Skip to content
Crosswalker
🚧 Early alpha — building the foundation. See the roadmap →

Foundation state of play — the web-of-webs and the open problems

Created Updated

Why this log exists

Section titled “Why this log exists”

After ~3 weeks away from the project, the abstract problem we’re solving in Foundation phase is hard to hold front-of-mind. There are now 27 dated log entries in zz-log/, 7 research challenges, and a multi-section roadmap with several forced-but-unmade decisions. The two large 04-10 syntheses covered most of the substance, but reading them cold doesn’t reboot the frame — only the contents.

This log attempts the frame. One metaphor, four kinds of structured graph, six open questions mapped onto the picture, and one canonical reading order. It deliberately makes no new claims and commits no decisions. Every assertion below is backed by a link to an existing log, challenge, concept page, or roadmap item.

The web-of-webs framing

Section titled “The web-of-webs framing”

The general problem is structured webs connecting to structured webs.

Each ontology Crosswalker handles — NIST 800-53, CIS Controls, MITRE ATT&CK, SCF — is itself a structured web: typed nodes (controls, techniques, sub-techniques), typed edges (parent/child, references, mitigates), and edge metadata (confidence, scope, applicability). When we import one, we’re reifying that web into the filesystem as folders, notes, frontmatter, and links.

Crosswalker’s job is what happens between those webs. There are at least four kinds of inter-web connection in play, plus one optional intermediary:

                                  ┌────────────────────────┐
                                  │  evidence-vault web    │
                                  │ (your policies, audits,│
                                  │  technical docs, etc.) │
                                  └───────────┬────────────┘
                                              │
                                              │ evidence-link edges
                                              │ (junction notes, 13-field schema)
                                              ▼
   ┌──────────────┐    crosswalk      ┌────────────────┐    crosswalk     ┌──────────────┐
   │  NIST 800-53 │  ◄──────────────► │  optional      │ ◄──────────────► │  CIS Controls│
   │  ontology    │      edges        │  pivot web     │      edges       │  ontology    │
   │  web         │                   │  (a.k.a. spine,│                  │  web         │
   │              │                   │  hub, canonical│                  │              │
   │              │                   │  intermediate) │                  │              │
   └──────┬───────┘                   └────────────────┘                  └──────┬───────┘
          │                                                                      │
          │     lifecycle edges (between versions of the same web)               │
          └──────────────────────────────────────────────────────────────────────┘
                  9 atomic primitives: add / remove / rename / split / merge / ...

The four edge types and the optional pivot (a.k.a. spine, hub, interlingua, canonical intermediate) are the things Foundation is figuring out.

The webs in play

Section titled “The webs in play”
WebWhat it isLinked to
Source-ontology webAn imported framework — folders + notes + typed links carrying the structure of NIST/CIS/MITRE/etc. into the vaultOntology lifecycle · Framework crosswalks · Why files
Evidence-vault webThe user’s own notes — policies, audit reports, configuration evidence, interview notesEvidence mapping for GRC teams · Link metadata system (historical)
Optional pivot web (a.k.a. spine, hub, interlingua, canonical intermediate)A canonical intermediate ontology between source webs — synthetic (computed via FCA), inherited (SCF, OSCAL catalog), or handcraftedChallenge 06 · Schema matching: pivot approach
Lifecycle layer — two distinct levels: (a) versions of an ontology over time, connected by structural-change atoms (add, remove, rename, split, merge, deprecate). (b) the meta-web: schemas Crosswalker itself invents for ontology lifecycle management (StewardshipProfile, junction-note schema, FrameworkConfig, etc.) also have lifecycles — Crosswalker eats its own dog food.Ontology evolution · Atomic operations research · StewardshipProfile (formerly EvolutionPattern) draft · 05-01 §2.5 meta-schema commitment

Every Foundation question is a question about one of these webs or the connections between them.

The connections in play

Section titled “The connections in play”
Edge typeConnectsStatusGoes to
Crosswalk edgessource-ontology web ↔ source-ontology webVocabulary direction set (STRM + SSSOM); form factor (junction notes? frontmatter? sidecar files?) still pendingFoundation synthesis §“Edge semantics”
Evidence-link edgesevidence-vault web → source-ontology webArchitecture committed: junction notes (one file per edge), 13-field flat-YAML schema, OSCAL by-component round-trip, git audit trailEvidence-link edge model synthesis · Challenge 07
Hierarchy edgeswithin a single ontology webFolder structure carries this — solved by the generation engineOntology lifecycle: Import
Lifecycle edgesbetween versions of the same source-ontology web9 atomic operations + 4 composites, provably complete per graph edit distance literatureAtomic operations research · Primitives depth

SSSOM dual-confidence borrowing applies to both crosswalk and evidence-link edges. BFO continuant/occurrent typing — if adopted — applies at the node level, not the edge level.

