Skip to content

Latest commit

 

History

History
119 lines (97 loc) · 7.88 KB

File metadata and controls

119 lines (97 loc) · 7.88 KB

JLDN Backlog Schema Specification

Document: docs/specification.md
Author: Jeff Langdon (JL Design Network)
Standard: JLDN Task & Backlog Management Protocol
Version: 2608.23.0-s


1. Architectural Philosophy

The JLDN Backlog Schema adheres to five core design principles:

  1. Document & Repository-Bound Isolation: Task datasets reside either in document JSON frontmatter or dedicated generational companion hubs (.dev/[GEN]/backlog.json). Tasks maintain clean scope and cross-reference lineage without external file corruption.
  2. Phase-Aware Lifecycle Tracking: Every task progresses through explicit developmental phases using a standardized [STATE]:[PHASE] colon-delimited syntax.
  3. Machine-Readable Structure: Tasks are stored as strictly validated JSON objects, allowing automated build pipelines, static analysis scripts, and LLM coding agents to parse, execute, and verify tasks programmatically.
  4. Empty Array Validity: The task array key MUST be a valid JSON array. An empty array ([]) is valid and indicates zero active or historical tasks.
  5. Strict JSON Formatting Standard: Backlog JSON MUST be valid, UTF-8 encoded JSON formatted with exactly 2-space indentation and no tab characters.

2. Storage Architecture & Storage Modes

Mode 1: Embedded Inline Frontmatter

For standalone specification documents, tasks reside directly within the todo array in the Markdown document's frontmatter header (--- block).

Mode 2: Companion Dataset Storage Pattern (The JLDN Standard)

For large rulesets and software repositories, tasks are stored in companion JSON datasets within the Generational Development Hub:

  • Root Project Backlog: .dev/backlog.json
  • Generation Master Task Register: .dev/[GEN]/backlog.json
  • Generation Conceptual Proposals: .dev/[GEN]/ideas.json
  • Generation Audit Ledger (Optional): .dev/[GEN]/audit-log.json — append-only record of every formal audit pass against this generation's work, written unconditionally (including zero-finding runs), so "never audited" and "audited clean" remain distinguishable. Referenced bidirectionally via source_audit_run (§4) on any ISSUE task the pass produces.

3. Unified [DOMAIN]-[KIND]-[NN] Taxonomy

Every task adheres to the standard prefix taxonomy:

  • Domains: Must be 3-5 uppercase alphanumeric characters. Must never match a Kind.
    • ROOT → Universal: Project-level milestones, CI/CD, and repository infrastructure.
    • DOCS → Universal: Internal docs/ wiki rules, mechanics, and documentation.
    • Custom Domains: Developers must define 3-5 character custom domains for their project's internal architecture or generational scope (e.g. AUTH, COMB, G08).
  • Kinds:
    • TODO → Actionable engineering and writing tasks.
    • IDEA → Conceptual proposals and architectural designs.
    • ISSUE → Tracked defects, regressions, and security advisories.

4. Task Object Schema Structure

{
  "id": "ROOT-TODO-01",
  "title": "Short Concise Task Title",
  "status": "in-progress:audit",
  "priority": "high-1",
  "protection": "protected",
  "existed_since": "2608.1.0-s",
  "details": "Detailed description of the mechanical, technical, or architectural work required.",
  "created_at": "2026-08-03T12:00:00+02:00",
  "owner": "Agent-Alpha/Jeff",
  "reason": "Explanation required whenever status is blocked or deprecated.",
  "child_of": "ROOT-TODO-00",
  "prerequisite": "ROOT-TODO-00.1a",
  "relates_to": "ROOT-TODO-00",
  "target_files": [
    "docs/specification.md",
    "CHANGELOG.md"
  ]
}

Property Definitions

  • id (String, Required): Standard taxonomy string ([DOMAIN]-[KIND]-[NN] or legacy TODO-[NN]). Scalable up to 4 digits (\d{2,4}). Supports single-letter sub-tasks (.1a to .1z).
  • title (String, Required): Short imperative summary (minimum 5 characters).
  • status (String, Required): Valid lifecycle state slug matching the 11-state matrix.
  • priority (String, Required): Granular 3-tier sub-priority (critical-1 to low-3) or legacy tier (critical, high, medium, low).
  • details (String, Required): Technical requirements description (minimum 15 characters).
  • created_at (String, Required): ISO-8601 UTC/local timestamp string (e.g. 2026-08-18T16:45:00+02:00).
  • existed_since (String, Mandatory on Creation): GVS version tag of origin ([YYMM].[SUBVERSION].[REVISION]-[TAG]).
  • blocked_since / deprecated_since (String, Conditional): Mandatory when status transitions to blocked or deprecated.
  • protection (String, Completed Tasks): "protected" (locked system element) or "open".
  • child_of (String, Optional): Direct vertical parent task ID establishing a strict DAG.
  • prerequisite (String, Optional): Prerequisite task ID that must reach terminal state before activation.
  • relates_to (String, Optional): Horizontal context association.
  • depends_on (Array of Strings, Optional): Other task IDs this task has a real, audit-surfaced dependency on. Distinct from prerequisite (a single blocking gate a task cannot activate without): depends_on is a plural, softer traceability record — a task can have several, and their completion isn't necessarily a hard activation gate the way prerequisite is.
  • source_doc (Array of Strings, Required whenever a task follows a plan → audit → findings development cycle): Paths to the task's build-plan and audit-findings documents, so the record of how a task was validated survives independently of any narrative outside the backlog itself.
  • target_files (Array of Strings, Optional): Localized file tracking for isolated environments without git dependency.
  • target_version / target_component / target_repository (String, Optional): Optional pointers to eliminate cross-project reference ambiguity.
  • source_audit_run (Object, Optional): Present on ISSUE tasks originating from a formal audit pass (e.g. a Red Team review). Structure: {generation, source, run_at}generation (GVS epoch the run occurred in), source (identifier for the audit or team that produced the finding, e.g. "red"), run_at (ISO-8601 timestamp of the run). Bidirectional with the companion audit ledger's own finding_ids array (§2) — a task's source_audit_run and the ledger entry's finding_ids cross-reference each other, so a claimed source is verifiable, not merely asserted.

5. Formalized 11-State Lifecycle Matrix

Phase / Group Status Slug Definition & Usage Guidelines
1. Initial Work pending Task created; initial work has not started.
in-progress Task is undergoing initial development.
2. Surface Review pending:review Initial implementation complete; queued for surface review.
in-progress:review Undergoing surface formatting and presentation review.
3. Deep Mechanical Audit pending:audit Draft complete; queued for deep mechanical/adversarial audit.
in-progress:audit Active Red Team auditing and loophole analysis underway.
4. Architectural Refactoring pending:refactor Flagged after audit for structural refactoring.
in-progress:refactor Active structural refactoring underway.
5. Control & Terminal completed Done & Verified (Immutable): Finished and locked.
blocked Work stuck due to external dependency.
deprecated Obsolete or cancelled task.

6. Validation Regular Expressions

  • Status: ^(pending|in-progress|pending:review|in-progress:review|pending:audit|in-progress:audit|pending:refactor|in-progress:refactor|completed|blocked|deprecated)$
  • Priority: ^(critical|high|medium|low)(-[1-3])?$
  • Protection: ^(protected|open)$
  • Taxonomy ID: ^([A-Z0-9]{3,5}-(TODO|IDEA|ISSUE)-\d{2,4}|TODO-\d{2,4})(\.\d+[a-z])?$
  • GVS Version: ^\d{4}\.\d+\.\d+-(a|as|b|bs|l|s|ts|z)$