AssistFlow

An Assisted Engineering framework: the developer remains the architect and reviewer, while AI handles implementation under supervision.

The developer sets the task, approves the plan, and controls every diff.

WARNING: Do not use these files as-is. Adapt them to your project and working style.

The Assisted Engineering Paradigm

Assisted Engineering is a development model where the human retains full authority over architecture, design decisions, and final acceptance. AI acts as an executor — it writes code, runs checks, and proposes changes, but every meaningful action requires explicit developer approval.

Division of responsibility:

Developer (Architect + Reviewer)
- Defines goals and constraints
- Reviews and approves plans before implementation
- Inspects every diff before commit
- Makes architectural decisions
- Accepts or rejects results
AI Agent (Executor)
- Analyzes the codebase and gathers context
- Proposes implementation plans
- Writes code following approved plan
- Runs tests and verification
- Prepares atomic commits for review

Control flow:

Developer: sets task
    → AI: proposes plan
        → Developer: reviews plan, approves or adjusts
            → AI: implements step by step
                → Developer: reviews each diff
                    → AI: commits approved changes

Installation

Install AssistFlow skills and agents into your IDE config directories (~/.cursor/, ~/.claude/, ~/.config/opencode/).

One-liner (requires Deno):

deno run --reload -A https://raw.githubusercontent.com/korchasa/flow/main/scripts/install.ts

Update to latest version:

deno run --reload -A https://raw.githubusercontent.com/korchasa/flow/main/scripts/install.ts --update

What happens:

The installer clones the repository into ~/.assistflow/ (shallow, ~1MB)
Detects which IDEs you have installed (Cursor, Claude Code, OpenCode)
Shows a plan of symlinks to create and asks for confirmation
Creates per-item symlinks for each skill and agent — your own files are never touched

Re-running is safe (idempotent). Stale symlinks from removed framework items are cleaned up automatically.

From a local clone:

git clone https://github.com/korchasa/flow.git
cd flow
deno task install

Development Setup

For contributors working on AssistFlow itself (not end-user installation):

Prerequisites: Deno, Git

git clone https://github.com/korchasa/flow.git
cd flow
deno task link

deno task link creates symlinks from .dev/ (the single source of truth for dev resources) to IDE-specific directories:

.dev/skills/ -> .cursor/skills/, .claude/skills/, .opencode/skills/
.dev/agents/ -> .cursor/agents/, .claude/agents/, .opencode/agents/
.dev/hooks/, .dev/hooks.json -> .cursor/hooks/, .cursor/hooks.json (Cursor-only)

The command is idempotent — safe to run multiple times. It will not overwrite existing real files (warns and skips instead).

Alternatively, deno task dev runs link automatically on startup.

Verify setup:

deno task check

How It Works

AssistFlow is a set of Skills and Agents — markdown instruction files that AI coding assistants (Cursor, Claude Code, OpenCode, etc.) load into context to follow structured workflows.

Skills (SKILL.md) — step-by-step workflows for specific tasks
Agents (.md) — role definitions with specialized capabilities
Documentation (documents/) — persistent project memory across sessions

AI models lose context between sessions. AssistFlow compensates by storing all decisions, requirements, and architecture in structured docs that the agent reads at the start of every session.

Product vs. Development Tooling

This repository contains two distinct layers. Do not confuse them:

framework/ — the product itself. Skills and agents that users copy into their projects (into .cursor/skills/, .claude/, etc.). This is what AssistFlow distributes.
.cursor/, .claude/ — internal development tooling. Skills and agents used to develop AssistFlow itself (benchmark runner, cursor-agent integration, code generation helpers). These are not part of the product and are not distributed to users.

