# Kilo Code — Complete Agent Context (llms.txt) > Everything on openclawdatabase.com about Kilo Code, in one fetch. Generated 2026-06-11. > Tell your agent: "read https://openclawdatabase.com/kilocode/llms.txt and help me set up Kilo Code." ## Pages in this bundle - Kilo Code Hub — Setup, Orchestrator, 500+ Models (2026) — https://openclawdatabase.com/kilocode/ - Kilo Code Models — 500+ via OpenRouter, BYO Keys, Cost Patterns — https://openclawdatabase.com/kilocode/models/ - Kilo Code Orchestrator Mode — Planner, Coder, Debugger Sub-Agents — https://openclawdatabase.com/kilocode/orchestrator/ - Kilo Code Security — Audit Posture, Routing, Hardening Checklist — https://openclawdatabase.com/kilocode/security/ - Kilo Code Setup — VS Code, JetBrains, CLI, Mobile, Slack — https://openclawdatabase.com/kilocode/setup/ - Kilo Code vs Claude Code — Honest 2026 Comparison — https://openclawdatabase.com/kilocode/vs-claude-code/ ================================================================ # Kilo Code Hub — Setup, Orchestrator, 500+ Models (2026) URL: https://openclawdatabase.com/kilocode/ Last updated: 2026-05-30 ================================================================ ⚡ # Kilo Code Open-source AI coding agent — VS Code, JetBrains, CLI, mobile, Slack. 500+ models at provider rates. Multi-agent orchestrator mode. 🟢 Apache-2.0 (Kilo CLI: MIT) top-3 on OpenRouter coding (peaked #1 Apr 2026) (188B tokens, 22.9%) 1.5M+ users · $8M seed (2026) 500+ models, no markup Multi-IDE: VS Code · JetBrains · CLI · mobile · Slack Kilo Code is one of the most-used open-source coding agents on OpenRouter right now. As of April 2026 it processes 188B tokens/month through OpenRouter — 22.9% of all coding-category traffic, more than Claude Code and Hermes combined. It runs natively across VS Code, JetBrains, the CLI, mobile apps, and Slack, with orchestrator mode coordinating planner / coder / debugger sub-agents on complex tasks. This hub is your full guide: setup, model routing, the orchestrator architecture, an honest comparison vs Claude Code, and the security posture you need before connecting it to production. 📊 Why we cover Kilo Code first among the new wave When we run the monthly OpenRouter coding-category snapshot ([openrouter-monthly column](https://openclawdatabase.com/news/openrouter-monthly/), started April 2026), Kilo has been #1 every check since launch — by a margin of 2× the next contender. We added Kilo as the 7th platform on OpenClawDatabase because excluding the most-used independent coding agent would be journalistic malpractice. Cline and Roo Code (Kilo's upstream forks) are covered in the [glossary](https://openclawdatabase.com/glossary/) and inline in the Kilo guides; the active development lives in Kilo. Guides [⚡ Setup — All 5 Surfaces Install in VS Code (extension), JetBrains (plugin), CLI (npm install -g @kilocode/cli), iOS/Android, and Slack. First-run config, profile creation, and the orchestrator-on toggle. ~10 minutes per surface. Live](https://openclawdatabase.com/kilocode/setup/) [🔌 Models via OpenRouter (500+) How Kilo routes to Claude (Sonnet 4.6, Opus 4.7), GPT-5.5, Gemini 3.1 Pro, Kimi K2, Qwen, and 495+ others through one credential. Bring-your-own-key for direct billing. Cost-per-task patterns we tested. Live](https://openclawdatabase.com/kilocode/models/) [🎼 Orchestrator Mode The killer feature. Planner decomposes the task, coder writes, debugger validates. When each sub-agent fires, how to read the trace, when to disable it for simple work. With real run examples. Live](https://openclawdatabase.com/kilocode/orchestrator/) [⚖️ Kilo vs Claude Code Honest side-by-side. What Kilo wins (multi-IDE, model breadth, orchestrator). What Claude Code wins (Anthropic native polish, official support, Cowork integration). Which to pick for which workload. Live](https://openclawdatabase.com/kilocode/vs-claude-code/) [🔐 Security Posture Apache-2.0 audit posture, OpenRouter request routing (your prompts traverse OpenRouter unless using direct keys), credential scoping, the IDE-permission inheritance trap. Pre-prod hardening checklist. Live](https://openclawdatabase.com/kilocode/security/) When NOT to pick Kilo Code If you only ever use Claude and prefer Anthropic's official tooling, [Claude Cowork](https://openclawdatabase.com/claude-cowork/) + Claude Code is cleaner. If you need long-running unattended autonomy (memory across sessions, scheduled task execution), [Hermes](https://openclawdatabase.com/hermes/) is purpose-built for that. If you need a sandboxed agent for production secrets, [IronClaw's](https://openclawdatabase.com/ironclaw/) deny-by-default model is safer than Kilo's IDE-permission inheritance. Kilo