What DigitalStack does

Log, reconstruct, submit, approve, import, and reconcile canonical time through self-scoped APIs, governed team decisions, and confirmation-bound AI tools.

Read canonical Execute work items, transition history, and board projections without exposing legacy tickets or making board coordinates authoritative.

Comment, transition, assign, schedule, estimate, label, link, block, and focus Execute work items through named, confirmation-bound commands separate from Work Item Enhance.

Create proposal-first enhancement runs over canonical work items: requirements, description drafts, design notes, implementation plans, acceptance criteria, test plans, estimates, and risk checks.

Read a deterministic daily-focus Context Package for the caller: portfolio attention, related work, cited evidence, coverage, and provenance without turning recommendations into schedule authority.

A one-call view of delivery status, alerts, phase/milestone progress, and directional financial and staffing signals for a project.

List, create, update, archive, and manage canonical project/engagement identity, members, and settings without exposing raw GUI-shaped project JSON.

Read governed delivery timeline geometry and run explicit Delivery milestone commands without letting advisory phase geometry become schedule authority.

Read sprint planning context, readiness, and proposal-only next moves without letting planning projections become authoritative execution placement.

Read governed delivery knowledge two ways: query the deterministic, evidence-backed Context Engine for a synthesized project, engagement, or portfolio briefing (findings, risks, blockers, coverage, freshness, provenance), or search the Engagement Knowledge Corpus for exact evidence passages — ranked full-text retrieval of ingested Context Sources with source, page/paragraph, and provenance-quality attribution. Search returns evidence, never a synthesized answer; neither surface exposes package-authoring authority or a competing source of truth.

Query grounded project context through an authorization-aware registry of eight read providers: Knowledge, Adoption, Delivery Health, Decisions, vSOW snapshots, Relationships, Context Opportunities, and derived Context Concepts. A request flows Sources → domain providers → an authorization-safe router → transports. The provider contract is read-only; each provider preserves its own permissions and authority instead of becoming a second source of truth. Opportunity results are the one exception to mutable_here=false because DigitalStack owns that advisory record itself. Concepts remain mutable_here=false and are derived on every read from authorized opportunity records, never persisted as a second object store. Recording an opportunity uses a separate canonical command service and is never approval — it never mutates, promotes, or supersedes governed truth. Evidence is not automatically a fact, an inference is not an approved decision, an opportunity is not an approved outcome, and partial results are expected and explicit. Distinct from the legacy Context Engine (deterministic briefing + corpus search) — see the Context Engine capability for those surfaces. Internal DigitalStack surfaces are beginning to consume the platform directly (the Execution Copilot adds supplemental grounding from it — controlled rollout, default off); not every product surface uses it yet.

Create and inspect project-scoped estimates, understand scope/effort/role/risk/timeline composition, compare immutable versions, draft governed changes, request human review, and freeze explicit baselines without granting advisory models scheduling or commitment authority.