Cortex is a structured cognitive operating layer for high-leverage AI collaboration.
It is not a note system. It is not a knowledge dump. It is not a session archive.
Cortex stores distilled cognitive leverage — reusable structures that improve clarity, reliability, and execution speed across projects.
Cortex contains only high-signal, reusable artifacts:
- Principles
- Patterns
- Contracts
- Specs
- Templates
- Canonical Prompts
- Playbooks
- Policies
- Philosophical Positions
- Operating Model Definitions
Every artifact must:
- Increase structured thinking
- Improve agent reliability
- Be reusable across projects
- Reduce cognitive entropy
If it does not compound reasoning power over time, it does not belong here.
Cortex does not store:
- Raw sessions
- Chat logs
- Telemetry
- Checkpoints
- Project state
- Experimental exhaust
- Brainstorm dumps
Git history is sufficient provenance.
Cortex is a reasoning layer, not an activity log.
Project-instance operational state belongs in a configurable boundary root.
-
Default boundary root is
.cortex/. -
Boundary root may be changed only through policy-governed contract update.
-
Policy:
policies/project_state_boundary_policy_v0.md -
Spec:
specs/project_state_boundary_spec_v0.md -
Contract:
contracts/project_state_boundary_contract_v0.json -
Enforcement gate:
scripts/project_state_boundary_gate_v0.py
Cortex exists to compound structured thinking.
It is optimized for:
- Clarity over accumulation
- Reusability over novelty
- Discipline over volume
- Architecture over tools
Cortex is a long-term cognitive multiplier.
Cortex standardizes on:
uvfor Python environment + command executionjustfor task orchestration
Primary operational entrypoints should be callable through just recipes that shell out via uv.
- Cortex Coach docs:
docs/cortex-coach/README.md - Maintainer gate:
just quality-gate(fallback:./scripts/quality_gate_v0.sh) - CI gate:
just quality-gate-ci(fallback:./scripts/quality_gate_ci_v0.sh)
Artifact Admission Rules: [ARTIFACT_ADMISSION.md]
[1] Reflection-loop pattern inspiration: aviadr1/claude-meta
[2] Comparative research reference and credited project: jayminwest/mulch (report: .cortex/reports/project_state/mulch_comparative_research_report_v0.md)