Skip to main content

Founder Feedback Action Plan — March 31

Source: Sergey's feedback on sprint review deliverables (access chain grouping, evidence model, ownership workflow, operating layer scope).

Feedback document: Founder Response: Access Chain Based on Observed Execution


Feedback Summary

Sergey provided feedback across four areas:

  1. Access Chain model — Collapse/group the execution surface. "Access Chain" is the canonical name (Notion already updated). The unit of risk, control, remediation, and prioritization — not just a grouped view.

  2. Evidence model — Direction is right, but the model mixes claim type and evidence strength. Separate into distinct fields. Simplify user-facing trust language. owner_role is a routing hint only, not the ownership solution.

  3. Ownership workflow — The ownership assignment workflow (PR #243) is the real mechanism for assigning accountable people. owner_role should not be framed as the ownership answer.

  4. Operating layer — Strongest product idea is routing + accountability + mitigation tracking + attestation on top of access-chain analysis. NOT full posture management. Phase ordering and API proposals are engineering proposals, not founder-approved roadmap.


Current State vs Feedback

1. Access Chain Naming & Model

AspectCurrent StateRequired Change
TerminologyIdentityAccessSurface, AuthorityPathDoc, authority-paths in URLsRename to AccessChain in all docs and new code
Product framingGrouped view toggle over paths (Option A)Virtual first-class object — access chain IS the product unit
Canonical axis"Unresolved" between identity/workload/domainIdentity is the default axis for v1 (decided)
UX modelParent row with expandable child pathsAccess-chain-first card: identity → what it reaches → why it matters → what to do
RemediationPer-path actions (5 actions for 1 identity)One remediation object per access chain
Prioritizationmax_severity + finding_count onlySurface score: blast radius, sensitivity, execution intensity, drift, cross-workload reuse, ownership quality
Time narrativeexecution_30d + last_execution_at aggregatedBehavioral profile: newly active, steadily active, bursty, expanding, dormant

2. Evidence Model

AspectCurrent StateRequired Change
ClassificationSingle EvidenceClassification axis (5 values mixing claim type + strength)Two fields: claim_type (what we assert) + evidence_strength (how confident)
User-facing language5 labels ("Proven from execution logs", "Correlated across data sources", etc.)Simplified set of 3-4 plain-English trust states
owner_roleDEFAULT_OWNER_ROLES maps finding type → team name, exposed in EvidenceClaimRename to routing_hint or suggested_team — clarify it's not ownership
UI badgesColor-coded by classification (PR #242)Update to reflect separated claim type + strength

3. Ownership Workflow

AspectCurrent StateRequired Change
Assignment mechanismOwnershipAssignmentDoc with assign/revoke (PR #243)Already correct — this IS the ownership solution
owner_role confusionNamed owner_role in EvidenceClaim, suggests it's ownershipRename + document as routing hint only
ScopeTargets path or entityShould target access_chain once the model is first-class

4. Operating Layer

AspectCurrent StateRequired Change
RoutingDEFAULT_OWNER_ROLES suggests a team per finding typeKeep as initial routing hint
AccountabilityOwnership assignment workflow existsExtend to access-chain level
Mitigation trackingReferenced in research (#215), not builtDesign needed — one mitigation per access chain
AttestationNot startedNear-term scope if it strengthens the operating loop
Scope framingNot explicitly scopedNOT posture management — operating layer on top of access-chain analysis

Action Plan

Wave 1: Reframe the Research Doc (documentation only)

Goal: Align the core research document with Sergey's access-chain-first model.

Changes to 2026-03-26-access-path-identity-scoped-grouping.md:

  1. Rename IdentityAccessSurfaceAccessChain throughout
  2. State explicitly: identity is the default canonical axis for v1
  3. Reframe Option A from "grouped view" to "virtual first-class object"
  4. Define one remediation object per access chain (Sergey's point #4)
  5. Sketch access-chain ranking model (blast radius, sensitivity, execution intensity, drift, cross-workload reuse, ownership quality)
  6. Add behavior/drift narrative at access-chain level (newly active, bursty, expanding, dormant)
  7. Rework UX section: access-chain-first card/detail model, paths as expandable evidence

Update founder feedback doc: Link to revised research, update status.

Wave 2: Evidence Model Design (new research doc)

Goal: Design the separated evidence model before touching code.

Deliverable: New research doc 2026-03-31-evidence-model-separation.md

  • Propose two-field model: claim_type + evidence_strength
  • Map current 5 classifications into the two-axis model
  • Define simplified user-facing trust language (3-4 plain-English states)
  • Impact assessment: evaluator rules, UI badges, evidence packs, API response shapes
  • Migration path from current single-axis model

Wave 3: Terminology Alignment (code + docs)

Goal: Eliminate owner_role confusion.

  • Rename owner_rolerouting_hint in EvidenceClaim and DEFAULT_OWNER_ROLESDEFAULT_ROUTING_HINTS
  • Update evaluator buildEvidenceClaim() and all rule files
  • Update UI if owner_role is displayed anywhere
  • Run full typecheck + tests after rename

Wave 4: Operating Layer Scope Doc (new research doc)

Goal: Define the operating layer without overcommitting to posture management.

Deliverable: New research doc 2026-03-31-operating-layer-scope.md

  • Scope: routing → assignment → mitigation tracking → attestation
  • Explicitly NOT posture management
  • Reference existing ownership workflow (PR #243) as foundation
  • Access-chain-level remediation design
  • Attestation workflow sketch (if near-term)
  • Phase ordering as engineering proposal, not roadmap commitment

Decisions Needed

DecisionOwnerOptions
Approve access-chain-first reframeSergeyAdopt as described, or adjust scope
Evidence model separation scopeSergey + IvanFull two-axis redesign now, or cosmetic rename + defer structural change
Operating layer scope boundarySergeyInclude attestation in near-term, or defer
owner_role rename timingIvanWave 3 (now) or bundle with evidence model redesign

Next Action

Status: research-in-progress

Decision needed from: Ivan (engineering scope and sequencing)

Immediate next steps:

  1. Wave 1 is in progress — reframing the access path grouping research doc
  2. Wave 2 evidence model design can start in parallel
  3. Waves 3-4 depend on decisions above

GitHub Issues: To be created after Wave 1-2 designs are reviewed