Name: tessl-labs/skill-optimizer
Rating: 0.892 (1 reviews)
Author: tessl-labs

tessl-labs/skill-optimizer

Improves your SKILL.md using tessl skill review plus validation and context. Reads skill bundle (SKILL.md + related files), validates syntax, explains WHY changes help, catches mistakes. Use when improving your own skill, skill file, skill description, reviewing skill quality, skill scoring, making your skill better, or learning the skill rubric.

1.25x

Quality

100%

Does it follow best practices?

Impact

88%

1.25x

Average score across 12 eval scenarios

Evaluation results

94%

26%

Skill Bundle Validation

Phase 4 syntax and reference validation

Criteria

Without context

With context

Python via ast.parse

100%

Python error identified

100%

JavaScript via node --check

100%

Command flag validation

40%

File reference check

53%

100%

Broken reference identified

100%

Validation before application

100%

Per-check pass/fail

100%

Fix suggestions

100%

Without context: $0.3517 · 1m 44s · 19 turns · 22 in / 5,392 out tokens

With context: $0.4499 · 2m 21s · 19 turns · 1,048 in / 6,466 out tokens

84%

Skill Improvement Recommendations

Prioritized recommendation generation from review output

Criteria

Without context

With context

Critical issues first

100%

High before Medium/Low

100%

Summary with priorities

40%

Expected improvement in summary

100%

Dimension score included

70%

80%

Before/after examples

75%

83%

Impact stated per recommendation

50%

75%

Educational WHY included

100%

All four issues addressed

100%

Approval framing

70%

60%

Without context: $0.1588 · 1m 32s · 8 turns · 13 in / 2,702 out tokens

With context: $0.2183 · 1m 23s · 13 turns · 1,041 in / 3,685 out tokens

100%

Skill Length Reduction

Progressive disclosure via reference file linking

Criteria

Without context

With context

Linking over inlining

100%

Reference file identified

100%

Severity mappings removed

100%

Flag tables removed

90%

100%

Template list removed

100%

SKILL.md substantially shorter

100%

Core examples preserved

100%

Before/after shown

100%

WHY explained

100%

REFERENCE.md not modified

100%

Without context: $0.2716 · 1m 33s · 12 turns · 19 in / 4,347 out tokens

With context: $0.6264 · 3m 6s · 26 turns · 1,057 in / 8,378 out tokens

100%

Skill Optimization Results Report

Phase 7 before/after score comparison

Criteria

Without context

With context

Overall before/after format

100%

Percentage delta shown

100%

Per-dimension breakdown

100%

Arrow notation or equivalent

100%

Dimension change labelled

100%

Dimensions impact explained

100%

Correct overall scores

100%

Completeness improvement noted

100%

Actionability improvement noted

100%

Conciseness unchanged noted

100%

Robustness improvement noted

100%

Without context: $0.1441 · 42s · 9 turns · 14 in / 1,956 out tokens

With context: $0.2314 · 1m 11s · 13 turns · 1,075 in / 3,272 out tokens

100%

29%

Skill Post-Edit Quality Audit

Phase 8 final accuracy check (5 criteria)

