Skip to main content

Dobby Overview

This longform module page is the manufacturer-docs entry point for the module. It uses the product overview when that surface is admitted and falls back to the first safe versioned detail page when the overview is still blocked. Use it first; drop into the versioned product pages only when you need deeper surface detail or channel-specific reference.

  • Canonical module key: dobby
  • Source repo: JaddaHelpifyr/jhf-dobby
  • Preferred versioned channel today: stable

Versioned product docs​

Use these direct entry points when you need exact product behavior, operations, or reference detail:

jhf-dobby is the Helpifyr adaptive-learning runtime. It turns governed runtime evidence into replay-backed recommendations, proposal records, and approval-aware adaptation flows.

When To Use Dobby​

Use Dobby when the stack needs governed learning, replay-backed evaluation, or approval-aware recommendation flow around runtime evidence. It is the right module for bounded adaptation proposals and recommendation posture, not for product-runtime truth, policy authorship, or business-domain ownership.

What Dobby Solves​

  • keeps learning-state orchestration in one bounded service instead of scattered scripts
  • preserves provenance and replay evidence for each candidate change
  • consumes canonical governance truth from Fabric instead of inventing local contract or admission truth
  • forces approval and policy integrity through Warp before activation paths can proceed

What Exists Today​

  • FastAPI runtime with health, readiness, learning, promotion, and metrics surfaces
  • Postgres-backed run, replay, proposal, and metric persistence
  • replay evaluation with threshold contracts and bounded budget controls
  • evidence-to-proposal replay loops for agent-work improvements without direct truth mutation
  • proposal lifecycle with approval checks, activation, rollback, and fail-closed degradation
  • optional Shuttle evidence enrichment and Bobbin-marked provenance publication
  • low-pressure service and worker runtime guardrails for shared hosts
  • canonical docs platform v1.6 base family rooted at:
    • README.md
    • docs/OVERVIEW.md
    • docs/INSTALL.md
    • docs/CONFIGURATION.md
    • docs/OPERATIONS.md
    • docs/TROUBLESHOOTING.md
    • docs/RELEASE_NOTES.md
    • docs_manifest.yml

What Dobby Owns​

  • adaptive-learning orchestration
  • durable learning-state persistence
  • replay verdicts and proposal-state transitions
  • Dobby runtime health, readiness, and metrics

What Dobby Consumes​

  • Fabric governance and admission truth
  • Warp approval truth
  • Shuttle read-only evidence
  • Bobbin provenance sink semantics

What Dobby Does Not Own​

  • policy registry or admission ownership
  • ERP/CRM/procurement source-of-truth data
  • direct repo mutation or autonomous code editing
  • hidden black-box supplier or people scoring

How To Verify The Current Repo-Owned Source Truth​

  • python scripts/validate_docs_platform_v1_6.py
  • python scripts/validate_contract_conformance.py
  • python scripts/validate_fabric_consumer_contract.py
  • python <local-workspace-redacted>/helpifyr-fabric/scripts/docs/validate_repo_docs_skeleton.py --repo <local-workspace-redacted>/_worktrees/jhf-dobby/111-wave6-public-docs-source-curation

These checks validate the current public-safe source docs and Dobby-owned contract/governance posture without claiming that foreign owner lanes are healthy.

License: AGPLv3. See ../LICENSE (LICENSE). Learn more at helpifyr.com.

Where To Go Next​

License: AGPLv3. See ../LICENSE (LICENSE). Learn more at helpifyr.com.