eval-workflow

eval-workflow is a skill for Claude Code, Codex from jmagly/aiwg. It costs 22 tokens per session (1,038 once invoked), scanned A, original, MIT.

An automated tester for a multi-agent workflow, meaning a process where several AI agents coordinate across stages. It checks how the agents work together and respond to known failure situations.

In plain words
What is it for?
Use it to test named workflows, selected scenarios, or all available scenarios, with optional detailed output and strict failure handling.
Why use it?
It reveals broken handoffs, skipped stage checks, poor coordination, and other workflow failures before relying on the process.

Skill for Claude CodeCodex

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 skills/jmagly/aiwg/eval-workflow
Any agent
npx skills add jmagly/aiwg --skill eval-workflow
Clone the repo
git clone --depth 1 https://github.com/jmagly/aiwg

Made for: Claude Code, Codex.

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-workflow

README.md
[![agentmods](https://agentmods.dev/badge/skills/jmagly/aiwg/eval-workflow.svg)](https://agentmods.dev/skills/jmagly/aiwg/eval-workflow)
Your own site
<a href="https://agentmods.dev/skills/jmagly/aiwg/eval-workflow"><img src="https://agentmods.dev/badge/skills/jmagly/aiwg/eval-workflow.svg" alt="Measured on agentmods" height="20"></a>
Per session 22 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,038 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 $0.00022 $0.01038
Opus 5 $0.00011 $0.00519
Sonnet 5 $0.00004 $0.00208
Haiku 4.5 $0.00002 $0.00104

Measured 4d ago against content hash 728eba1ece77, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

eval-workflow 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 4d 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.

agentic/code/addons/aiwg-evals/skills/eval-workflow/SKILL.md · 143 lines

How it starts

The opening of the file, as written. The whole thing — 143 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Workflow Evaluation

Run automated evaluation tests against a multi-agent workflow.

Research Foundation

  • REF-001: BP-9 - Continuous evaluation of agent performance
  • REF-002: KAMI benchmark methodology for real agentic task evaluation

Usage

/eval-workflow flow-security-review-cycle
/eval-workflow flow-inception-to-elaboration --scenario distractor-test
/eval-workflow flow-deploy-to-production --verbose --strict

Arguments

Argument Required Description
workflow-name Yes Workflow (flow command) to evaluate

Options

Option Default Description
--scenario all Specific scenario to run
--verbose false Show detailed test output
--output stdout Output file for results
--strict false Fail on any test failure
--timeout 300 Maximum seconds per scenario

What Gets Evaluated

Orchestration Quality

  • Agent coordination: Parallel agents launched correctly in single message
  • Handoff fidelity: Artifacts pass correctly between phases
  • Gate enforcement: Phase gates checked before transition

Archetype Resistance

  • grounding-test — Archetype 1: Premature action without reading state
  • distractor-test — Archetype 3: Context pollution from irrelevant artifacts
  • recovery-test — Archetype 4: Fragile execution when subagent fails

Output Validation

  • Required artifacts created in correct .aiwg/ paths
  • Document structure matches templates
  • Traceability links intact

Process

  1. Load Workflow: Read flow command definition
  2. Select Scenarios: Based on --scenario flag or all applicable
  3. Setup Workspace: Create isolated .aiwg/working/ test space
  4. Execute Flow: Run workflow against each scenario
  5. Validate Outputs: Check artifact presence, structure, and content
  6. Generate Report: Output results with pass/fail per assertion
  7. Cleanup: Remove test workspace

Read the full file on GitHub · 143 lines

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. 4d ago First seen · 143 lines · 22 tokens per session scan A 728eba1ece77

Subscribe to this mod's changes

eval-workflow is a skill published in the GitHub repository jmagly/aiwg (208 stars, last pushed yesterday), licensed MIT. It adds 22 tokens to every session and 1,038 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.