Criteria

Without context

With context

Code syntax check included

100%

Python syntax error found

100%

Command flags check included

100%

File references check included

100%

File reference passes

100%

Use when clause check included

100%

Use when clause fails

100%

Known concepts check included

70%

100%

Known concepts issue found

83%

100%

Readiness summary

100%

Without context: $0.2139 · 1m 10s · 10 turns · 15 in / 3,070 out tokens

With context: $0.5286 · 2m 47s · 25 turns · 325 in / 7,485 out tokens

92%

64%

Skill Optimization Automation

tessl skill review command and workflow scripting

Criteria

Without context

With context

tessl skill review command

100%

Review before changes

100%

Review after changes

100%

Validation before apply

100%

Python ast.parse validation

100%

node --check JS validation

100%

Command --help flag validation

File reference validation

100%

Before/after score output

87%

100%

Script accepts SKILL.md path

100%

Phases are ordered

50%

100%

Without context: $0.2895 · 1m 31s · 12 turns · 15 in / 5,460 out tokens

With context: $0.5264 · 2m 30s · 23 turns · 1,053 in / 7,963 out tokens

69%

34%

Skill Score Maximization

Scoring dimension weights and quick wins

Criteria

Without context

With context

Completeness weight correct

Conciseness weight correct

Actionability weight correct

Use when clause highest impact

33%

100%

Use when quantified

50%

Revised description includes Use when

100%

Executable code recommended

100%

Known concepts flagged

50%

100%

High-impact first ordering

10%

100%

Dimension coverage

100%

Without context: $0.1991 · 1m 13s · 12 turns · 16 in / 3,206 out tokens

With context: $0.4782 · 2m 37s · 24 turns · 1,054 in / 6,921 out tokens

100%

Skill Quality Improvement

Only modify SKILL.md, not other bundle files

Criteria

Without context

With context

REFERENCE.md not recreated

100%

No REFERENCE.md changes proposed

100%

SKILL.md produced

100%

Use when clause added

100%

Inline duplication removed

100%

REFERENCE.md linked

100%

Core examples retained

100%

SKILL.md shorter

100%

Change log documents SKILL.md changes

100%

Change log explains why

100%

Without context: $0.3107 · 1m 44s · 19 turns · 26 in / 3,792 out tokens

With context: $0.4584 · 2m 18s · 19 turns · 1,047 in / 6,663 out tokens

100%

14%

Skill Improvement Proposal

Approval-gated recommendations before applying

Criteria

Without context

With context

SKILL.md not modified

100%

Priority summary present

50%

100%

Expected improvement in summary

25%

100%

Current score per recommendation

100%

Before/after per recommendation

83%

100%

Impact per recommendation

100%

Educational WHY per recommendation

100%

Critical first

100%

Approval framing

100%

Without context: $0.1453 · 53s · 10 turns · 14 in / 2,599 out tokens

With context: $0.2988 · 1m 41s · 11 turns · 1,040 in / 4,599 out tokens

46%

Skill Portfolio Improvement Plan

bulk-optimize redirect for large-scale skill work

Criteria

Without context

With context

bulk-optimize mentioned

57%

bulk-optimize for PR/repo work

72%

Single-skill approach not recommended

10%

Rationale provided

30%

Without context: $0.1654 · 58s · 9 turns · 14 in / 2,372 out tokens

With context: $0.3006 · 1m 53s · 15 turns · 347 in / 4,660 out tokens

85%

-1%

Progressive Disclosure Evaluation

Criteria

Without context

With context

Identifies good references

100%

Explains why good

100%

Identifies poor references

100%

Explains why poor

100%

Token efficiency framing

60%

80%

Routing gate test

100%

70%

Improves CONFIGURATION.md

100%

Improves GUIDE.md

100%

Improves EXAMPLES.md

100%

Improves ADVANCED.md or REFERENCE.md

100%

Questions blind split recommendation

Without context: $0.1705 · 57s · 8 turns · 13 in / 2,961 out tokens

With context: $0.2769 · 1m 46s · 13 turns · 1,074 in / 5,453 out tokens

97%

Bundle File Audit

Criteria

Without context

With context

Lists all bundle files

100%

Identifies referenced files

100%

Identifies orphaned files

100%

TRANSACTIONS.md recommendation

100%

PERFORMANCE.md recommendation

100%

SECURITY.md recommendation

80%

100%

LEGACY_EXAMPLES.md recommendation

100%

DRAFT_REPLICATION.md recommendation

100%

Bloat reduction framing

40%

Clear routing signals emphasis

80%

100%

Link vs remove justification

100%

Without context: $0.1513 · 49s · 8 turns · 13 in / 2,103 out tokens

With context: $0.2708 · 1m 15s · 12 turns · 15 in / 3,653 out tokens

Install with Tessl CLI

npx tessl i tessl-labs/skill-optimizer

Evaluated: about 4 hours ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Skill Bundle Validation Skill Improvement Recommendations Skill Length Reduction Skill Optimization Results Report Skill Post-Edit Quality Audit Skill Optimization Automation Skill Score Maximization Skill Quality Improvement Skill Improvement Proposal Skill Portfolio Improvement Plan Progressive Disclosure Evaluation Bundle File Audit