Continue the conversation — chat opens pre-seeded with the current signal, caps, and movement.
Privacy-first open-source VS Code agent with enterprise governance layer, backed by $32M in Series A and Seed funding (July 2025, led by Emergence Capital and Pace Capital). Differentiator: complete transparency (every file read, command run, token spent visible), BYOK model flexibility, and client-side architecture ensuring code never leaves the local environment.
Cline competes on openness and enterprise control rather than pure capability. It offers complete data sovereignty and model flexibility. The October 2025 Enterprise launch with SSO, RBAC, and audit trails addresses the governance gap that previously limited adoption. Samsung, SAP, Oracle, and Salesforce (Agentforce) have deployed Cline across tens of thousands of developers. The December 2025 LG CNS partnership for "AI-Native Development" (AIND) signals serious enterprise momentum.
Critical watch item: a recurring security cadence — Aug 2025 Mindgard vulns (slow 35+ day response), Feb 2026 Clinejection + npm supply-chain compromise (OpenClaw, ~4,000 downloads), and May 2026 CVE-2026-44211 (Kanban WebSocket hijack). All are now patched and the Feb post-mortem materially hardened the release pipeline (OIDC provenance, third-party CI/CD audits), but the disclosure cadence plus the absence of SOC 2 / ISO 27001 keep advancement gated despite strong enterprise traction (Oracle NetSuite SDA, Samsung, SAP, Salesforce Agentforce).
Adoption & Proof Points
- $32M funding (July 2025): Series A + Seed led by Emergence Capital and Pace Capital, with 1984 Ventures, Essence VC, Cox Exponential, and angels including Jared Friedman, Logan Kilpatrick, Addy Osmani
- Fortune 500 deployments: Samsung (DX division beta, full rollout planned), SAP, Oracle deployed across tens of thousands of developers
- Salesforce: Agentforce Vibes built on Cline's open architecture
- LG CNS: Partnership (Dec 2025) for AIND enterprise agentic solution
- Community scale: 4+ million VS Code installs, 57k+ GitHub stars, 20k Discord members
- Growth: 4,704% GitHub growth (Octoverse 2025)—crowned fastest-growing AI open source project
- Team tier: Free through Q1 2026, then $20/user/month (first 10 seats always free). Enterprise pricing on request.
Recommended Use Cases
- Enterprise teams wanting open-source flexibility with governance controls and $32M-backed product roadmap
- Organizations with existing cloud contracts (Bedrock, Vertex, Azure) seeking BYOK model routing
- Developers prioritizing transparency and approval-based workflows
- Air-gapped and Zero Trust environments requiring local LLM support where cloud-based alternatives fail compliance review
- Teams building custom tooling via Cline CLI, MCP integrations, and Skills compatibility
Risks & Limitations
- SECURITY PATTERN (watch item): three serious events in ~9 months — Aug 2025 Mindgard 4 critical vulns (slow 35+ day response), Feb 2026 Clinejection + npm supply-chain compromise installing OpenClaw (~4,000 downloads), May 2026 CVE-2026-44211 (CVSS 9.3-9.7 Kanban WebSocket hijack). ALL are now patched, and the Feb post-mortem materially improved process (OIDC provenance attestations, removal of GH Actions cache from credential workflows, third-party CI/CD audits, formal VDP in progress). No critical-security-vuln cap applies (nothing currently unpatched), but the disclosure cadence keeps the tool at Tracked and warrants ongoing monitoring; pin package versions and treat the local Kanban/WebSocket surface and MCP endpoints as untrusted.
- No certifications: no SOC 2 / ISO 27001 — primary blocker for regulated enterprise procurement; enterprise features are largely community-driven setups with limited official support.
- Cost unpredictability: BYOK token spend can run $50-200/mo for heavy agentic use; the built-in cost tracker shows real-time spend but there is no spending cap or budget alert.
- No codebase indexing: deliberate design choice (file-by-file traversal + 1M-token context, no RAG/embeddings) — defensible but means no semantic recall / cross-repo awareness that indexed tools like Cursor provide.
- No tab-autocomplete: not a Copilot replacement; you explicitly request code in chat, slower for small edits.
- Setup complexity: MCP servers, multi-provider API keys, and auto-approve rules require meaningful configuration; not beginner-friendly.
- Architectural ambiguity: weaker on high-level, ambiguous tasks (e.g., 'make this more performant') vs. concrete implementation.
- MCP risk: treat MCP endpoints as untrusted, pin versions.
- Not fire-and-forget: human-in-the-loop / per-step approval by default, requires active guidance.
- Benchmark caveat: strong Terminal-Bench numbers are vendor-reported (not third-party adjudicated); Claude Code still leads SWE-bench Verified (~80.8%).
Capabilities & Integration
Agentic depth: Deep Planning mode for codebase investigation, Focus Chain for persistent task tracking, Auto Compact for context summarization, YOLO mode for full autonomy. Plan → Act workflow separates investigation from execution. Background Edits (Dec 2025) allows editing without cursor interruption. Explain Changes provides inline AI explanations for code review.
Context handling: Memory Bank for persistent context across tasks, intelligent context management removing outdated file reads, Focus Chain reinjects todo lists every 6 messages to prevent drift.
Integration surface: VS Code extension (primary), Cline CLI for terminal-first workflows, JetBrains plugin (IntelliJ, PyCharm, WebStorm). MCP Marketplace (Feb 2025) provides extensible tool integrations. Skills compatibility (v3.48.0, Jan 2026) for Claude Code interoperability. Websearch tooling through Cline provider.
Model support: Claude Opus 4.5, Sonnet 4.5 (96.2% diff-edit success per telemetry), Gemini 3 Pro with thinking levels, GPT-5, Kimi K2, Grok 4.1. BYOK any model provider (Anthropic, OpenAI, Google, AWS Bedrock, Azure, Vertex, OpenRouter, local models via Ollama/LM Studio).
