Feature Map

Ownerless contributor map of the major Livepeer Docs repository systems, philosophy, governance contracts, and implementation surfaces.

Feature Map

This repository is a docs-as-infrastructure system, not a folder of pages. The intended operating model is ownerless: every governed surface declares its canonical source, validator, repair path, and primary gate so a contributor or agent can diagnose and repair drift without private maintainer context.

Use this page as the high-level system map. Use the linked feature pages for implementation details and the Gap Analysis for purge, migration, and remediation work.

---

Core philosophy

Principle	What it means in this repo	Primary evidence
Ownerless governance	A surface is not production-ready unless it has one canonical source, one validator, one repair path, and one primary gate	policies/ownerless-governance.mdx · canonical: operations/governance/config/repo-governance-surfaces.json (5 unified surfaces, 4 ownerless-ready, 1 advisory post-D-ACT-10); legacy ownerless-governance-surfaces.json (8 entries) superseded
Source of truth before convenience	Generated outputs, indexes, and registries must be regenerated from canonical inputs, never hand-edited	policies/source-of-truth-policy.mdx · operations/governance/config/generated-artifacts.json (29 artefacts)
Root causes over bypasses	Failures get repaired in the policy, generator, source data, or validator that owns the drift. Ignoring files or hiding failures is not sufficient	AGENTS.md · .githooks/ · policies/quality-gates.mdx
Layered enforcement	Fast local checks stay in pre-commit; branch and AI task checks stay in pre-push; broad browser/link/network sweeps belong in scheduled CI	policies/infrastructure-principles.mdx · operations/tests/WHEN-TESTS-RUN.md
Public docs and repo ops are one product	Content, components, data feeds, automation, AI artefacts, and governance docs evolve together	docs.json · v2/ · snippets/ · operations/ · docs-guide/
Honesty about partial state	”Ownerless” covers routine drift, not policy authorship or destructive operations. Live state never lies in canonical docs	This map + Gap Analysis are dated; every count traces to a generator or a verified command

---

Feature index

#	Feature	Main page	Core surfaces	State (2026-05-23)
1	AI features	AI Features	AGENTS.md, ai-tools/, 6 native adapters, llms.txt, sitemap-ai.xml, hosted MCP, Mintlify chat assistant	Production with llms.txt + sitemap-ai.xml 47 days stale (root cause: cron-dry-run); 34 local SKILL.md (18 in draft)
2	UI system	UI System	snippets/components/, snippets/templates/, snippets/composables/, .vscode/*.code-snippets, style.css	Production; 132 registry exports + 37 templates + 8 Tier-1 composables + 312 VS Code snippets; 4-surface count drift (cleanup queue)
3	Automations	Automations	.github/workflows/, .github/workspace/, operations/scripts/, operations/tests/, .githooks/	Production (2026-05-22 4-tier composable refactor): 11 workflows + 321 scripts + 18 locked decisions; cron-is-dry-run bug = P0
4	Data integrations	Data Integrations	api/, snippets/data/, operations/scripts/integrators/, operations/scripts/generators/content/reference/	11 integration families. Contracts pipeline = gold-standard; OpenAPI specs 67 days stale (weakest link); Luma feed silently dead
5	Adaptive architecture	Adaptive Architecture	operations/governance/config/, operations/scripts/remediators/, .githooks/, lpd repair	4 of 5 unified surfaces formally ownerless-ready post-D-ACT-10 (1 advisory: github-workspace-governance); 5-stage control loop (detect/explain/repair/verify/record); honest framing
6	Contributor tools	Contributor Tools	tools/lpd CLI, tools/dev/, tools/editor-extensions/, .githooks/, .vscode/, tools/lib/	Production. lpd CLI (13 subcommands + 5 group shorthands) + 4 VS Code extensions + scoped preview + 312 snippets
7	Content writing pipeline	Content Writing	_MY_PROCESS/, CONTENT-WRITING/, Frameworks.mdx, per-tab _workspace/canonical/	9-stage project pipeline. 0 of 5 tabs has IA Approved (all stuck at Phase 2). 14 supporting AI skills (18 of 34 in draft)
8	Checks framework	Checks Framework	master-checks.mdx, 5 per-tab canonical/checks.mdx, prompts under CONTENT-WRITING/Prompts/, 14 skills	11-phase per-page review pipeline + 9 universal check categories. veracity-pass.md DRAFT – blocks Phase 6
9	Gap analysis	Gap Analysis	workspace/reports/repo-ops/, this feature map, the 10 deep-slice audits	Implementation backlog. P0/P1/P2 priorities with file paths + acceptance criteria; refreshed 2026-05-23

