Workspace

Workspace overview

Workspace is the current experience hub for governed knowledge, grounded answers, and agent operations. The mental model is five capability areas that share one policy plane for ACL checks, citation integrity, capability lock enforcement, and acceptance gates.

Workspace is the experience lens

Workspace is the operator-facing lens over Intake, Atlas, Lens, Warrant, Conduit, Lineage, Crucible, and Ports. Teams work through one hub while module ownership, evidence quality, and decision policy stay explicit.

Bring in knowledge

Move governed team material into one ACL-aware corpus without changing the systems where work begins.

Current modules: IntakeLineage

Read-only Confluence, SharePoint, Slack, and Markdown connectors with resumable checkpoints.
Binary .pptx, .docx, .txt, .pdf, and Markdown upload through the normalization pipeline.
Team file stores for decks and templates, with owning-team access and explicit cross-team grants.
Source permissions preserved before retrieval ranking; every revision remains traceable.

Knowledge base and vault · Content connectors · Import and onboarding · Documents

Structure and govern it

Turn documents into typed knowledge and versioned commercial truth with named ownership.

Current modules: AtlasWarrantLineage

Typed entities and relations with evidence, authority, lifecycle, owners, and review history.
Versioned structured records for matrices, deal configurations, pricing rules, and estimation factors.
Deterministic pricing, estimator ranges with assumptions, alignment checks, and proposal generation.
Versioned assets and team-installed skills tied to stores the caller can already read.

Capability map · Concept governance · Data lineage · Sales AI composition

Ask and work with live data

Answer from governed sources and current systems, or decline when the evidence is insufficient.

Current modules: LensConduitWarrantLineage

Hybrid lexical/vector retrieval, graph fusion, reranking, and token-budgeted context assembly.
Configurable claim-level citation enforcement (off, shadow, or enforce) with Source [S] and Data [D] citations plus clarify/conflict/abstain outcomes; raising enforce is refused while an active acceptance gate fails.
Read-only named queries with typed parameters, row entitlements, caps, and encrypted credentials.
Frozen result snapshots and hashes so a cited number remains verifiable after the source changes.

Retrieval pipeline · Semantic and data layer · Answer policy · Workspace settings

Build agents and skills

Package a repeatable job behind chat, durable runs, the SDK, or a narrowly scoped MCP mount.

Current modules: PortsWarrantLensConduit

Immutable agent versions with exact tool allowlists, knowledge scope, budgets, memory, and guardrails.
Capability locks enforced at publish, run time, and /api/mcp/agents/{slug}; role and ACL checks still intersect every call.
Configurable HITL rules: approve, reject, edit, or earned autopilot with risk scoring and retrospective review.
Team-installed, versioned skills that inject instructions and expand context only to accessible bound stores.

Creating agents · Approvals and feedback · Tools, MCP, and HITL · MCP server

Measure and operate

Treat quality, security posture, cost, and rollout safety as product capabilities.

Current modules: CrucibleLineagePorts

Golden sets, graded qrels, real-pipeline evaluation, RAGAS/CRAG/IR metrics, and significance-tested comparisons.
Acceptance gates that can veto consequential settings changes, plus scheduled drift and alerts.
Append-only audit, run traces, cost attribution, approval SLA escalation, export, and offboarding.
Configurable provider posture including zero-retention requirements, plus complete OpenAPI and typed SDK surfaces.

Evaluation framework · Tenancy and security · Multi-provider AI · Scheduling, drift, and alerts · Workspace settings · Typed SDK · API reference

Overview flows

These three runtime views show how team file stores, policy checks, and scoped tool access compose into consistent user outcomes.

Team knowledge flow

Downloads

Concepts

Workspace current diagrams
Workspace team knowledge flow
Workspace

Keywords

Upload .pptx / .docx / .txt / .pdf
Team file store
source + revision trail
Team-installed skill
retrieval + cited chat
ownership + explicit grants
normalize + chunk + embed
entities + structured records
bound to accessible stores
Answer with source citations
Intake
Atlas
Lineage
Lens

Source and generation provenance

Status: current

Generated at: 2026-08-12T23:42:43.067Z

Source hash: 58a87f3d15acf72fe55ce761f44567898092c83898f81f30169d1191155d6d36

Metadata payload hash: e02bf7208ea4926b22c8ec7361e518d401e2653ff8b20532ece0ae72292bfb89

Canonical appearance

src/components/docs/workspace-current-diagrams.ts:4 route /docs/workspace

All appearances

canonical — src/components/docs/workspace-current-diagrams.ts:4 route /docs/workspace

No mirrored appearances.

