eval-improve

eval-improve is a command for Claude Code from dynamics365ninja/d365fo-mcp-server. It costs 25 tokens per session (212 once invoked), scanned A, original, MIT.

A command for running an evaluation-improvement process on a D365FO agent, an AI assistant for Microsoft Dynamics 365 Finance and Operations. It finds common failure patterns, fixes the highest-priority actionable defect, tests the fix, and prepares a pull request.

In plain words
What is it for?
Use it to analyze evaluation results, reproduce failures as repository tests, improve the agent or its validators, run the test suite, check held-out cases, and prepare a pull request without committing or pushing automatically.
Why use it?
It provides a repeatable way to turn observed agent failures into tested improvements. The process checks the change against both known examples and held-out examples not used during the fix.

Command for Claude Code

Written for Claude Code: allowed-tools in frontmatter. Also seen: mentions subagents.

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 commands/dynamics365ninja/d365fo-mcp-server/eval-improve
Clone the repo
git clone --depth 1 https://github.com/dynamics365ninja/d365fo-mcp-server

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 eval-improve

README.md
[![agentmods](https://agentmods.dev/badge/commands/dynamics365ninja/d365fo-mcp-server/eval-improve.svg)](https://agentmods.dev/commands/dynamics365ninja/d365fo-mcp-server/eval-improve)
Your own site
<a href="https://agentmods.dev/commands/dynamics365ninja/d365fo-mcp-server/eval-improve"><img src="https://agentmods.dev/badge/commands/dynamics365ninja/d365fo-mcp-server/eval-improve.svg" alt="Measured on agentmods" height="20"></a>
Per session 25 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 212 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 $0.00025 $0.00212
Opus 5 $0.00013 $0.00106
Sonnet 5 $0.00005 $0.00042
Haiku 4.5 $0.00003 $0.00021

Measured 6d ago against content hash b3044d2b0629, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

eval-improve 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/commands/eval-improve.md · 18 lines

What it actually says

Launch the eval-improver subagent (Task tool, subagent_type: eval-improver) to run the improver role of the D365FO agent eval loop.

Focus for this run: $ARGUMENTS (If blank, take the top-priority actionable cluster from npm run eval:clusters.)

The subagent already carries the full protocol (survey corpus → confirm rubric class → reproduce as a VM-free repo test → fix tool/knowledge/validator → validate full suite + held-out split → open a PR citing corpus evidence). Do not re-explain it. Relay the subagent's verdict and the resulting diff/PR back to me. Do not commit or push unless I ask.

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 · 18 lines · 25 tokens per session scan A b3044d2b0629

Subscribe to this mod's changes

eval-improve is a command published in the GitHub repository dynamics365ninja/d365fo-mcp-server (138 stars, last pushed yesterday), licensed MIT. It adds 25 tokens to every session and 212 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.