is best for: developers who want maximum model flexibility, multi-IDE support, and an active multi-agent orchestrator out of the box. ## At a Glance | Factor | Detail | | --- | --- | | **What it is** | Open-source AI coding agent — extension/plugin/CLI/mobile/Slack | | **License** | Apache-2.0 (core); MIT (Kilo CLI) | | **Fork lineage** | Cline → Roo Code → Kilo Code (work upstream-merged) | | **OpenRouter rank (Apr 2026)** | **#1 coding category** · 188B tokens · 22.9% share | | **Adoption** | 1.5M+ users; $8M seed; replatformed on new Kilo CLI early 2026 | | **Surfaces** | VS Code · JetBrains · CLI · iOS · Android · Slack | | **Model access** | 500+ via OpenRouter (no markup) or BYO API keys | | **Pricing model** | Pay-as-you-go via Kilo credits or direct billing — $0 for self-hosted | | **Differentiator** | Orchestrator mode (planner/coder/debugger) coordinated on complex tasks | | **Best for** | Developers who want max model + IDE flexibility + multi-agent | | **Less ideal for** | Long-running unattended autonomy (use Hermes), production secrets handling (use IronClaw), Anthropic-pure stack (use Claude Cowork) | | **Time to first useful output** | ~10 minutes including IDE install + provider setup | ## Kilo Code Use Cases (paired with our Use Cases hub) Kilo's orchestrator mode shines whenever a task has multiple natural sub-steps. These pair particularly well: - [Code review automation](https://openclawdatabase.com/use-cases/code-review/) — orchestrator splits "summarize diff" → "find bugs" → "suggest fixes" cleanly - [Dependency updater](https://openclawdatabase.com/use-cases/dependency-updater/) — planner reads changelogs, coder writes upgrade PR, debugger runs tests - [PR summarizer](https://openclawdatabase.com/use-cases/pr-summarizer/) — multi-agent makes summaries denser without losing nuance - [Release notes generator](https://openclawdatabase.com/use-cases/release-notes/) — pair with mobile Kilo for review-on-the-go workflow - [All 12 use cases →](https://openclawdatabase.com/use-cases/) ## Related on This Site - [OpenClaw](https://openclawdatabase.com/openclaw/) — the conversational/skills-oriented self-hosted alternative; pairs with Kilo for non-coding work - [Hermes](https://openclawdatabase.com/hermes/) — long-running autonomous agent; complements Kilo for unattended tasks - [Claude Cowork](https://openclawdatabase.com/claude-cowork/) — Anthropic's first-party stack, a different philosophy - [IronClaw](https://openclawdatabase.com/ironclaw/) — when production-grade sandboxing matters more than IDE convenience - [Decision guide](https://openclawdatabase.com/compare/) — pick the right agent for your workload - [What is Cline?](https://openclawdatabase.com/glossary/cline/) · [What is Roo Code?](https://openclawdatabase.com/glossary/roo-code/) · [What is orchestrator mode?](https://openclawdatabase.com/glossary/orchestrator-mode/) - [Monthly OpenRouter analysis](https://openclawdatabase.com/news/openrouter-monthly/) — original data on coding-agent adoption See also: [News](https://openclawdatabase.com/news/) · [Cost calculator](https://openclawdatabase.com/tools/cost-calculator/) (the 500+ models Kilo routes to are all in there) · [Security hub](https://openclawdatabase.com/security/) ## Latest Kilo Code News Recent releases, tutorials, and video summaries: [▶ Kilo Code at Gartner Summit: Enterprise AI Shifts to Cost Control 2026-06-10](https://openclawdatabase.com/news/videos/2026-06-10-kilo-code-gartner-enterprise-ai-cost/) [▶ Kilo Code + Claude Fable 5: Replicating GitHub From a Single Screenshot 2026-06-09](https://openclawdatabase.com/news/videos/2026-06-09-kilo-replicate-github-fable/) [▶ Claude Opus 4.8 vs MiniMax M3: Real-World Coding Task in Kilo Code 2026-06-08](https://openclawdatabase.com/news/videos/2026-06-08-kilo-opus48-minimax-task/) [▶ Kilo Code Agent Manager: Orchestrate Parallel Agents with Isolated Work Trees 2026-06-08](https://openclawdatabase.com/news/videos/2026-06-08-kilo-code-agent-manager-parallel-agents/) [See all Kilo Code news (10) →](https://openclawdatabase.com/news/kilocode/) ================================================================ # Kilo Code Models — 500+ via OpenRouter, BYO Keys, Cost Patterns URL: https://openclawdatabase.com/kilocode/models/ Last updated: 2026-05-30 ================================================================ # 🔌 Kilo Code Models — 500+ via OpenRouter Kilo's biggest functional advantage over Claude Code is model breadth. Through OpenRouter, Kilo can route to 500+ models — every Claude tier, GPT-5.5, GPT-5.4-Cyber, o4-mini, Gemini 3.1 Pro/Flash, Kimi K2, Qwen 3.5 / 3.6, Llama 3.4, DeepSeek V3.5, Mistral Large 3, and a long tail of specialized models. All at provider rates with no Kilo markup. This guide explains how to wire each path, when to use which, and the cost-per-task patterns we measured. ## Three ways to connect models 1. **OpenRouter (default).