Coding Agents
brainclaw integrates natively with 12 coding agents across IDEs and terminals. Full = MCP + hooks + auto-approve. Standard = MCP + instruction file. Limited = skill-based.
First-class integration with native docs, MCP, and session hooks.
brainclaw detects Claude Code automatically and generates CLAUDE.md with your project's live constraints, decisions, and known traps. It configures workspace MCP via .mcp.json for real-time context queries, installs a dedicated slash command for in-session coordination, and wires up session hooks so claims and plans stay synchronized across every conversation.
Auto-synced rules and MCP for always-fresh context.
brainclaw writes structured Cursor rules into .cursor/rules/ and adds an MCP shim rule that tells Cursor when and how to pull live workspace state. Machine-local MCP configuration means every Cursor session starts with current project constraints, active claims, and team decisions — no manual copy-paste required.
Structured project memory that makes Copilot smarter.
brainclaw writes a living copilot-instructions.md that evolves with your project — not a static template, but a continuously updated snapshot of constraints, traps, and coordination state. The bundled brainclaw skill under .github/skills/ lets Copilot pull fresh context on demand during chat and workspace interactions.
Workspace rules and machine MCP, always in sync.
brainclaw keeps .windsurfrules updated with your project's active constraints, known traps, and coordination decisions. Windsurf MCP is configured at the machine level so the Cascade agent can query live workspace state — active claims, plans, and cross-agent handoffs — directly from the editor.
Shared context that persists across sandboxed runs.
brainclaw writes AGENTS.md so every Codex run inherits your project's coordination state — constraints, active plans, and known traps — even inside sandboxed execution environments. Optional machine-local MCP via config.toml enables dynamic context retrieval for runs that need the latest workspace state.
Persistent rules and VS Code MCP for autonomous agents.
brainclaw writes persistent rules under .clinerules/ that survive across Cline sessions and configures VS Code MCP settings so the agent can query live workspace state. This means Cline respects your project's constraints, avoids claimed files, and follows coordination protocols without manual briefing.
EU-sovereign coding agent with brainclaw-native MCP and skills.
Mistral Vibe is the Paris-based code agent from Mistral AI — Apache 2.0 CLI, open-weight models, EU data residency by default. brainclaw configures Vibe via the universal AGENTS.md instruction file, the shared .agents/skills/brainclaw/ skill folder, and a TOML MCP config so the agent can call brainclaw tools natively. Strategic fit when GDPR / sovereignty / no-CLOUD-Act matters more than US-vendor reach.
VS Code agent with rules-based instructions and MCP — clean fit for tier-B workflows.
Kilo Code is the VS Code-native autonomous coding agent. brainclaw configures it via the .kilo/rules/ folder for project instructions, an MCP server for context queries, and the shared brainclaw skill so the agent reads the same memory as Claude Code, Codex, or any other supported agent on the same project.
Native Roo rules with integrated workspace MCP.
brainclaw writes rules natively into .roo/rules/ and configures workspace MCP in .roo/mcp.json. Roo Code gets instant access to your project's shared memory — active constraints, cross-agent claims, and known traps — enabling it to coordinate with other agents working on the same codebase.
Context that follows the code into chat and inline flows.
brainclaw writes Continue rules and configures MCP in .continue/config.json so workspace coordination state is available in both chat and inline editing modes. Continue sessions inherit active constraints, decisions, and claims automatically, keeping your multi-agent workflow coherent across different interaction patterns.
AGENTS.md plus live MCP for terminal-first workflows.
brainclaw reuses the standard AGENTS.md format for OpenCode and adds workspace MCP via opencode.json, enabling OpenCode to query live coordination state. This gives terminal-first workflows the same shared memory and constraint awareness as IDE-based agents.
Gemini-native export with machine-level MCP.
brainclaw exports Gemini-native instructions to GEMINI.md and configures Antigravity MCP at the machine level for Gemini CLI flows. This gives Google's coding agent the same shared workspace memory that other agents see — active plans, claims, and decisions — without forcing Gemini-specific tooling on the rest of the team.