aictx

AI coding agents are getting better at writing code, but they still have a continuity problem.

Every new session often starts like a rookie joining the repo for the first time: reading broad docs, rediscovering structure, missing previous decisions, repeating failed attempts, and asking for context that already existed.

AICTX solves this by giving coding agents repo-local operational continuity.

It is not another coding agent.
It is the continuity layer for the agents you already use.

AICTX stores the useful execution state of a project inside the repo itself: active Work State, handoffs, decisions, failures, validation evidence, next actions, Task Context Packs and Continuity Quality.

What makes it different from generic memory tools:

• Repo-local: continuity lives with the project, not in a remote black box
• Inspectable: state is stored as files you can open, review, diff, delete or selectively commit
• Portable: teams can share a safe subset of continuity through Git
• Cross-agent: Codex, Claude Code, GitHub Copilot and other agents can read and contribute to the same shared continuity state
•Visual: Continuity View generates Mermaid diagrams to show work progress, decisions, handoffs and execution flow
• Execution-focused: it remembers what happened in the repo, not just what was said in chat
• MCP + CLI: agents can use local MCP tools or fall back to simple commands
• Easy to adopt: install it, initialize the repo, and keep using your coding agent normally

AICTX is not another heavy memory system.

It is continuity for the coding agents you already use.

Stop onboarding your coding agents like rookies every session.