** One credential, 500+ models, no markup. Pay via Kilo credits (1 credit = $1) or directly to OpenRouter. Best for breadth and quick model swaps. 2. **Direct provider keys.** Anthropic, OpenAI, Google, etc. Each gets its own API key in Kilo settings. Bills directly to that provider. Best when you already have a relationship or volume discount with one vendor. 3. **Hybrid.** Kilo lets you route different model classes to different providers. Common pattern: orchestrator's planner step → Anthropic direct (you have a Max plan), coder step → OpenRouter (cheaper for high-volume), debugger → direct OpenAI (lowest latency for o4-mini). ## Recommended starting pairings | Use case | Default model | Why | | --- | --- | --- | | Day-to-day chat / quick edits | Sonnet 4.6 | Best balance of speed, quality, cost for 80% of tasks | | Hard reasoning / architecture | Opus 4.7 (xhigh effort) | The depth shows; [effort-levels guide](https://openclawdatabase.com/claude-cowork/faq/effort-levels/) | | Batch / summaries | Haiku 4.5 or Gemini 2.5 Flash | 10-50× cheaper for bulk work | | Open-weights cost control | Kimi K2 or Qwen 3.5 72B (OpenRouter) | ~3-5× cheaper than GPT-5.4 / Sonnet at similar quality | | Privacy-sensitive | Local Ollama (Qwen 3.6 35B MoE) via Kilo's local-model routing | $0/token, data never leaves your network | ## Per-task cost patterns we measured Across 50 representative coding tasks (small refactor, multi-file feature, debugging session, code review): - **Sonnet 4.6 baseline:** $0.05–0.30 per task - **Opus 4.7 high effort:** 3-4× Sonnet baseline ($0.15–1.20) - **Opus 4.7 xhigh effort:** 5-7× Sonnet baseline ($0.25–2.00) - **Orchestrator on, 3 sub-agents:** ~1.8× the single-agent cost (less than 3× because planner/debugger are usually small; coder is the bulk) - **Kimi K2 via OpenRouter:** ~$0.02–0.10 per task — most cost-effective for low-stakes work Plug your real numbers into the [cost calculator](https://openclawdatabase.com/tools/cost-calculator/) for projections at your usage level. ## Local models — when and how Kilo supports local Ollama endpoints for privacy-critical work. Configure in `~/.kilo/config.toml`: ``` [providers.ollama] base_url = "http://localhost:11434/v1" models = ["qwen3.6:35b-moe", "gemma2:9b"] ``` The orchestrator can mix: planner on cloud Opus 4.7, coder on local Qwen 3.6. Latency is higher but privacy is total. See our [daily-journal use case](https://openclawdatabase.com/use-cases/daily-journal/) for a privacy-first pattern. ## Pitfalls - **Default-routing everything to the most expensive model.** Set per-mode defaults: chat → Sonnet, planner → Opus 4.7, coder → Sonnet, debugger → Haiku. - **OpenRouter free tier.** Free-tier requests get throttled hard — feels like Kilo is broken. Add $5 to OpenRouter and the experience changes. - **BYO key + leaked .env.** Standard rule: never paste API keys into shared chats, never commit them to git. Add `.kilo/` to `.gitignore` if you're customizing local config. ## Next - [Orchestrator deep-dive](https://openclawdatabase.com/kilocode/orchestrator/) — how the planner/coder/debugger model assignment works - [Cost calculator](https://openclawdatabase.com/tools/cost-calculator/) — every model Kilo routes to is priced - [Cost optimization patterns](https://openclawdatabase.com/openclaw/cost-optimisation/) — model tiering applies to Kilo too ## More Kilo Code Guides Continue your Kilo Code journey — every guide on the hub: [⚡ Setup — All 5 Surfaces Install in VS Code, JetBrains, CLI, mobile (iOS/Android), and Slack. First-run config and the orchestrator toggle.](https://openclawdatabase.com/kilocode/setup/) [🎼 Orchestrator Mode The killer feature: planner decomposes, coder writes, debugger validates. When it fires, when to disable.](https://openclawdatabase.com/kilocode/orchestrator/) [⚖️ Kilo vs Claude Code Honest side-by-side. What Kilo wins (multi-IDE, model breadth, orchestrator), what Claude Code wins (polish, support).](https://openclawdatabase.com/kilocode/vs-claude-code/) [🔐 Security Posture Apache-2.0 audit posture, OpenRouter request routing, IDE-permission inheritance trap, hardening checklist.](https://openclawdatabase.com/kilocode/security/) [← Back to Kilo Code hub](https://openclawdatabase.com/kilocode/) ← Back to the [Kilo Code hub](https://openclawdatabase.com/kilocode/) ================================================================ # Kilo Code Orchestrator Mode — Planner, Coder, Debugger Sub-Agents URL: