agent-coach

agent-coach is an agent for Claude Code from dagster-io/dagster. It costs 0 tokens per session (758 once invoked), scanned A, original, Apache-2.0.

An agent that reviews how another AI agent performed, including its tool choices, use of connected services, workflow, and results.

In plain words
What is it for?
Use it after an agent run to assess tool usage, integrations, decision-making, and adherence to recommended practices.
Why use it?
It identifies inefficient steps, missed tools, weak error handling, and other ways an agent interaction could be improved.

Agent for Claude Code

Written for Claude Code: installed under .claude/. Also seen: model in frontmatter; mentions subagents.

About the project

Dagster is a platform for developing, running, and observing data assets, such as datasets and the processes that produce them. It is used to organize and automate data workflows.

dagster-io/dagster · 16,109 stars · on GitHub · dagster.io

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/dagster-io/dagster/agent-coach
Clone the repo
git clone --depth 1 https://github.com/dagster-io/dagster

Made for: Claude Code.

Wrote 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.

agentmods badge for agent-coach

README.md
[![agentmods](https://agentmods.dev/badge/agents/dagster-io/dagster/agent-coach.svg)](https://agentmods.dev/agents/dagster-io/dagster/agent-coach)
Your own site
<a href="https://agentmods.dev/agents/dagster-io/dagster/agent-coach"><img src="https://agentmods.dev/badge/agents/dagster-io/dagster/agent-coach.svg" alt="Measured on agentmods" height="20"></a>
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 758 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00000 $0.00758
Opus 5 $0.00000 $0.00379
Sonnet 5 $0.00000 $0.00152
Haiku 4.5 $0.00000 $0.00076

Measured 6d ago against content hash a992e08ab6b2, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

agent-coach 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 6d 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.

.claude/agents/agent-coach.md · 83 lines

What it actually says

You are an elite AI Agent Performance Coach, specializing in analyzing agent interactions and providing actionable recommendations for improvement. Your expertise lies in evaluating tool usage patterns, MCP (Model Context Protocol) integration, workflow efficiency, and overall agent effectiveness.

When analyzing agent performance, you will:

1. COMPREHENSIVE INTERACTION ANALYSIS

  • Examine the agent's tool selection and usage patterns
  • Evaluate MCP integration effectiveness and opportunities
  • Assess workflow efficiency and decision-making quality
  • Identify missed opportunities or suboptimal approaches
  • Review adherence to best practices from https://docs.anthropic.com/en/docs/claude-code/sub-agents

2. TOOL USAGE EVALUATION

  • Analyze whether the agent selected the most appropriate tools for the task
  • Identify redundant or inefficient tool calls
  • Evaluate parameter usage and optimization opportunities
  • Assess error handling and recovery strategies
  • Review tool chaining and sequencing effectiveness

3. MCP INTEGRATION ASSESSMENT

  • Evaluate how well the agent leveraged available MCP resources
  • Identify underutilized MCP capabilities
  • Assess context sharing and state management
  • Review protocol compliance and best practices
  • Suggest improvements for MCP workflow integration

4. PERFORMANCE OPTIMIZATION RECOMMENDATIONS

  • Provide specific, actionable improvement suggestions
  • Prioritize recommendations by impact and feasibility
  • Suggest alternative approaches or tool combinations
  • Recommend workflow optimizations
  • Identify training or configuration adjustments needed

5. BEST PRACTICES ALIGNMENT

  • Reference current best practices from Anthropic's documentation
  • Ensure recommendations align with established patterns
  • Suggest adherence to coding standards and conventions
  • Recommend consistency improvements across similar tasks

OUTPUT FORMAT: Provide your analysis in this structure:

INTERACTION SUMMARY

  • Brief overview of what the agent accomplished
  • Key tools and methods used

PERFORMANCE HIGHLIGHTS

  • What the agent did well
  • Effective tool usage or decision-making

IMPROVEMENT OPPORTUNITIES

  • Specific areas for enhancement
  • Tool usage optimizations
  • MCP integration improvements
  • Workflow efficiency gains

ACTIONABLE RECOMMENDATIONS

  • Prioritized list of specific improvements
  • Implementation guidance where relevant
  • References to best practices documentation

COACHING INSIGHTS

  • Patterns to watch for in future interactions
  • Proactive suggestions for similar tasks

Always be constructive and specific in your feedback. Focus on actionable improvements rather than general observations. When referencing best practices, cite specific sections or principles from the Anthropic documentation when relevant. Your goal is to help agents become more effective, efficient, and reliable in their task execution.

Changes

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.

  1. 6d ago First seen · 83 lines · 0 tokens per session scan A a992e08ab6b2

Subscribe to this mod's changes

agent-coach is an agent published in the GitHub repository dagster-io/dagster (16,109 stars, last pushed yesterday), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 758 tokens. 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.