agentic-evaluator

Evaluates any repository's agentic development maturity. Use when auditing a codebase for best practices in agents, skills, instructions, MCP config, and prompts. Produces a scored report with specific remediation steps.

Install with Tessl CLI

npx tessl i github:0xrabbidfly/eric-cartman --skill agentic-evaluator

What are skills?

1.50x

Quality

81%

Does it follow best practices?

Impact

93%

1.50x

Average score across 3 eval scenarios

SKILL.md

Review

Evals

Evaluation results

97%

40%

Agentic Maturity Audit: DataSync Platform

Foundation scoring and lean context evaluation

Criteria

Without context

With context

Noise items flagged

76%

100%

Signal items preserved

100%

Root instructions quality scored

100%

Missing .github/ structure flagged

25%

100%

Missing MCP config flagged

100%

Discovery scan locations

25%

62%

Score breakdown table

37%

100%

Correct grade assigned

100%

Issues by priority

100%

Recommendations included

100%

README AI documentation credited

14%

100%

Lean context principle cited

100%

Without context: $0.2465 · 1m 43s · 8 turns · 13 in / 5,163 out tokens

With context: $0.6968 · 3m 16s · 20 turns · 4,797 in / 11,153 out tokens

87%

32%

Skills Audit: FinTrack API Platform

Skills quality, size guidelines, and progressive disclosure

Criteria

Without context

With context

Oversized skill flagged

66%

100%

Progressive disclosure compliance

80%

70%

Remediation file structure

70%

90%

Quality dimension ratings

20%

100%

Quality dimension total scored

100%

Frontmatter validation - api-development

25%

62%

Frontmatter validation - kafka-consumer

25%

62%

SkillsBench findings applied

87%

100%

Domain sensitivity noted

37%

100%

Right-sized skill acknowledged

87%

62%

Skills inventory included

90%

100%

Without context: $0.3401 · 2m 10s · 12 turns · 17 in / 5,537 out tokens

With context: $0.6226 · 3m 30s · 13 turns · 4,786 in / 12,000 out tokens

96%

20%

Agentic Tooling Review: CloudOps Automation Platform

Complete evaluation: agents, instructions, consistency, report generation

Criteria

Without context

With context

TerraformProvisioner naming flagged

100%

TerraformProvisioner tools missing

100%

Unresolved cross-reference flagged

100%

pagerduty-mcp not in MCP config

100%

Instructions missing applyTo

100%

Instructions right-sized

100%

66%

MCP config credits

66%

100%

Root instructions quality credited

66%

Noise in root instructions flagged

100%

P0/P1/P2 issue prioritization

100%

Correct letter grade

100%

Artifacts inventory

50%

100%

Recommendations structure

75%

100%

Without context: $0.3759 · 2m 15s · 12 turns · 17 in / 7,069 out tokens

With context: $0.7324 · 3m 32s · 17 turns · 2,843 in / 12,778 out tokens

Evaluated: 1 day ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Agentic Maturity Audit: DataSync Platform Skills Audit: FinTrack API Platform Agentic Tooling Review: CloudOps Automation Platform

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.