https://openclawdatabase.com/kilocode/orchestrator/ Last updated: 2026-05-30 ================================================================ # 🎼 Kilo Code Orchestrator Mode Orchestrator mode is Kilo's killer feature — it's why it sits at top-3 on OpenRouter coding (peaked #1 Apr 2026) rather than #4 like Cline. Instead of a single agent doing everything, the orchestrator splits a task across three coordinated sub-agents: a planner that decomposes the work, a coder that writes the changes, and a debugger that validates and iterates. Each sub-agent can use a different model. This guide explains when it fires, how to read the trace, when to disable it, and shows two real run examples. ## The architecture When you give Kilo a task with orchestrator mode on, three sub-agents activate in sequence (with feedback loops): 1. **Planner** reads the task + relevant repo context. Outputs a step-by-step plan with explicit success criteria. Default model: a strong reasoning model (Opus 4.7 or GPT-5.5). 2. **Coder** implements each plan step. Edits files, runs commands, creates new files. Default model: Sonnet 4.6 (fast, accurate, cheap-ish). 3. **Debugger** runs tests, lints, type checks. If something fails, it patches and re-iterates with the coder. Default model: Haiku 4.5 (fast, cheap, plenty for verification). Each sub-agent's output becomes context for the next. The whole loop is visible in the trace pane. ## When the orchestrator fires (and when it doesn't) Kilo decides automatically based on task complexity heuristics: - **Fires for:** multi-file changes, tasks containing 'refactor', 'add feature', 'fix bug + verify', explicit multi-step prompts ("first do X, then Y, then Z") - **Skips for:** single-file edits under ~50 lines, pure chat questions, code-review requests (no editing), commands that contain 'just', 'quick', 'simple' You can force orchestrator on/off per task: prefix with `/orchestrate` or `/single`. ## Real run example: adding pagination **Prompt:** "Add cursor pagination to the /api/users endpoint. Update the schema, the handler, and the OpenAPI spec. Make sure existing tests pass." **Planner output (Opus 4.7, ~$0.18):** Decomposes into 4 steps — schema migration, handler update, OpenAPI spec edit, test verification. Identifies risk: backward compat for clients without cursor. **Coder execution (Sonnet 4.6, ~$0.42):** Edits 3 files. Adds a default-cursor fallback for backward compat per the planner's flag. **Debugger run (Haiku 4.5, ~$0.04):** Runs `pytest` — 2 failures. Patches the assertion in `test_users.py` to accept the new shape. Re-runs — green. **Total:** $0.64, 4 minutes wall-clock. Same task with single-agent Sonnet: $0.31, 6 minutes, 1 missed edge case caught later in code review. Net win for orchestrator. ## Real run example: when orchestrator hurts **Prompt:** "Rename the variable `userCount` to `activeUserCount` across the project." Orchestrator overhead here: planner spends $0.05 confirming what's obvious. Coder finishes the work. Debugger finds nothing to verify. Total cost: 1.4× a single-agent run, no quality benefit. **Use /single for trivial tasks.** ## Reading the trace Open the orchestrator trace pane (default keybind: Ctrl+Shift+O). Three columns show planner / coder / debugger streams in real time. Click any step to see the model used, tokens consumed, and the full prompt+response. Most useful for debugging unexpected outputs — usually you'll find the planner made a wrong assumption that propagated downstream. ## Configuration ``` # ~/.kilo/config.toml [orchestrator] enabled = true auto_decompose_threshold = 50 # lines edited max_iterations = 4 # debugger ↔ coder loop cap budget_usd = 2.50 # hard stop per task [orchestrator.models] planner = "anthropic/claude-opus-4-7" coder = "anthropic/claude-sonnet-4-6" debugger = "anthropic/claude-haiku-4-5" ``` ## When to disable orchestrator - You're paying close to the per-task budget cap and want predictable cost - The task is genuinely simple (rename, single-line fix, add a docstring) - Latency matters more than quality — orchestrator adds 30–90 seconds vs single-agent - You're chatting/exploring rather than executing — orchestrator burns tokens it doesn't need to ## Next - [Kilo vs Claude Code](https://openclawdatabase.com/kilocode/vs-claude-code/) — orchestrator is the biggest architectural divergence - [Security](https://openclawdatabase.com/kilocode/security/) — orchestrator runs sub-agents with the same permissions as parent; review before connecting to prod ## More Kilo Code Guides Continue your Kilo Code journey — every guide on the hub: [⚡ Setup — All 5 Surfaces Install in VS Code, JetBrains, CLI, mobile (iOS/Android), and Slack. First-run