Open Foundation problems, mapped to the web-of-webs

Section titled “Open Foundation problems, mapped to the web-of-webs”

Six forced-but-unmade decisions. Each is fundamentally a question about one of the webs above or the edges between them.

Question (web-of-webs framing)StatusWhere it lives
Do we add an intermediary web (pivot / spine / hub / interlingua) between source-ontology webs? Or keep direct pairwise edges and accept O(n²) maintenance?Open research — biggest open questionChallenge 06 · Foundation synthesis §“synthetic spine” · Roadmap: Foundation
What labels go on web → web edges? And what metadata envelope wraps each label?Ready to commit — STRM + SSSOMFoundation synthesis §“edge semantics” · Roadmap: Foundation
What does an evidence-vault → ontology edge physically look like? What schema? What status vocabulary? What audit trail?Ready to commit — junction notes, 13-field schema, 3D status, git historyEvidence-link synthesis · Challenge 07 · Roadmap: Foundation
How do we describe a web’s lifecycle? Per-framework stewardship defaults? Per-version transformation recipes? Both stacked?Needs a position; research direction setStewardshipProfile (formerly EvolutionPattern) draft · Primitives depth · User-first maintenance · Roadmap: Foundation
Do nodes carry a formal ontological type? Adopt BFO continuant/occurrent as node.type? Skip BFO entirely?Design callFoundation synthesis §“EvolutionPattern needs formal grounding”
How do we query the entire web-of-webs at scale? When does Tier 2 sql.js kick in? What’s the SQL schema? What does Bases force?Needs perf testing + schema designProgressive tier architecture pillar · Foundation synthesis §“files-canonical ceiling” · Architecture at scale: Challenge 02

Things that are not on this list because they’re already settled at Foundation level: file-first storage (why-obsidian-why-files), config-as-code (vision alignment §2), layered architecture (layered architecture vision), the 9 atomic operations (atomic operations research).

Where we left off (3-week refresh)

Section titled “Where we left off (3-week refresh)”

Last real activity: 2026-04-10. Twelve commits landed across two parallel synthesis sessions (foundation AM, evidence-link PM). Eight .workspace/ research drafts (~225 KB total) fed those syntheses; two of them remain undistilled into KB form.

The system is now in a decide-next state. Suggested phasing:

  • Phase A — Lock in what’s ready (pure paperwork, unblocks docs):
    • Commit STRM + SSSOM crosswalk vocabulary → new dated log + roadmap update
    • Commit evidence-link junction-note architecture → new dated log + roadmap update
    • Commit BFO scope decision (probably continuant/occurrent only) → new dated log
  • Phase B — Run the gating research:
  • Phase C — Downstream design (only after A+B):

Phase A items don’t change behavior — they convert “the syntheses said this” into “we committed to this.” Phase B is the actual remaining Foundation research. Phase C unblocks development.

Reading order for fresh agents

Section titled “Reading order for fresh agents”

If you (or a fresh agent) need to spin up on Foundation work from cold:

  1. This log — orientation
  2. Terminology — vocabulary
  3. Ontology lifecycle — the 5-phase cycle (Acquire → Import → Enrich → Maintain → Share)
  4. Foundation research synthesis (04-10) — the morning synthesis covering 4 parallel research sessions
  5. Evidence-link edge model synthesis (04-10) — the afternoon synthesis resolving Challenge 07
  6. Roadmap: Foundation section — the live decision checklist
  7. Challenge 06 and Challenge 07 — the two open research briefs
  8. The Apr 8–9 cluster for the primitives / atomic-operations background
  9. The Apr 3 cluster for the original architectural commitments

That’s about 4 hours of careful reading; this log alone should take 10 minutes and refresh enough mental scaffolding to make the rest of the path navigable.

Section titled “Related”

The two 04-10 syntheses (must-read):

Apr 8–9 cluster (primitives, atomic operations, user-first UX):

Apr 3 cluster (architectural foundations, highlights):

Research challenges (fresh-agent briefs):

Concept pages (terminology and conceptual depth):

Roadmap and challenge index:

Undistilled .workspace/ drafts (gitignored, repo root):

These two 04-10 research drafts haven’t been distilled into KB logs yet — flagged for a future cleanup pass:

  • Claude-Ontology primitives for knowledge work architecture.md (~92 KB) — largest of the 04-10 batch
  • Claude-Ontology primitives for resilient knowledge work.md (~35 KB)

Six other .workspace/ drafts from the 04-10 batch have already been distilled into the two syntheses above but remain in .workspace/ until the cleanup pass.