CtrlK
BlogDocsLog inGet started
Tessl Logo

tessl-labs/skill-optimizer

Optimize your skills and tiles: review SKILL.md quality, generate eval scenarios, run evals, compare across models, diagnose gaps, and re-run until scores improve.

88

1.07x
Quality

94%

Does it follow best practices?

Impact

88%

1.07x

Average score across 24 eval scenarios

SecuritybySnyk

Passed

No known issues

Overview
Quality
Evals
Security
Files

criteria.jsonevals/scenario-15/

{
  "context": "Tests whether the agent can identify orphaned files in a skill bundle (files that exist but are never referenced in SKILL.md) and provide appropriate recommendations. The bundle contains 8 files: SKILL.md references MIGRATIONS.md and POOLING.md, leaving 5 orphaned files (TRANSACTIONS.md, PERFORMANCE.md, SECURITY.md should be linked; LEGACY_EXAMPLES.md and DRAFT_REPLICATION.md should be removed).",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Lists all bundle files",
      "description": "Identifies all 8 files in the bundle (SKILL.md, MIGRATIONS.md, POOLING.md, TRANSACTIONS.md, PERFORMANCE.md, SECURITY.md, LEGACY_EXAMPLES.md, DRAFT_REPLICATION.md)",
      "max_score": 10
    },
    {
      "name": "Identifies referenced files",
      "description": "Correctly identifies that MIGRATIONS.md and POOLING.md are referenced in SKILL.md",
      "max_score": 10
    },
    {
      "name": "Identifies orphaned files",
      "description": "Correctly identifies all 5 orphaned files (TRANSACTIONS.md, PERFORMANCE.md, SECURITY.md, LEGACY_EXAMPLES.md, DRAFT_REPLICATION.md) - files that exist but are never linked from SKILL.md",
      "max_score": 15
    },
    {
      "name": "TRANSACTIONS.md recommendation",
      "description": "Recommends linking TRANSACTIONS.md (valuable content about atomic operations) with clear routing signals like 'for atomic operations and rollback patterns' or similar",
      "max_score": 10
    },
    {
      "name": "PERFORMANCE.md recommendation",
      "description": "Recommends linking PERFORMANCE.md (valuable optimization content) with clear routing signals like 'for indexing strategies and batch operation optimization' or similar",
      "max_score": 10
    },
    {
      "name": "SECURITY.md recommendation",
      "description": "Recommends linking SECURITY.md (valuable security content) with clear routing signals like 'for credential management and connection security practices' or similar",
      "max_score": 10
    },
    {
      "name": "LEGACY_EXAMPLES.md recommendation",
      "description": "Recommends REMOVING LEGACY_EXAMPLES.md - it's deprecated content from v1.x that shouldn't be in the bundle anymore",
      "max_score": 10
    },
    {
      "name": "DRAFT_REPLICATION.md recommendation",
      "description": "Recommends REMOVING DRAFT_REPLICATION.md - it's unfinished/unimplemented content that shouldn't be in a published tile",
      "max_score": 10
    },
    {
      "name": "Bloat reduction framing",
      "description": "Frames orphaned files as adding bloat to the tile - mentions tile size, unnecessary files, or reducing bundle size",
      "max_score": 5
    },
    {
      "name": "Clear routing signals emphasis",
      "description": "When recommending links, emphasizes the need for clear routing signals (WHEN to open each file) rather than generic 'see FILE.md for more info' links",
      "max_score": 5
    },
    {
      "name": "Link vs remove justification",
      "description": "Provides clear reasoning for why some orphaned files should be linked (valuable content) vs removed (deprecated, unfinished, or no longer relevant)",
      "max_score": 5
    }
  ]
}

evals

README.md

tile.json