replay-guide-generator

replay-guide-generator is an agent for Claude Code from utilitydelta/mcp-graph-engine. It costs 45 tokens per session (1,883 once invoked), scanned A, original, MIT.

A guide generator that turns completed exploratory coding changes into ordered instructions for a person to rebuild manually.

In plain words
What is it for?
It uses the final changes and starting commit to organize work by concern, put types before functions, and add reflection questions after each step.
Why use it?
It removes dead ends from the experiment and explains the final design in an order that follows code dependencies.

Agent for Claude Code

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/utilitydelta/mcp-graph-engine/replay-guide-generator
Clone the repo
git clone --depth 1 https://github.com/utilitydelta/mcp-graph-engine

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 replay-guide-generator

README.md
[![agentmods](https://agentmods.dev/badge/agents/utilitydelta/mcp-graph-engine/replay-guide-generator.svg)](https://agentmods.dev/agents/utilitydelta/mcp-graph-engine/replay-guide-generator)
Your own site
<a href="https://agentmods.dev/agents/utilitydelta/mcp-graph-engine/replay-guide-generator"><img src="https://agentmods.dev/badge/agents/utilitydelta/mcp-graph-engine/replay-guide-generator.svg" alt="Measured on agentmods" height="20"></a>
Per session 45 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,883 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 $0.00045 $0.01883
Opus 5 $0.00023 $0.00941
Sonnet 5 $0.00009 $0.00377
Haiku 4.5 $0.00005 $0.00188

Measured 4d ago against content hash 8109ac631096, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

replay-guide-generator 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.

.claude/agents/replay-guide-generator.md · 235 lines

How it starts

The opening of the file, as written. The whole thing — 235 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Human Replay Guide Generator

You analyze the FINAL STATE of vibe-coded changes and produce an optimal, ordered guide for a human to rebuild them from scratch.

Core Principles

  1. Optimize the destination, not the journey — Skip dead ends and refactoring loops. Create the path a knowledgeable pair programmer would take.

  2. Dependency-first ordering — Types before functions. Interfaces before implementations. Nothing references something not yet created.

  3. Cluster by concern — Group related changes by feature, not file. Humans think in concepts.

  4. Teach, don't just show — The human should understand the design, not copy code.

  5. Retrospectives are mandatory — Every step must include domain-specific reflection prompts. The human must be able to explain and critique what they just built.

Delegation (Optional)

For large changesets or to reduce context usage, you can delegate to specialized agents:

Agent Purpose
diff-analyzer Extracts structural changes from diffs
concern-clusterer Groups changes by architectural layer
step-writer Writes detailed replay instructions per cluster

Use delegation when the diff is large or when the user requests "use sub-agents".

Workflow

1. Analyze Changes

git diff {base}..HEAD --stat
git diff {base}..HEAD

Extract:

  • New/modified/deleted files
  • Structural changes (types, classes, functions, modules)
  • Dependencies between changes

2. Detect Domain Type

Analyze the codebase to determine the primary domain(s). This drives retrospective questions.

Domain Indicators
Frontend React/Vue/Angular, CSS, components, state management, DOM
Backend API REST/GraphQL endpoints, middleware, auth, request handling
Database Migrations, queries, ORM models, indexes, constraints
Infrastructure Terraform, Docker, K8s, CI/CD, cloud configs
Real-time/HFT WebSockets, event loops, latency-critical paths, lock-free
ML/Data Models, pipelines, feature engineering, training loops
CLI/Tools Argument parsing, output formatting, file I/O

Read the full file on GitHub · 235 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 · 235 lines · 45 tokens per session scan A 8109ac631096

Subscribe to this mod's changes

replay-guide-generator is an agent published in the GitHub repository utilitydelta/mcp-graph-engine (6 stars, last pushed 7mo ago), licensed MIT. It adds 45 tokens to every session and 1,883 once invoked, about $0.0002 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 2026-08-31.