Name: tessl/plugin-creator
Rating: 93.77 (1 reviews)
Author: tessl

tessl/plugin-creator

Create a Tessl plugin: turn agent workflows and guardrails into a versioned bundle of skills, rules, hooks, and MCP servers, then validate and publish it.

1.31x

Quality

88%

Does it follow best practices?

Impact

95%

1.31x

Average score across 9 eval scenarios

Securityby

Low

Low-risk findings worth noting

Evaluation results

97%

13%

Build a Claude Code idle-notification hook

Use nativeHooks only for an agent-specific lifecycle event

Criteria

Baseline

With context

Native hook tier selected

100%

Agent-specific event represented

87%

100%

Bundled entrypoint anchored

78%

Hook entrypoint ships

100%

Native boundary justified

100%

No false generic workaround

100%

No direct Claude config edit

100%

Valid plugin metadata

100%

Build a cross-agent post-edit validation plugin

Build a cross-agent PostToolUse hook with a bundled entrypoint

Criteria

Baseline

With context

Generic PostToolUse hook declared

100%

Edit and Write matcher

100%

Valid command hook shape

100%

Plugin path anchored correctly

100%

Hook entrypoint ships

100%

Generic tier preferred

100%

No prose-only substitute

100%

No direct agent config edits

100%

Valid plugin metadata

100%

31%

Build a live component-search MCP plugin

Build a plugin-bundled HTTP MCP server declaration

Criteria

Baseline

With context

Root MCP file created

100%

Valid MCP envelope

100%

Correct HTTP server config

50%

100%

Explicit manifest pointer

100%

Capability not prose

66%

100%

No credentials invented

100%

No direct agent config edits

60%

100%

No unrelated primitives

25%

100%

Valid plugin metadata

100%

17%

Build a PR Review Skill for the Engineering Team

Tessl plugin scaffolding conventions

Criteria

Baseline

With context

Verb-form skill name

100%

plugin.json name format

100%

plugin.json semver version

100%

plugin.json description present

100%

SKILL.md description covers what

100%

SKILL.md description covers when

50%

100%

References directory used

100%

SKILL.md links to references

100%

Single-skill shape chosen

100%

plan.md justifies shape choice

100%

82%

68%

Package a commit-message skill as a Tessl plugin

Author location — avoid the .tessl/plugins install cache, use a repo-owned source dir

Criteria

Baseline

With context

Source NOT under .tessl/plugins/

100%

Repo-owned source directory used

100%

Install cache left intact

50%

100%

Wired into tessl.json via file:

Valid plugin produced

80%

100%

22%

Plugin Shape Planning for a Backend Engineering Team

Primitive selection and composition planning

Criteria

Baseline

With context

Skill for PR review

100%

Rule for import convention

100%

MCP server for metrics API

50%

100%

Cross-agent hook for post-edit validation

33%

100%

Justification for each primitive

87%

100%

No padding

60%

100%

Plan is concise

70%

100%

Plugin shape stated

90%

100%

No secrets hard-coded for MCP

100%

Skill vs rule distinction clear

100%

47%

Publishing the data-pipeline-tools Plugin

Plugin publish checklist with provenance, visibility, lint, dry-run, auth, and versioning

Criteria

Baseline

With context

Provenance concern raised

100%

Public visibility confirmed

100%

Public is irreversible warning

100%

Lint step included

100%

Dry-run step included

58%

100%

Auth requirement mentioned

100%

Version bump step

100%

Retirement via archive, not unpublish

40%

100%

83%

Refactor a Sprawling Database Admin Skill

Decompose monolithic skill into focused sub-skills

Criteria

Baseline

With context

Benefit-led rationale

100%

83%

Independent verifiability mentioned

30%

100%

No eval-running

100%

Verb-form skill names

30%

One responsibility per skill

100%

Shared material in references/

80%

Orchestrator or explicit delegation

70%

80%

Cross-skill dependencies explicit

50%

plugin.json present

100%

Decomposition structure shown

100%

95%

-5%

Skill Improvement Context Analysis

Infer vs ask — context analysis without over-questioning

Criteria

Baseline

With context

Identifies test gap

100%

Identifies migration gap

100%

Identifies smoke test gap

100%

Identifies env var gap

100%

No redundant questions

100%

70%

Questions target unknowns only

100%

80%

No fabricated evidence

100%

Preserves original intent

100%

Description has 'what'

100%

Description has 'when'

100%

Sources cited

100%

No invented PR feedback

100%

Evaluated: 7 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Build a PR Review Skill for the Engineering Team Refactor a Sprawling Database Admin Skill Publishing the data-pipeline-tools Plugin Skill Improvement Context Analysis Package a commit-message skill as a Tessl plugin Plugin Shape Planning for a Backend Engineering Team Build a cross-agent post-edit validation plugin Build a Claude Code idle-notification hook Build a live component-search MCP plugin