RUNLOCALAIv38
->Will it run?Best GPUCompareTroubleshootStartLearnPulseModelsHardwareToolsBench
Run check
RUNLOCALAI

Independently operated catalog for local-AI hardware and software. Hand-written verdicts. Source-cited claims. Reproducible commands when we have them.

OP·Fredoline Eruo
DIR
  • Models
  • Hardware
  • Tools
  • Benchmarks
TOOLS
  • Will it run?
  • Compare hardware
  • Cost vs cloud
  • Choose my GPU
  • Prompting kits
  • Quick answers
REF
  • All buyer guides
  • Learn local AI
  • Methodology
  • Glossary
  • Errors KB
  • Trust
EDITOR
  • About
  • Author
  • How we make money
  • Editorial policy
  • Contact
LEGAL
  • Privacy
  • Terms
  • Sitemap
MAIL · MONTHLY DIGEST
Get monthly local AI changes
Monthly recap. No spam.
DISCLOSURE

Some links on this site are affiliate links (Amazon Associates and other first-class retailers). When you buy through them, we earn a small commission at no extra cost to you. Affiliate links do not influence our verdicts — there are cards we rate highly that we don't have affiliate relationships with, and cards that sell well that we refuse to recommend. Read more →

© 2026 runlocalai.coIndependently operated
RUNLOCALAI · v38
  1. >
  2. Home
  3. /Maps
  4. /Coding agents (May 2026)
Ecosystem map · Updated May 6, 2026

The coding-agent ecosystem

Six zones covering every coding-agent paradigm. Read /systems/agent-execution-systems first if you need the architectural depth on planning loops, tool dispatch, sandbox isolation, and MCP/memory integration. Pair with /frontier/coding-agents for the momentum view.

By Fredoline Eruo · Reviewed monthly
ℹWhat changed this month
  • OpenClaw foundation governance landed. Founder joined OpenAI; project transitioned to foundation ownership. Stabilized long-term ownership without (yet) slowing the cadence.
  • Roo Code shut down May 15, 2026. Migration target: Cline. Listed in the shutdown tracker zone as a deliberate ecosystem-volatility signal — the coding-agent category remains turbulent.
  • OpenHands v1.6 shipped Planning Mode. Closed the “agent loops without making progress” failure mode that plagued earlier autonomous-agent products.
  • Goose's MCP-first design pulled ahead in extensibility-driven deployments. Block's enterprise governance push made it the right pick for teams that need audit + RBAC + per-team MCP allowlisting.
Zones
  1. Autonomous open-source agents
  2. Surgical-edit CLI
  3. IDE-resident agents
  4. Closed-source flagships
  5. MCP-native infrastructure
  6. Shutdown + migration tracker

Autonomous open-source agents

The category leaders by paradigm. OpenHands has the longest track record + most stable production deployment. OpenClaw is the velocity leader (350k+ stars after April 2026 release; foundation governance). Goose is Block's MCP-first contender, lighter and more extensible.

agentOSS

OpenHands

★ 72k

AI-driven development agent that completes engineering tasks end-to-end — branches, code, PRs. v1.6 added a Planning Mode that drafts a plan before executing. Local-LLM-friendly via Ollama, vLLM, and

orchestratorOSS

OpenClaw

★ 347k

Personal AI agent with a local-first gateway architecture. Connects your local LLMs (Ollama, llama.cpp) to the messaging surfaces you already use — WhatsApp, Telegram, Slack, Discord, iMessage, and 20

agentOSS

Goose

★ 18k

Open-source extensible AI agent now governed by the Agentic AI Foundation (AAIF) at the Linux Foundation. Started inside Block (formerly Square). 25+ provider support including Ollama, Ramalama, Docke

Surgical-edit CLI

Different paradigm — git-integrated CLI editor for tight-control workflows. Aider holds this niche; mature, stable, well-loved by developers who want the agent under their command rather than driving the workflow.

agentOSS

Aider

★ 30k

Terminal-based AI pair programmer. Run in your project directory, describe a change, it edits files and creates meaningful git commits. Works with any LLM — local Ollama, Anthropic, OpenAI, etc.

