Workspace intelligence

Workspace intelligence gives operators and agents one bounded answer to the question “what’s going on across this workspace right now?” without opening a separate inspector for each subsystem. It composes cartographer posture, stigmergy heat, enforcement status, and workstream signals into a single read-only workspace_intelligence JSON packet, delivered on workspace_snapshot, GET /v1/workspace-environment, and MCP bmo_get_workspace_snapshot.

It does not authorize mutation, replace file reads, or assert filesystem truth. Individual slices (heat, markers, trail, signal pressure) remain the drill-in surfaces when you need raw detail.

Maturity: Advanced advisory surface. Use it when you want a single cross-subsystem summary before opening cartography, stigmergy explain, or enforcement deposit flows.

What the packet contains

Each packet (workspace-intelligence-packet/v1) may include independent source entries with their own freshness and degradation:

Source	What it summarizes
`cartographer`	Bounded living-architecture evidence when zone paths overlap generated proof
`workspace_strategy`	Cross-session strategy memory posture for the workspace
`stigmergy`	Signal-fabric pressure and zone context for the scoped glob
`enforcement`	Substrate posture and active primitives when enforcement is enabled
`workstreams`	Open workstream / review context when scoped
`change_throughput`	Parallel-writer advisory totals and deposit follow-through telemetry

The packet may also carry enforcement_deposit_recommendations: advisory suggestions naming a primitive kind, target zone, evidence refs, and suggested mode/TTL with authority manual-deposit-required. These are not deposits — use the existing enforcement_deposit surface explicitly when you act.

Optional intelligence_lifecycle on the parent workspace snapshot reports cross-surface readiness vocabulary (cold, warming, ready, degraded, stale, absent, unknown) when lifecycle producers are attached (for example LSP and in-flight analyzer work; verify-run may appear on dedicated verify trailers rather than every snapshot read). Lifecycle metadata is advisory; it does not block reads or authorize mutation.

When to open what

Use this routing before diving into a single tab or tool:

Question	Start here
Is the observer healthy?	`/workspace-snapshot`
Where is pressure spatially?	`/cartography` (Overview · Heat · Markers · Trail)
Which zones are hot or contested?	GRADIENT · signal zones — `/signal-zones`, `bmo signal zones`
Why does this zone matter?	`bmo stigmergy explain` or `/signal-zones`
What tension categories apply?	`bmo stigmergy explain` embedded `tension_manifest` or tension topic
Parallel writers / claims?	`/workspace-throughput`
Path claim overlaps?	`/workspace-claims`
Should I deposit enforcement?	`/enforcement` — packet recommendations are hints only

Operator surfaces

Surface	Entry
Agent tool	`workspace_snapshot` — read `workspace_intelligence` and optional `intelligence_lifecycle` keys
HTTP	`GET /v1/workspace-environment` (Bearer auth)
MCP	`bmo_get_workspace_snapshot` (list in `options.mcp_server.exposed_tools`; not exposed by default)
TUI	`/workspace-snapshot` → links to `/cartography`; cartography Overview synthesizes the packet
CLI	`bmo config show-workspace-snapshot`; `bmo workspace preflight`

Stigmergy explain/triage/zones compose a deeper capability readout on Stigmergic delivery; the workspace intelligence packet is the snapshot-local composition hub.

Configuration

Most sources read persisted or in-process state; there is no single options.workspace_intelligence block. Enable the underlying producers:

[options.workspace_observer]
# enabled = true  # default; required for snapshot root

[options.workspace_markers]
# enabled = true  # default-on; set false to opt out

[options.signal_fabric]
# enabled = true  # default-on; set false to opt out

[options.enforcement]
enabled = true  # off until enabled; source + deposit recommendations when on

[options.workspace_strategy_memory]
# enabled = true  # default; strategy source in packet

Cartographer context appears when generated proof exists and zone paths overlap evidence. Its packet source also reports advisory posture: ready, stale, unavailable, malformed, or empty, plus bounded source/projected counts and proof provenance (task cartographer:generate for maintainers). See the configuration reference for merge precedence.

Work intelligence substrate (deeper readouts)

When you pass workstream_id or session_id to stigmergy explain, triage, or signal zones, BMO may attach substrate evidence: patch proposals, review findings, arena candidates, and memory hints. That composition is read-only with explicit_only mutation on capability guidance.

Advanced intent/influence substrate lanes are maintainer-documented only — see Workspace intelligence topic.

Start with Stigmergic delivery for the operator loop; use this page when you need the snapshot packet contract.

Scenario captures

Echo-seeded tmux captures document the routing story above. The journey recording shows hub → zones → explain; the explain PNG is the final frame only. Deposit follow-through uses the JSON excerpt capture (not the throughput TUI). Maintainer beat specs: intelligence scenario corpus (anchor: contested parallel delivery — swappable for adjacent domains).