cornerstones 0.4.5

Agent First market data and risk intelligence platform

Cornerstones

Managed market context for serious operator and agent workflows.

What Cornerstones is

Cornerstones is a product surface for reading market facts, structure, evidence, and decision-ready context through one managed interface.

It is designed for teams and agents that need:

• trustworthy market reads
• no-key selected basic market reads
• issued-key verification for /v1/status
• admin-only /v1/changelog
• authenticated higher-value context surfaces
• explicit degraded or fallback semantics instead of hidden guesswork

Product model

Cornerstones organizes its read surfaces into five layers:

1. Market Truth — quotes, bars, charts, and asset facts
2. Market Structure — higher-value structure and liquidity reads
3. Intelligence Contract — macro, cross-asset, sentiment, and interpretation layers
4. Evidence and Events — browse, export, and delivery-ready evidence/event flows
5. Agent Context — decision-ready context packages for downstream systems and agents

Public-safe discovery flow

Start here when you are evaluating the product shape:

cornerstones guide
cornerstones --help
cornerstones context overview

Then move into one domain at a time through domain help and bounded commands.

Onboarding truth

Current public-safe onboarding is still a managed issued-key flow.

That means:

• use docs and CLI help first
• point the local client at your service URL
• log in with an issued API key
• then run bounded discovery and context commands

This README does not claim that public self-serve packaging, pricing, browser account management, or anonymous full-surface access are already fully launched.

Current product shape

Today the product surface includes:

• market reads across multiple asset classes
• no-key selected basic market reads for first-touch agent/customer evaluation
• charts, evidence, events, and alerts primitives
• decision-ready context packages for downstream agents
• explicit package and tier boundaries on higher-value reads

Stability and truthfulness principles

Cornerstones keeps these rules explicit:

• bounded discovery beats unlimited export claims
• authenticated premium-depth surfaces stay clearly scoped
• responses may return degraded, fallback, not_implemented, or unavailable when that is the truthful answer
• customer-facing docs should stay product-safe and avoid upstream integration details

Internal / operator materials

Detailed implementation notes, runtime integration specifics, and operator-facing history were moved out of the public root docs.

Operator snapshot references preserved locally:

• docs/operator/README_INTERNAL_2026-04-22.md
• docs/operator/CHANGELOG_INTERNAL_2026-04-22.md

Development

For local development and operator workflows, use repo-local runbooks and operator docs instead of treating this root README as an exhaustive implementation manual.

CI guardrails live in .github/workflows/ci.yml and keep current-facing docs aligned with shipped operator surfaces.

Operator auth usage is exposed through:

• CLI: cornerstones admin auth-usage
• API: GET /v1/admin/auth-usage