Developer Workflow

1. Project Setup

Initialize the project structure and documentation:

Run flow-init to analyze the codebase and generate AGENTS.md, SRS, SDS
Configure development commands for your stack

2. Task Cycle

Every task follows the same supervised loop:

Task — describe what needs to be done
Plan (flow-plan) — AI proposes a plan in GODS format. You review, adjust, approve
Execute — AI implements the approved plan. You watch the diffs
Verify — deno task check (or your project's equivalent) must pass. No exceptions
Commit (flow-commit) — AI prepares atomic commits. You review before push

3. Maintenance

flow-maintenance — project health audit
flow-investigate — root cause analysis for complex bugs
flow-answer — codebase Q&A without modifications

Available Skills

Core Workflow (daily use)

flow-init — project initialization
flow-plan — task planning (GODS format)
flow-commit — atomic commits with QA
flow-review — QA + code review of current changes
flow-reflect — self-analysis of recent work
flow-maintenance — project health check
flow-investigate — deep bug investigation
flow-answer — codebase questions

Extending AssistFlow

flow-engineer-command — create/modify a command
flow-engineer-skill — create/modify a skill
flow-engineer-rule — create/modify a rule
flow-engineer-hook — create/modify a hook
flow-engineer-subagent — create/modify a subagent

Setup & Configuration

flow-setup-code-style-ts-deno — Deno/TS code style
flow-setup-code-style-ts-strict — strict TypeScript
flow-skill-configure-deno-commands — configure Deno tasks
flow-skill-ai-skel-ts — scaffold AI agent skeleton

Specialized Skills

flow-skill-draw-mermaid-diagrams — Mermaid diagrams
flow-skill-write-agent-benchmarks — agent benchmarks
flow-skill-write-dep — Development Enhancement Proposals
flow-skill-write-gods-tasks — GODS-format tasks
flow-skill-write-prd — Product Requirements Documents
flow-skill-write-in-informational-style — informational writing style
flow-skill-manage-github-tickets-by-mcp — GitHub issue management
flow-skill-playwright-cli — browser automation
flow-skill-fix-tests — fix failing tests
flow-skill-conduct-qa-session — Q&A sessions
flow-skill-analyze-context — token usage analysis
flow-skill-deep-research — multi-source web research
flow-skill-engineer-prompts-for-instant — prompts for fast models
flow-skill-engineer-prompts-for-reasoning — prompts for reasoning models
flow-skill-deno-cli — Deno CLI operations
flow-skill-deno-deploy — Deno Deploy management

Key Principles

Developer controls, AI executes — no autonomous commits, no unsupervised architectural changes
Explicit workflows — every task type has a defined skill with clear steps
Persistent memory — documentation in documents/ bridges the gap between sessions
Single verification gate — deno task check is the source of truth for project health
IDE-agnostic — skills work across Cursor, Claude Code, OpenCode, and other AI-assisted editors

Project Structure

framework/              # THE PRODUCT — distributed to users
  skills/               #   Skills (SKILL.md per folder)
  agents/               #   Agents (.md files)
documents/              # Project documentation (SRS, SDS, whiteboard)
scripts/                # Deno task scripts
benchmarks/             # Evidence-based agent benchmarks
AGENTS.md               # Project vision, rules, agent instructions

.cursor/                # INTERNAL — AssistFlow development tooling (not distributed)
  skills/               #   Dev-only skills (benchmarks, code generation, etc.)
  agents/               #   Dev-only agents (benchmark-runner, etc.)
.claude/                # INTERNAL — Claude Code settings for this repo

Documentation as Memory

Documentation is not optional — it is the only mechanism that preserves context between AI sessions.

AGENTS.md — project vision, constraints, mandatory rules
requirements.md (SRS) — functional and non-functional requirements
design.md (SDS) — architecture, components, data models
whiteboard.md — current task plan (GODS: Goal, Overview, Done, Solution)

The agent reads these at session start. If the docs are outdated, the agent works with wrong assumptions. Keep them accurate.

License

MIT

Name	af-skill-conduct-qa-session
Description	How to conduct a Q&A session with the user

af-skill-conduct-qa-session

SKILL.md