What is agentflow CLAUDE.md?

A project guide for AgentFlow, an application that coordinates AI coding agents into task pipelines. It documents the project’s frontend, backend, database, and agent-server structure.

Use it when developing, testing, linting, building, or extending the AgentFlow codebase.

It gives coding agents the project context they need before changing code, reducing guesses about commands, folders, and system design.

agentflow CLAUDE.md — Instructions file for coding agents by harun-yardimci: A project guide for AgentFlow, an application that coordinates…

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 instructions/harun-yardimci/agentflow/claude-md

Clone the repo

git clone --depth 1 https://github.com/harun-yardimci/agentflow

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/harun-yardimci/agentflow/claude-md.svg)](https://agentmods.dev/instructions/harun-yardimci/agentflow/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/harun-yardimci/agentflow/claude-md"><img src="https://agentmods.dev/badge/instructions/harun-yardimci/agentflow/claude-md.svg" alt="Measured on agentmods" height="20"></a>

Per session 808 This file is loaded in full into every session.

When invoked 808 The same file — it is already loaded in full.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00808	$0.00808
Opus 5	$0.00404	$0.00404
Sonnet 5	$0.00162	$0.00162
Haiku 4.5	$0.00081	$0.00081

Measured 5d ago against content hash 39f23a984d1a, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

agentflow CLAUDE.md 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 5d 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.md · 75 lines

How it starts

The opening of the file, as written. The whole thing — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AgentFlow

AI agent orchestration platform — pipeline-based task runner for Claude, Codex, and Gemini CLI agents.

Quick Reference

npm run dev          # Start frontend + backend (hot reload)
npm test             # Run tests (vitest)
npm run lint         # ESLint
npm run build        # Build frontend (vite)
npm run mcp          # Start MCP server (stdio)

Backend runs on port 3100. Frontend proxies /api/* and /mcp/* to backend via Vite config.

Architecture

Frontend: React 19 + TypeScript strict + Tailwind v4 + Vite 7 (src/)
Backend: Express 5 + better-sqlite3 + Zod validation (server/)
MCP Server: @modelcontextprotocol/sdk, stdio + SSE transports (mcp/)
Database: SQLite in data/agentflow.db (gitignored)
Execution: Worker pool spawns CLI processes (server/engine/)

Project Structure

src/                 React frontend
  components/        UI components (pipelines/, agents/, settings/, shared/)
  context/           React context providers
  lib/               API client, utilities
  types/             TypeScript interfaces
server/              Express backend
  db/                Schema, connection, seed, migrations
  engine/            Task runner, worker pool, worktree manager, event bus
  executor/          CLI executor, templates (claude/codex/gemini), output parser
  routes/            REST API endpoints
  services/          Business logic (pipeline, task, context, breakdown)
  safety/            Output safety rules
  notifications/     Telegram, Slack integrations
mcp/                 MCP server (server.ts + index.ts entry)
bin/                 CLI entry point (cli.js)

Key Design Decisions

Task execution: CLI agents are spawned as child processes. stdin MUST be closed (proc.stdin.end()) even when useStdin=false — Claude CLI hangs on open stdin pipes.
Git isolation: Each task runs in its own git worktree to prevent conflicts. Worktrees are created/cleaned up by server/engine/worktree-manager.ts.
Cascade logic: When a task completes, its dependents are checked. auto approval tasks go straight to running, manual tasks go to awaiting_approval. This is atomic in a SQLite transaction.
Output chaining: Task output is auto-saved to pipeline_ctx table so downstream tasks can reference it.
Interactive mode: Tasks can pause for questions or tool approval via a control protocol (server/executor/control-protocol.ts).
Font: IBM Plex Mono — no geometric unicode characters (they don't render). Use CSS dots, SVG, or HTML entities instead.
Dark theme: surface-0 (#080808) to surface-4, accent-orange (#D97706).

Read the full file on GitHub · 75 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.

5d ago First seen · 75 lines · 808 tokens per session scan A 39f23a984d1a

Subscribe to this mod's changes

agentflow CLAUDE.md is an instructions file published in the GitHub repository harun-yardimci/agentflow (7 stars, last pushed 2d ago), licensed Apache-2.0. It adds 808 tokens to every session, about $0.0040 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.

Building agents

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,182 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

agentflow CLAUDE.md