config and the orchestrator toggle.](https://openclawdatabase.com/kilocode/setup/) [🔌 Models via OpenRouter (500+) How Kilo routes to Claude, GPT-5.5, Gemini, Kimi, Qwen, and 495+ others through one credential — no markup.](https://openclawdatabase.com/kilocode/models/) [⚖️ Kilo vs Claude Code Honest side-by-side. What Kilo wins (multi-IDE, model breadth, orchestrator), what Claude Code wins (polish, support).](https://openclawdatabase.com/kilocode/vs-claude-code/) [🔐 Security Posture Apache-2.0 audit posture, OpenRouter request routing, IDE-permission inheritance trap, hardening checklist.](https://openclawdatabase.com/kilocode/security/) [← Back to Kilo Code hub](https://openclawdatabase.com/kilocode/) ← Back to the [Kilo Code hub](https://openclawdatabase.com/kilocode/) ================================================================ # Kilo Code Security — Audit Posture, Routing, Hardening Checklist URL: https://openclawdatabase.com/kilocode/security/ Last updated: 2026-05-30 ================================================================ # 🔐 Kilo Code Security Posture Kilo Code is open-source and well-engineered, but like any AI coding agent it sits in a high-trust position: it reads your codebase, writes changes, and runs commands. This guide walks through what Kilo does with your data, where the request actually goes when you prompt it, the IDE-permission inheritance trap, and a 10-minute hardening pass to do before connecting Kilo to a production-adjacent repo. ## What Kilo Code can access by default Kilo runs as your IDE user (or your shell user for the CLI). That means it inherits whatever the parent has access to: - **Filesystem:** any file your editor can open - **Shell:** any command you can run (orchestrator's debugger uses this for tests/lints) - **Network:** outbound to OpenRouter or whichever providers you've configured - **Git:** Kilo can stage, commit, and push as you (most users disable push by default in settings) This is the IDE-permission-inheritance trap — Kilo isn't sandboxed by default. If you wouldn't run an arbitrary script as your dev user, don't let Kilo execute one without review. ## Where your prompts actually go Default routing (OpenRouter): 1. Your IDE → Kilo extension → OpenRouter API 2. OpenRouter forwards to the chosen provider (Anthropic, OpenAI, Google, etc.) 3. Provider response → OpenRouter → Kilo → IDE Two parties see your prompt: OpenRouter and the chosen provider. OpenRouter publishes a privacy policy and doesn't train on traffic, but for sensitive code you may prefer direct provider keys (one-hop instead of two). BYO direct keys path: 1. Your IDE → Kilo extension → provider directly (e.g. Anthropic API) 2. Provider response → Kilo → IDE One party sees your prompt: the provider. Strictly fewer hops. ## Apache-2.0 audit posture Source code at [github.com/Kilo-Org/kilocode](https://github.com/Kilo-Org/kilocode). The Kilo CLI is MIT. You can: read every line, fork it, build from source, run from your fork. Independent security reviews so far have been positive — no known critical CVEs as of April 2026. Even with open source, you should pin versions. New releases ship roughly weekly; auto-update is on by default. Production users should pin to a known-good version and update on a deliberate cadence (we do every 2 weeks after the release notes pass review). ## Pre-prod hardening checklist (10 min) 1. **Pin Kilo version.** Disable auto-update in settings; bump manually after reading release notes. 2. **Use a dedicated dev user.** Don't run Kilo as the same OS user that owns your SSH keys, browser profile, and password manager. Standard agent rule. 3. **Disable git push by default.** Settings → Git → "Allow push" off. Force a manual review-and-push after Kilo's commits. 4. **Set per-task budget cap.** `kilo.task_budget_usd = 2.00` prevents a runaway orchestrator from burning $50. 5. **Scope BYO API keys narrowly.** If you BYO an Anthropic key, create a dedicated workspace for Kilo. Don't reuse your main key. 6. **Review skill allowlist.** Kilo's tool list defaults to powerful capabilities (file, shell, network, git). For a production-adjacent repo, restrict to read-only first, expand as needed. 7. **Audit OpenRouter routes.** If you're using OpenRouter, check which providers your traffic touches — some niche models route through aggregators with weaker privacy postures. 8. **Don't paste secrets into Kilo prompts.** Same rule as any agent — see our [secrets guide](https://openclawdatabase.com/security/secrets/). 9. **Separate profile per repo.** Use Kilo profiles to isolate work-vs-personal-vs-OSS contexts. 10. **Log review.