improvement-loop
Use after resolving a bug, failed task, or unexpected agent behavior to improve the pipeline skills, agents, hooks, or scripts that contributed to the problem. Also proactively suggest improvements when recurring patterns or inefficiencies are observed.
Install with Tessl CLI
npx tessl i github:aaddrick/claude-desktop-debian --skill improvement-loop81
Quality
72%
Does it follow best practices?
Impact
93%
1.25xAverage score across 3 eval scenarios
Optimize this skill with Tessl
npx tessl skill review --optimize ./.claude/skills/improvement-loop/SKILL.mdDiscovery
67%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
The description effectively communicates when to use the skill with clear trigger conditions, but lacks specificity in the concrete actions it performs. The trigger terms are adequate but could better match natural user language. The scope is reasonably distinct but may conflict with debugging or refactoring skills.
Suggestions
Add more specific concrete actions like 'update error handling logic', 'add retry mechanisms', 'improve logging', or 'refactor validation rules' to increase specificity.
Include more natural user trigger terms such as 'fix', 'broken', 'error', 'not working', 'keeps failing' that users would actually say when encountering problems.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Names the domain (pipeline improvement) and some actions ('improve the pipeline skills, agents, hooks, or scripts'), but lacks concrete specific actions like 'update error handling', 'add validation', or 'refactor retry logic'. | 2 / 3 |
Completeness | Clearly answers both what ('improve the pipeline skills, agents, hooks, or scripts') and when ('after resolving a bug, failed task, or unexpected agent behavior' and 'when recurring patterns or inefficiencies are observed') with explicit trigger conditions. | 3 / 3 |
Trigger Term Quality | Includes some relevant terms ('bug', 'failed task', 'agent behavior', 'recurring patterns', 'inefficiencies') but misses common user phrases like 'fix', 'broken', 'error', 'not working', 'keeps failing', or 'debug'. | 2 / 3 |
Distinctiveness Conflict Risk | Somewhat specific to pipeline/agent improvement context, but could overlap with general debugging skills, code review skills, or refactoring skills due to broad terms like 'bug' and 'improvements'. | 2 / 3 |
Total | 9 / 12 Passed |
Implementation
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-structured skill with excellent workflow clarity and actionability. The mandatory gate check before improvements and the five-step cycle with verification are particularly strong. However, the skill is verbose for its purpose—some guidance is repeated across sections (gate check, anti-patterns), and the inline tables could be extracted to reference files for better progressive disclosure.
Suggestions
Consolidate redundant content: the gate check appears in both a diagram and a numbered list—keep one format
Extract the routing and verification tables to a separate REFERENCE.md file, keeping only the most common cases inline
Tighten the 'Proactive Detection' section—the example dialogue template is useful but the surrounding explanation is verbose
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is comprehensive but includes some redundancy (the gate check is explained multiple times, anti-patterns are listed in multiple formats). The DOT diagrams add visual clarity but also token overhead. Some sections could be tightened. | 2 / 3 |
Actionability | Provides concrete, actionable guidance throughout: specific commit message format, exact verification steps per change type, clear routing table for which tool to use, and specific example dialogue for asking users. The five-step cycle is immediately executable. | 3 / 3 |
Workflow Clarity | Excellent multi-step workflow with explicit gate check before any improvement work. Clear sequence (capture → classify → change → verify → commit) with validation checkpoints. The mandatory gate check and verification table provide strong feedback loops for this potentially risky operation. | 3 / 3 |
Progressive Disclosure | Content is well-organized with clear sections and tables, but everything is inline in one file. References to writing-skills and writing-agents skills are mentioned but could be better signaled. The skill is long (~200 lines) and could benefit from splitting detailed tables into reference files. | 2 / 3 |
Total | 10 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Reviewed
Table of Contents
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.