sequenced with best-practice checkpoints like 'log every 100 batches' and 'close writer when done', but there are no explicit validation/error-recovery feedback loops (e.g. verifying logs opened correctly or recovering from a failed write), leaving minor validation gaps.	4 / 5
Progressive Disclosure	Structure is good with clearly labeled sections and a 'See Also' pointing to three real one-level-deep reference files (visualization.md, profiling.md, integrations.md), but the SKILL.md still inlines substantial material (full integration training loops, advanced features) that could live in those references, so it is not maximally split.	4 / 5
	Total	16 / 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.

The description is concrete and tool-specific with good action coverage and natural trigger terms, but it lacks an explicit 'Use when...' trigger clause, which caps its completeness and leaves the 'when' only weakly implied. Adding a clear invocation trigger would lift it from solid to excellent.

Suggestions

Add an explicit 'Use when...' clause, e.g. 'Use when training or debugging ML models and the user wants to visualize metrics, compare runs, or profile performance.'

Include a few more natural synonyms/trigger phrases such as 'log metrics', 'experiment tracking', or 'monitor training' to broaden trigger coverage.

Mention embedding projection and image/text logging in the description to make capability coverage comprehensive.

Dimension	Reasoning	Score
Specificity	Lists several concrete actions ('Visualize training metrics', 'debug models with histograms', 'compare experiments', 'visualize model graphs', 'profile performance') naming the TensorBoard domain, but coverage overlaps somewhat and a few capabilities from the body (embeddings, text/images) are omitted, keeping it just below a 5.	4 / 5
Completeness	It clearly states what the skill does (multiple concrete visualization/debugging actions) but provides no explicit 'Use when...' clause telling Claude when to invoke it; the 'when' is only weakly implied by the action list, which per the guidelines caps completeness at 3.	3 / 5
Trigger Term Quality	Includes natural terms users would say ('training metrics', 'experiments', 'model graphs', 'profile performance') and the tool name, but lacks common synonyms like 'logging metrics', 'experiment tracking' as explicit triggers and no file extensions, so a few natural phrases are missing.	4 / 5
Distinctiveness Conflict Risk	The TensorBoard-specific framing ('TensorBoard - Google's ML visualization toolkit') gives it a clear niche with minimal conflict risk against unrelated skills, though it could overlap with generic ML experiment-tracking skills, leaving minor overlap risk.	4 / 5
	Total	15 / 20 Passed

Validation

81%

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 — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (639 lines); consider splitting into references/ and linking	Warning
metadata_version	'metadata.version' is missing	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	13 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/ml-training/tensorboard/SKILL.md
Commit: 4e060d6

Reviewed: 13 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.