Secure Coding Agent
Resume and operate coding agents remotely while humans retain approval over consequential actions.
Interactive platform map
Architecture in context
The focused blueprint, its required foundation, and declared recommendations.
Secure Coding Agent
Resume and operate coding agents remotely while humans retain approval over consequential actions.
What this replaces
Autonomous coding agents are valuable, but unrestricted file writes and shell commands create unacceptable operational risk.
What your team gains
A remote coding experience exposed through chat/Telegram with explicit approve-or-deny checkpoints for writes and shell operations.
What is inside the blueprint
How it fits the platform
Open WebUI or Telegram -> coder gateway -> Claude Code session -> approval gate -> filesystem/shell action -> session history/telemetry.
telegram-coder, owui-coder, telegram-gateway, Open WebUI, session-recall-mcp, Neo4j stack
- AICORTEX Core PlatformCoder gateways and approval flows run on Core's container platform.
From prerequisites to operation
- Claude Code/runtime environment
- Telegram bot credentials
- Session persistence
- Defined approval policy
- Deploy coder gateway modes
- Connect OpenWebUI pipe and Telegram bot
- Map sessions
- Define gated operations
- Integrate session recall/graph optionally
- Approval scope
- Allowed commands/actions
- Session mapping
- Bot access lists
- Timeout and retry behavior
- Session cleanup
- Secret management
- Gateway health
- Auditability of approved actions
- Dependency compatibility
What this unlocks with other layers
Controlled Agency
Agents can act across systems while risky writes and commands remain human-gated.