Generation versions

App: eli-ai 0.1.0

Mermaid: 11.16.0 · Mermaid CLI: 11.16.0

Node: v26.3.1 · Yarn: 4.17.1

Renderer config hash: 68c10966fe84406ee626034d58bfabd555df9f65f691204b7c46db24038da101

Renderer theme hash: c80287a78d80ad63d27bd5ca348b2ef9a7e2f44da289e436be6484ea28a1b033

Adapter versions: diagramGenerator=2, drawioFlowchart=1, drawioGantt=1, drawioSequence=1, drawioState=1

Full sidecar JSON: team-knowledge-flow-58a87f3d.json

Grounded answer flow

Downloads

Concepts

Workspace current diagrams
Workspace grounded answer flow
Workspace

Keywords

append-only provenance
User + live role + content ACL
hybrid retrieval + graph fusion
authority + citation policy
optional governed live query
Answer / clarify / conflict / abstain
evaluation + acceptance gates
Lens
Warrant
Conduit
Lineage
Crucible

Source and generation provenance

Status: current

Generated at: 2026-08-12T23:39:05.035Z

Source hash: 01b02f3dc18010b6efcfcd001606218a20b80b5ef54330e68782eee978060b41

Metadata payload hash: 320aed688db727e16f057c6740a5ad35131151e2841eb1641e487a29e2ad8f10

Canonical appearance

src/components/docs/workspace-current-diagrams.ts:15 route /docs/workspace

All appearances

canonical — src/components/docs/workspace-current-diagrams.ts:15 route /docs/workspace

No mirrored appearances.

Generation versions

App: eli-ai 0.1.0

Mermaid: 11.16.0 · Mermaid CLI: 11.16.0

Node: v26.3.1 · Yarn: 4.17.1

Renderer config hash: 68c10966fe84406ee626034d58bfabd555df9f65f691204b7c46db24038da101

Renderer theme hash: c80287a78d80ad63d27bd5ca348b2ef9a7e2f44da289e436be6484ea28a1b033

Adapter versions: diagramGenerator=2, drawioFlowchart=1, drawioGantt=1, drawioSequence=1, drawioState=1

Full sidecar JSON: grounded-answer-flow-01b02f3d.json

Capability-locked agent flow

Downloads

Concepts

Workspace agent flow
Workspace current diagrams
Workspace

Keywords

documents · graph · data · ...
Agent builder
Immutable agent version
Capability lock
Workspace chat harness
Exactly scoped tools
HITL policy
reviews + gates + drift
approve · reject · edit · autopilot
/api/mcp/agents
{slug}
slug
Crucible

Source and generation provenance

Status: current

Generated at: 2026-08-12T23:37:10.924Z

Source hash: 9adcef5fb0cc0ea4695812a34aa1c726888c345b2beb93251227b8a48571354d

Metadata payload hash: 2fc5553f3b8b09cef9d2d14a499248f50da31765e28dede83cb9577ba38411f2

Canonical appearance

src/components/docs/workspace-current-diagrams.ts:25 route /docs/workspace

All appearances

canonical — src/components/docs/workspace-current-diagrams.ts:25 route /docs/workspace

No mirrored appearances.

Generation versions

App: eli-ai 0.1.0

Mermaid: 11.16.0 · Mermaid CLI: 11.16.0

Node: v26.3.1 · Yarn: 4.17.1

Renderer config hash: 68c10966fe84406ee626034d58bfabd555df9f65f691204b7c46db24038da101

Renderer theme hash: c80287a78d80ad63d27bd5ca348b2ef9a7e2f44da289e436be6484ea28a1b033

Adapter versions: diagramGenerator=2, drawioFlowchart=1, drawioGantt=1, drawioSequence=1, drawioState=1

Full sidecar JSON: capability-locked-agent-flow-9adcef5f.json

Entry points

Entry	Scope
Workspace UI	Chat, documents, graph, records, assets, stores, skills, approvals, evals, and settings
REST API	92 OpenAPI operations with workspace/user API keys and live role narrowing
Typed SDK	@eli-ai/contracts + @eli-ai/client, split by capability module
Workspace MCP	/api/mcp — the governed workspace tool surface
Per-agent MCP	/api/mcp/agents/{slug} — exactly the capability-lock projection
Workers and schedules	Durable runs, connector syncs, evaluations, reports, alerts, and approval SLA escalation

The operation count above is sourced from PUBLIC_API_OPERATIONS so the page stays current as routes evolve.