|
Use this page for its local concern first. Return to Documentation Index when you need the broader map. |
|
Track A produces proof-grade artifacts such as bundles, manifests, hashes, and verification results. Track B produces hypotheses, exploratory reasoning, and live workspace investigation state. Use the proof path when later humans or CI must trust the output. Use the live path when the work is still moving. |
This is the main conceptual entrypoint for cx.
Start with the fastest proof of value, then learn the deeper model.
The job of this document is mapping, not semantics. For canonical definitions, see Mental Model.
Ultra-Minimal Onboarding
Run this first:
cx mcp
See the agent work on live code.
Learn later:
-
why
Track B = hypothesis generation -
why
Track A = proof generation -
why notes are the durable cognition layer
That order matters. Operators usually trust the model faster after they have watched one live MCP session solve a real problem.
Start Here
If someone is new to cx, do not begin with the full architecture
lecture. Begin with the live surface:
-
Run
cx mcp. -
Let the agent do real work against live code.
-
Then explain why the system is split the way it is.
The progressive onboarding model is:
-
cx mcpshows the agent working on the current checkout. -
cx notesexplains how durable reasoning is preserved. -
cx bundleexplains how that work becomes a promotable artifact.
Only after the operator has seen the system in motion should you introduce the deeper model:
-
Track B = hypothesis generation
-
Track A = proof generation
-
Notes are the durable cognition layer
Choose the mode that matches the job you need to do right now.
Use cx mcp
Need fast interactive AI help on live code? Use cx mcp.
Use it when you need:
-
live workspace reads and search
-
interactive reasoning over the current checkout
-
note discovery and review during active work
-
planning help before you freeze an artifact
This is the right choice when the workspace is still moving and you need the agent to understand what is true right now.
Workflow example: Friday to Monday
Use cx bundle
Need a reproducible, promotable artifact? Use cx bundle.
Use it when you need:
-
a deterministic handoff for CI, review, or downstream automation
-
manifest, lock file, and checksum sidecar protection
-
later validation and verification
-
extraction under manifest and hash guardrails
This is the right choice when the work must stop being exploratory and become an artifact contract.
Workflow example: Friday to Monday
Use cx notes
Need durable design memory? Use cx notes.
Use it when you need:
-
stable architectural intent in version control
-
linked repository knowledge that survives sessions
-
searchable note history and graph review
-
durable context for later humans and agents
This is the right choice when the reasoning itself needs to persist, not just the current task result.
Workflow example: Safe Note Mutation
How The Modes Fit Together
These are not competing tools. They are the three operating surfaces of the same system:
-
Start with
cx mcpwhen you need live understanding. -
Record durable conclusions with
cx notes. -
Freeze trusted handoff state with
cx bundle.
Another way to say the same thing:
-
cx mcpis where hypotheses are generated and tested against live state. -
cx notesis where high-signal reasoning becomes durable repository memory. -
cx bundleis where proof is generated for review, CI, and handoff.
See Mental Model, Friday to Monday, and Safe Note Mutation.