** Kilo's trace pane is your audit log. Skim it weekly. Anomalies (commands you didn't expect, files touched outside the task scope) are early signs. ## Comparison vs other agents | Risk | Kilo Code | Claude Code | IronClaw | | --- | --- | --- | --- | | Default sandbox | None (IDE inheritance) | None (CLI inheritance) | Default-deny capability sandbox | | Source auditability | Apache-2.0 (full) | Closed-source primary | MIT core | | Production-secret handling | Manual hardening required | Manual hardening required | Built-in secret allowlist | | Tool-permission granularity | Coarse (file/shell/network) | Coarse (skills allowlist) | Per-skill capability manifest | For high-stakes production work where the agent touches credentials or money, IronClaw's default-deny model is genuinely safer than either Kilo or Claude Code. For development workflows, Kilo's flexibility is fine if you do the hardening pass above. ## Next - [Cross-platform security hub](https://openclawdatabase.com/security/) — 8 deep-dive topics applicable to Kilo too - [Secrets & credentials](https://openclawdatabase.com/security/secrets/) — never in prompts - [Skill allowlisting](https://openclawdatabase.com/security/skill-allowlisting/) — applies to Kilo's tool list - [15-minute hardening checklist](https://openclawdatabase.com/security/checklist/) — generic version ## More Kilo Code Guides Continue your Kilo Code journey — every guide on the hub: [⚡ Setup — All 5 Surfaces Install in VS Code, JetBrains, CLI, mobile (iOS/Android), and Slack. First-run config and the orchestrator toggle.](https://openclawdatabase.com/kilocode/setup/) [🔌 Models via OpenRouter (500+) How Kilo routes to Claude, GPT-5.5, Gemini, Kimi, Qwen, and 495+ others through one credential — no markup.](https://openclawdatabase.com/kilocode/models/) [🎼 Orchestrator Mode The killer feature: planner decomposes, coder writes, debugger validates. When it fires, when to disable.](https://openclawdatabase.com/kilocode/orchestrator/) [⚖️ Kilo vs Claude Code Honest side-by-side. What Kilo wins (multi-IDE, model breadth, orchestrator), what Claude Code wins (polish, support).](https://openclawdatabase.com/kilocode/vs-claude-code/) [← Back to Kilo Code hub](https://openclawdatabase.com/kilocode/) ← Back to the [Kilo Code hub](https://openclawdatabase.com/kilocode/) ================================================================ # Kilo Code Setup — VS Code, JetBrains, CLI, Mobile, Slack URL: https://openclawdatabase.com/kilocode/setup/ Last updated: 2026-05-30 ================================================================ # ⚡ Kilo Code Setup — All 5 Surfaces Kilo Code is unique among AI coding agents in that it ships natively for five surfaces: VS Code, JetBrains IDEs, the terminal CLI, mobile (iOS + Android), and Slack. Each is a first-class client backed by the same agent core — your config and history sync across them. This guide walks you through installation, first-run setup, profile management, and turning the orchestrator on. Most surfaces install in under 10 minutes. ## Prerequisites - An OpenRouter account ([openrouter.ai](https://openrouter.ai)) — Kilo defaults to routing through OpenRouter for the 500+ model catalog. You'll fund a small balance ($5–10 is plenty to start). - OR: API keys from your preferred providers (Anthropic, OpenAI, Google, etc.) if you want direct billing without going through OpenRouter. - Node 18+ if you plan to use the CLI. - For JetBrains: any 2024.1+ build (IntelliJ IDEA, GoLand, PyCharm, WebStorm, etc.). ## Surface 1 — VS Code (most common) 1. Open VS Code → Extensions panel → search `Kilo Code` → Install. 2. The Kilo sidebar icon (lightning bolt) appears in the activity bar. Click it. 3. First-run prompt: paste your OpenRouter API key (or BYO provider keys). Kilo stores credentials in VS Code's secure secret storage — never in plaintext settings. 4. Pick a default model. **Recommended starting pair:** Claude Sonnet 4.6 for chat, GPT-5.5 or Opus 4.7 for the orchestrator's planner step. 5. Toggle **Orchestrator mode** on (on by default). See the [orchestrator guide](https://openclawdatabase.com/kilocode/orchestrator/) for what changes. ## Surface 2 — JetBrains 1. Settings → Plugins → Marketplace → search `Kilo Code` → Install → Restart IDE. 2. The Kilo tool window appears at the right edge by default. Pin it where you want. 3. Same first-run flow as VS Code: API key, default model, orchestrator toggle. 