IDE-resident agents

Live inside VS Code (or compatible IDE). Cline picked up Roo Code's user base after that project shut down in May 2026 — current frontrunner for autonomous-task IDE agents. Continue covers a broader surface (chat + autocomplete + commands) at the cost of less aggressive autonomous-task quality.

agentOSS

Cline

★ 50k

VS Code extension agent — ~4M installs in 2026. Plan/Act mode, autonomous file edits with diff approval, terminal access. The leading open-source IDE agent.

agentOSS

Continue

★ 25k

Open-source VS Code and JetBrains assistant. Configurable autocomplete + chat + agent modes. Strong with local Ollama backends.

Closed-source flagships

The polished commercial offerings. Claude Code is Anthropic's premium agent; Cursor is the IDE-as-agent product. Both deliver capability ceilings the open-source category can't match yet — but at subscription-tier cost and without self-hostability.

agent

Claude Code

★ 30k

Anthropic's terminal-native coding agent. Tops SWE-bench Verified at 87.6% and SWE-bench Pro at 64.3% in 2026. Deep MCP integration, agentic file editing, and a $20/mo Pro tier are the standout signal

ide

Cursor

Anysphere's AI-native IDE. Forks VS Code with Cursor Tab inline completion, agentic chat, and background agents. Best 'flow' for inline completion in 2026.

MCP-native infrastructure

The MCP servers most coding-agent deployments wire in. Filesystem MCP for file access; Git MCP for repo metadata. Pair with any of the agents above to get the canonical /stacks/local-coding-agent capability surface.

serverOSS

MCP Filesystem Server

★ 60k

Anthropic's reference MCP server for filesystem access. Read, write, search, move, and list files inside a configured allowlist of directories. The canonical example for understanding how MCP tool exp

serverOSS

MCP Git Server

★ 60k

Reference MCP server for local Git repository operations. Status, diff, log, blame, branch listing — read-side operations against a checked-out repo without round-tripping to GitHub. Pairs with mcp-se

serverOSS

MCP PostgreSQL Server

★ 60k

Reference MCP server that exposes a Postgres database as a query surface. Read-only by default — but worth flagging that early versions had a SQL-injection class issue where the read-only wrapper coul

Shutdown + migration tracker

The agent ecosystem remains volatile. Roo Code shut down May 15, 2026 — listed here as a deliberate ecosystem-volatility signal. Migration target: Cline. Bet infrastructure on stable upstreams; track the volatile ones.

agentOSS

Roo Code (sunsetting May 15, 2026)

★ 16k

Open-source AI dev-team extension for VS Code (1.55M installs, 23.8k GitHub stars). **Discontinued: all Roo Code products — Extension, Cloud, and Router — shut down on May 15, 2026** with refunds for

Best agent stack by goal

Self-hosted autonomous coding on private codebase: OpenHands + Mem0 + filesystem/git MCP + vLLM + RTX 4090. See /stacks/local-coding-agent.
Latest features + Anthropic-style reasoning: OpenClaw + same MCP/memory layer + Claude Sonnet via API (or local 32B AWQ). The cloud-Claude hybrid pattern.
MCP-heavy enterprise workflows: Goose + per-team MCP allowlist + audit logs. Block's enterprise governance is real — the right pick when MCP extensibility is the primary requirement.
Surgical edits with tight control: Aider + Ollama + git. Each session = one focused commit. The right pick when you want the agent under command rather than driving.
IDE-resident autonomous tasks: Cline + Ollama. The Roo Code migration target; lives inside VS Code; designed for “help me with this task while I keep working in the same window.”
Everyday IDE assistance (broad surface): Continue + Ollama. Chat + autocomplete + commands; less autonomous-task focus than Cline; broader ergonomics.

Going deeper

  • /systems/agent-execution-systems — protocol-engineering depth on the architecture under each of these tools.
  • /frontier/coding-agents — the momentum view: who's exploding, who's cooling.
  • OpenHands operational review, OpenClaw operational review — the operator-grade L1.5 reviews.
  • /stacks/local-coding-agent, /stacks/memory-enabled-agent — canonical deployment recipes.