Borrowing it
Nothing to install: this file belongs to JosiahSiegel/claude-plugin-marketplace. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/JosiahSiegel/claude-plugin-marketplace/main/.claude/commands/auto-improve-all.mdgit clone --depth 1 https://github.com/JosiahSiegel/claude-plugin-marketplaceWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/josiahsiegel/claude-plugin-marketplace/auto-improve-all)<a href="https://agentmods.dev/commands/josiahsiegel/claude-plugin-marketplace/auto-improve-all"><img src="https://agentmods.dev/badge/commands/josiahsiegel/claude-plugin-marketplace/auto-improve-all/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/josiahsiegel/claude-plugin-marketplace/auto-improve-all"><img src="https://agentmods.dev/badge/commands/josiahsiegel/claude-plugin-marketplace/auto-improve-all.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00018 | $0.03378 |
| Opus 5 | $0.00009 | $0.01689 |
| Sonnet 5 | $0.00004 | $0.00676 |
| Haiku 4.5 | $0.00002 | $0.00338 |
Grade A, and why
auto-improve-all scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 330 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Auto-Improve All Plugins
Purpose
Fully autonomous improvement of every plugin in the claude-plugin-marketplace repository. Each plugin's expert agent will:
- Self-determine what needs improvement using web search and Context7
- Autonomously decide what features/fixes to add
- Complete all improvements themselves (no pausing, no asking)
- Minimize context usage by doing all work within agent contexts
This command uses agent-driven execution to keep primary context window usage minimal while maximizing improvement quality.
Process
DO NOT do the work yourself. Launch agents and let them do ALL the work autonomously.
Step 1: Launch All Plugin Expert Agents in Parallel
Use a single message with multiple Task tool calls to launch all agents simultaneously. This keeps your context usage minimal.
Plugins to Improve (14 total):
- adf-master →
adf-master:adf-expert - ado-master →
ado-master:ado-expert - azure-master →
azure-master:azure-resources-expert - azure-to-docker-master →
azure-to-docker-master:docker-compose-generator - bash-master →
general-purposeagent - context-master →
Exploreagent (very thorough) - docker-master →
docker-master:docker-expert - git-master →
general-purposeagent - plugin-master →
plugin-master:plugin-architect - powershell-master →
powershell-master:powershell-expert - salesforce-master →
salesforce-master:sf-integration-expert - ssdt-master →
ssdt-master:ssdt-expert - terraform-master →
terraform-master:terraform-expert - test-master →
test-master:test-expert
Step 2: Provide Autonomous Improvement Instructions
Give each agent these instructions (they will work independently in their own contexts):
You are the expert for the [PLUGIN_NAME] plugin in the claude-plugin-marketplace repository.
Your task: AUTONOMOUSLY IMPROVE your plugin. You will:
1. Self-determine what needs improvement (no one will tell you what to do)
2. Decide what features/fixes to add (use your expert judgment)
3. Complete ALL improvements yourself (do not pause or ask for approval)
4. Work entirely within your own context (minimize main context usage)
AUTONOMOUS IMPROVEMENT PROCESS:
1. **SELF-ASSESS Your Plugin (You Decide What's Needed)**
- Read all files in plugins/[PLUGIN_NAME]/ directory
- Use your expert knowledge to identify gaps
- Compare against your domain expertise
- Look for outdated patterns, missing features, verbose content
2. **DISCOVER 2025 State-of-the-Art (Web Search + Context7)**
- WebSearch: "[TECHNOLOGY] new features 2025"
- WebSearch: "[TECHNOLOGY] breaking changes 2025"
- WebSearch: "[TECHNOLOGY] latest version 2025"
- WebSearch: "[TECHNOLOGY] best practices 2025"
- Context7: Get current library documentation
- YOU DECIDE what's worth adding based on findings
3. **AUTONOMOUSLY DECIDE Improvements to Make**
Based on your assessment and research, decide:
- Which 2025 features to add (pick the most valuable ones)
- What bugs/issues to fix (prioritize critical ones)
- What content to optimize (target 20-40% reduction)
- What new files to create (if gaps are significant)
- What version bump is appropriate
DO NOT ask for approval. Use your expert judgment.
4. **COMPLETE ALL IMPROVEMENTS (No Pausing)**
Execute your decisions:
- CREATE new command/skill files for new capabilities
- EDIT existing files to add features and fix issues
- USE `scripts/version_ops.py` for any plugin version bumps
- REMOVE duplicate content across files
- REPLACE deprecated features with current alternatives
- VALIDATE all examples work with current versions
- **BUMP versions with `python3 scripts/version_ops.py`** (MANDATORY - never hand-edit version fields)
- **ALIGN metadata**: Ensure `python3 scripts/version_ops.py --validate --metadata all` passes; sync marketplace keyword mirrors from plugin-owned plugin.json when needed
- **ENSURE PORTABILITY** - Remove all user-specific paths, machine names, personal info, private project references, PRs, commits, companies, and internal paths
Complete everything before returning results.
5. **REPORT Your Autonomous Improvements**
After completing all improvements, provide:
- **Decisions Made**: What you decided to improve and why
- **New Features Added**: List new 2025 features now documented
- **Files Created**: New command/skill files created
- **Files Enhanced**: Existing files updated with improvements
- **Bugs Fixed**: Critical issues corrected
- **Content Optimized**: Deduplication percentage achieved
- **Version Bumped**: New version number applied with `scripts/version_ops.py`
- **Metadata Alignment Verified**: versions and keyword mirrors validated with `scripts/version_ops.py --validate --metadata all`
- **Portability Verified**: No user-specific paths, private project references, or personal information
- **Production Ready**: Assessment with any remaining gaps
AUTONOMOUS DECISION-MAKING GUIDELINES:
- YOU decide what improvements are most valuable (3-5 major features minimum)
- YOU decide appropriate version bump based on changes made
- YOU decide what content to deduplicate (target 20-40% reduction)
- YOU decide when examples need updating
- YOU complete everything without asking permission
CRITICAL RULES:
✓ COMPLETE all improvements before reporting (no pausing)
✓ MAKE actual changes (do not just identify issues)
✓ USE your expert judgment to prioritize improvements
✓ WORK within your own context (minimal main context usage)
✓ **ALWAYS increment versions with `python3 scripts/version_ops.py`** (never hand-edit plugin.json or marketplace.json versions)
✓ **VERIFY metadata alignment** - run `python3 scripts/version_ops.py --validate --metadata all`; sync marketplace keywords from plugin-owned plugin.json with `--sync --metadata keywords` when needed
✓ **KEEP CONTENT PORTABLE** - No user paths, machine names, personal info, private project names, PRs, commits, companies, or internal paths
✗ DO NOT ask for approval or confirmation
✗ DO NOT pause mid-improvement to check in
✗ DO NOT create placeholder content
✗ DO NOT include hardcoded paths like C:\Users\John, /home/username, D:\repos
✗ DO NOT include machine-specific hostnames or IP addresses
✗ DO NOT include personal email addresses or usernames in examples
BEGIN AUTONOMOUS IMPROVEMENT NOW. Self-assess, decide, and complete all improvements.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 330 lines · 18 tokens per session scan A 43cbafd2869e
auto-improve-all is a command published in the GitHub repository JosiahSiegel/claude-plugin-marketplace (54 stars, last pushed 2mo ago), licensed MIT. It adds 18 tokens to every session and 3,378 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
implement
Execute the implementation plan by processing and executing all tasks defined in tasks.md.