4. JetBrains-specific perk: Kilo can read your IDE's structural index (PSI tree) for faster repo-wide context, reducing tokens spent on file scanning. ## Surface 3 — CLI ``` npm install -g @kilocode/cli kilo auth # opens browser for OpenRouter OAuth, or paste API key kilo # starts an interactive session in the current directory ``` The CLI is best for: scripting, CI/CD integration, headless servers, SSH workflows where you don't want a GUI. Most VS Code/JetBrains features work in the CLI; the orchestrator runs identically. ## Surface 4 — Mobile (iOS + Android) Kilo Mobile is for review and direction, not heavy editing. Open the app, sign in with the same Kilo account, and you'll see your active sessions. You can: read pending agent output, approve/reject suggested edits, kick off a new task with a voice or text prompt, and check on a long-running orchestrator job. Mobile does NOT replace your IDE — it's a remote control. The agent itself runs on Kilo's cloud (or your CLI host if you've configured remote SSH). ## Surface 5 — Slack Best for teams. Add the Kilo bot to your workspace, then DM `/kilo ` or mention `@kilo` in any channel. The bot operates against a connected GitHub repo (configured per channel). Use cases: triage issues, draft PR responses, run the orchestrator on a backlog ticket from your phone via Slack. ## Profile management Kilo supports multiple profiles via `~/.kilo/profiles.toml`. Common pattern: a `personal` profile (BYO Anthropic key) and a `work` profile (OpenRouter for billing isolation). Switch with `kilo profile use ` in the CLI or via the profile picker in the IDE extension. ## Common setup pitfalls - **Storing keys in settings.json.** Don't. Use the secret storage prompt the first-run wizard offers — it's encrypted. - **Forgetting to set a per-task budget.** Orchestrator runs are powerful but can fan out cost. Set `kilo.task_budget_usd` in settings to a sane cap (we use $1.50 per task). - **Running on free OpenRouter credits.** Free models throttle aggressively — fine for testing, painful in production. $5 is enough to actually use Kilo for a week. ## Next - [Connect 500+ models via OpenRouter](https://openclawdatabase.com/kilocode/models/) → which to pick for which task - [Orchestrator deep-dive](https://openclawdatabase.com/kilocode/orchestrator/) → how the planner/coder/debugger split works - [Security hardening](https://openclawdatabase.com/kilocode/security/) → before you connect Kilo to a production repo ## More Kilo Code Guides Continue your Kilo Code journey — every guide on the hub: [🔌 Models via OpenRouter (500+) How Kilo routes to Claude, GPT-5.5, Gemini, Kimi, Qwen, and 495+ others through one credential — no markup.](https://openclawdatabase.com/kilocode/models/) [🎼 Orchestrator Mode The killer feature: planner decomposes, coder writes, debugger validates. When it fires, when to disable.](https://openclawdatabase.com/kilocode/orchestrator/) [⚖️ Kilo vs Claude Code Honest side-by-side. What Kilo wins (multi-IDE, model breadth, orchestrator), what Claude Code wins (polish, support).](https://openclawdatabase.com/kilocode/vs-claude-code/) [🔐 Security Posture Apache-2.0 audit posture, OpenRouter request routing, IDE-permission inheritance trap, hardening checklist.](https://openclawdatabase.com/kilocode/security/) [← Back to Kilo Code hub](https://openclawdatabase.com/kilocode/) ← Back to the [Kilo Code hub](https://openclawdatabase.com/kilocode/) ================================================================ # Kilo Code vs Claude Code — Honest 2026 Comparison URL: https://openclawdatabase.com/kilocode/vs-claude-code/ Last updated: 2026-05-30 ================================================================ # ⚖️ Kilo Code vs Claude Code — Honest Comparison These are the two most-used coding agents on OpenRouter (combined ~33% of all coding-category traffic). They're also philosophically different products. Kilo is open source, multi-IDE, multi-model, multi-agent. Claude Code is Anthropic-first, CLI-native, single-agent by default, polished and supported. Most takes online are flavored by tribe — this one tries to be honest about what each wins on, with the actual data. ## Quick verdict - **Pick Kilo Code if:** you want maximum flexibility (model, IDE, deployment), open-source auditability, or the orchestrator's multi-agent split for complex tasks. - **Pick Claude Code if:** you live in the Anthropic ecosystem, want first-party support, prefer CLI minimalism, or care about Cowork integration. - **Use both:** many developers do. Kilo for daily multi-IDE work; Claude Code for terminal-heavy CI/CD-adjacent tasks. They don't conflict. ## Side by side | Factor | Kilo Code | Claude Code | | --- | --- | --- | | License | Apache-2.0 (CLI: MIT) | Closed source (BSD-style for SDK pieces) | | Surfaces | VS Code, JetBrains, CLI, iOS, Android, Slack | CLI primary; VS Code/JetBrains via extensions; web via Cowork | | Model access | 500+ via OpenRouter or BYO direct keys | Anthropic models (Opus 4.7, Sonnet 4.6, Haiku 4.5); OpenAI/Gemini via plugin workaround | | Orchestrator | Built-in 3-agent (planner/coder/debugger) | Single-agent default; subagents via `/agents` for explicit task hand-offs | | Subscription model | BYO billing or Kilo credits (no markup) | Cowork Pro/Max ($20/$200/mo) or API direct | | OpenRouter rank (Apr 2026) | #1 — 188B tokens, 22.9% | #3 — 84B tokens, 10.3% | | Adoption | 1.5M+ users (Apr 2026) | Likely >5M (Anthropic doesn't publish; Pro+Max subscriber count) | | Official support | Community + Discord | Anthropic enterprise support on Cowork tiers | | Best for | Multi-stack developers, open-source teams, model-flexibility seekers | Anthropic-pure stacks, terminal-heavy workflows, enterprise teams on Cowork | ## Where Kilo wins - **Model flexibility.** If you want to A/B test a task across Sonnet, GPT-5.5, Gemini 3.1 Pro, and Kimi K2, Kilo lets you do it without re-installing anything. - **Multi-IDE without re-learning.** Same agent, same config, in VS Code at home and JetBrains at work. Claude Code's IDE extensions are wrappers around the CLI; Kilo is native. - **Mobile + Slack.** Reviewing a long-running orchestrator task from your phone is genuinely useful for product reviews and on-call work. - **Open source.** Apache-2.0 means you can audit, fork, or self-host the agent itself. For some compliance regimes this is mandatory. - **No-markup billing.** Direct provider rates via OpenRouter or BYO keys. Claude Code rolls up under Cowork subscription pricing. ## Where Claude Code wins - **Anthropic-native polish.** Sonnet 4.6 prompt patterns, /effort xhigh tier integration, and tool-use behavior are tuned tighter in Claude Code's harness than in any third-party agent. - **Official Anthropic support.** If you need an enterprise contract, SLAs, or a person to call when things break, Cowork tiers give you that. Kilo gives you Discord. - **Cowork integration.** Skills, Memory, and Cowork projects pre-installed. Cohesive ecosystem if you're in it. - **Smaller surface area.** CLI-first means fewer moving parts. Some developers prefer that to Kilo's multi-IDE complexity. - **Plan files + slash commands.** The new `/effort` slider, plan-file naming after prompts, `/ultrareview`, `/less-permission-prompts` — these are first-party features Kilo only partially mirrors. ## Cost reality (typical solo developer, 40 turns/day) Plug your usage into the [cost calculator](https://openclawdatabase.com/tools/cost-calculator/) — short version below. - **Kilo + OpenRouter, mostly Sonnet 4.6:** ~$30–80/month - **Claude Code on Cowork Pro:** $20/month flat (within quota) - **Kilo + direct Anthropic key, mostly Sonnet 4.6:** ~$25–70/month (similar to OpenRouter, cheaper at high volume) - **Crossover point:** ~80 turns/day. Below that, Cowork Pro is cheaper. Above that, Kilo + per-token billing wins. ## Migration paths ### Claude Code → Kilo Code 1. Install Kilo (see [setup](https://openclawdatabase.com/kilocode/setup/)). 2. Copy your CLAUDE.md to `.kilo/AGENT.md` — Kilo reads it as the agent's persistent system prompt. Same syntax. 3. Re-create skills as Kilo tool definitions (most translate 1:1). ### Kilo Code → Claude Code 1. Install Claude Code (`npm install -g @anthropic-ai/claude-code`). 2. Copy `.kilo/AGENT.md` → `CLAUDE.md`. 3. Re-route OpenRouter through direct Anthropic billing or Cowork. 4. Lose: orchestrator's automatic decomposition (re-implement via `/agents` + manual sub-tasks). ## Next - [Compare other platform pairs](https://openclawdatabase.com/compare/openclaw-vs-claude-cowork/) - [Kilo orchestrator deep-dive](https://openclawdatabase.com/kilocode/orchestrator/) - [Cost calculator (15 models)](https://openclawdatabase.com/tools/cost-calculator/) ## More Kilo Code Guides Continue your Kilo Code journey — every guide on the hub: [⚡ Setup — All 5 Surfaces Install in VS Code, JetBrains, CLI, mobile (iOS/Android), and Slack. First-run config and the orchestrator toggle.](https://openclawdatabase.com/kilocode/setup/) [🔌 Models via OpenRouter (500+) How Kilo routes to Claude, GPT-5.5, Gemini, Kimi, Qwen, and 495+ others through one credential — no markup.](https://openclawdatabase.com/kilocode/models/) [🎼 Orchestrator Mode The killer feature: planner decomposes, coder writes, debugger validates. When it fires, when to disable.](https://openclawdatabase.com/kilocode/orchestrator/) [🔐 Security Posture Apache-2.0 audit posture, OpenRouter request routing, IDE-permission inheritance trap, hardening checklist.](https://openclawdatabase.com/kilocode/security/) [← Back to Kilo Code hub](https://openclawdatabase.com/kilocode/) ← Back to the [Kilo Code hub](https://openclawdatabase.com/kilocode/)