The default eval model has changed to GLM 5.1.

GLM 5.1 brings stronger reasoning and improved accuracy. You can still select any model when starting a run. Read the full announcement →

simon/cli

Private

Auto-generated tile from GitHub (107 skills)

Eval Details

Eval Run Status

Completed

Version

1ff5bc820f52cb485a7f88fce01c16d2b3e48847

Agent

Claude

Model

Claude Sonnet 4.6

Score

Agent success rate when using this plugin

81%

Improvement

Agent success rate improvement when using this plugin compared to baseline

1.8x

Baseline

Agent success rate without this plugin

45%

Evaluation results

100%

70%

Document Content Retrieval Script

Schema-first API inspection

Criteria

Without context

With context

Schema inspection step

15%

100%

Help command reference

100%

Correct CLI resource syntax

100%

Params flag used

100%

Schema-driven flag construction

13%

100%

No secret output

100%

82%

Automated Meeting Notes Updater

Plain text vs rich text appending

Criteria

Without context

With context

+write for plain text

100%

batchUpdate for table

100%

Correct +write flags

100%

batchUpdate uses --json

100%

Reason for split documented

100%

Confirmation before write

No hardcoded credentials

100%

80%

40%

Policy Document Revision Rollout

Batch update safety with dry-run

Criteria

Without context

With context

Dry-run preview pass

22%

100%

Dry-run is separate from live run

100%

User confirmation before live run

Correct batchUpdate syntax

100%

Loops over all IDs

100%

Runbook dry-run explanation

27%

100%

No secrets exposed

100%

82%

67%

Document Inventory Export Tool

Output formatting and pagination

Criteria

Without context

With context

--format flag used

100%

--page-all flag present

100%

--output flag for file saving

FORMAT argument controls --format

100%

Pagination flags documented

100%

Output filename matches format

100%

61%

Legal Document Processing Pipeline

Service account auth and PII screening

Criteria

Without context

With context

GOOGLE_APPLICATION_CREDENTIALS env var

72%

22%

No credential values output

90%

100%

--sanitize flag on get

Credential path as variable

100%

Setup guide covers no-secret rule

100%

Sanitize explained in setup guide

16%

100%