---

System map

Layer	Runtime role	Canonical sources	Derived outputs	Primary gates
Navigation and routing	Mintlify site definition (versions, tabs, groups, redirects, branding, route membership)	docs.json	Mintlify site, generated docs indexes, route lists	operations/scripts/validators/content/structure/lint-structure.js, lpd test --staged
Content	User-facing docs, internal docs, legacy docs, section workspaces	v2/, docs-guide/, frozen v1/	docs-index.json, generated catalogs, browser reports	Content validators, page rendering, link audits
Components and templates	Shared presentation + authoring primitives	snippets/components/, snippets/templates/, snippets/composables/	Component registry, public component-library/ tab, VS Code snippets	Component validators, template generator tests
Data	Structured data consumed by pages + integrator components	api/, snippets/data/, external source configs	Generated API docs, snapshots, feed data, reference maps	OpenAPI reference audit, freshness checks, data-fetch workflows
Operations	Scripts, tests, hooks, CI workflows, repair commands	operations/scripts/, operations/tests/, .github/workflows/, .githooks/, tools/	Reports, generated governance maps, CI artefacts	Pre-commit, PR CI, scheduled workflows
Governance	Machine-readable policy, surface ownership by contract, root allowlist, generated-artefact manifest, approval policy	operations/governance/config/ (6 JSON configs), docs-guide/policies/, docs-guide/frameworks/, docs-guide/standards/, docs-guide/decisions/	Repo governance maps, root allowlist, compliance reports	Governance validators, approval labels, hook checks
AI agent layer	Tool-specific adapters + reusable skills for agents	AGENTS.md, 6 native adapters (.claude/, .codex/, .cursor/, .windsurf/, .augment/, .github/copilot-instructions.md), ai-tools/ai-skills/	Agent packs, synced Codex skills, skill indexes	Agent-docs freshness, Codex branch-contract checks
Content writing pipeline	The 9-stage project pipeline + 11-phase per-page review	workspace/plan/active/_MY_PROCESS/, CONTENT-WRITING/, master-checks.mdx, per-tab canonical/checks.mdx	Approved page drafts, review packets, decisions registry entries	Human gates per phase recorded in tab-status.md

---

Three locked-decision registries

The repo carries 32 locked structural decisions across three ID-prefixed registries. All cross-referenced from the unified index at docs-guide/decisions/registry.md.

Registry	Lives at	Count	Coverage
Content writing decisions	workspace/plan/active/CONTENT-WRITING/decisions/decision-registry.md	1 (D-NAV-01)	Cross-tab navigation patterns
GitHub Actions + governance decisions	.github/workspace/decisions-log.mdx	18 (D-ACT-01..10 + D-GOV-01..08)	Workflow architecture, taxonomy, pipeline tags, dispatcher model, 4-tier composable
Docs-guide structure decisions	docs-guide/decisions/docs-guide-structure.md	13 (D-DG-01..13)	docs-guide IA migration; canonical/ retirement; standards/frameworks/policies/decisions/contributing tiers; thin-adapter rule (D-DG-11)

---

Current inventory snapshot (live, 2026-05-23)

Inventory	Count or state
docs.json registered v2 routes	686
Active v2 MDX outside workspace + archive lanes	1,128
Active GitHub workflows	11 (6 dispatch-{concern}.yml + 5 interface-governance-*) post-2026-05-22 4-tier refactor
Archived workflows	1 in deprecated/ + 61 in workflows/x-archive/ + contents of .github/x-archive/ (third lane)
Active operations scripts	321 (102 dispatch + 58 integrators + 55 validators + 37 remediators + 31 generators + 25 audits + 8 interfaces)
Archived operations scripts	38 in archive/ + x-archive/
Script JSDoc compliance	327 of 321 use canonical @purpose (~100%); 45 use retired @category; 4 use retired @domain
Active JSX components	35 active + 24 archived (59 file count); 132 registry exports
Templates	37 across 14 sub-lanes
Composables	8 Tier-1 + 73 per-tab fragments under pages/
VS Code snippets	312 across 5 .code-snippets files
Assets	236 files (~101.5 MB)
AI skills (local SKILL.md)	34 active + 1 archived; 18 of 34 status: draft
AI portable templates	53
AI portable skill exports	53 (agent-packs/skills/); manifest registers 1
AI tools registry entries	317 (ai-tools-registry.json)
AI artefacts at root	llms.txt (188 entries) + sitemap-ai.xml (181 URLs, 6 of 10 tabs covered) + robots.txt + humans.txt
Native agent adapters	6 (Claude, Codex, Cursor, Windsurf, Augment, Copilot) + AGENTS.md baseline (119 lines) + .mintlify/Assistant.md
.claude/references/ library	25 files, ~2,860 lines, 13 categories
docs-guide canonical files	13 frameworks + 18 policies + 5 standards + 11 features + 4 decisions + 5 contributing + 5 tooling + 6 catalog + 8 docs-library + repo-ops + canonical
Locked decisions across 3 registries	32 total (1 D-NAV + 18 D-ACT/D-GOV + 13 D-DG)
Ownerless surfaces (formally contracted)	8 of 28 in the wider registry
Generated artefacts (registered)	29 in generated-artifacts.json
Per-tab _workspace/ files	3,110 across 11 tab subtrees + v2/_workspace root
Per-tab canonical/checks.mdx files	5 (gateways, orchestrators, about, delegators, developers1)
V2 folder cleanup matrix backlog	307 rows (235 _workspace + 72 x-deprecated; gateways carries 199)
Script footprint audit findings	4 medium + 20 low (no critical/high)
BACKLOG items registered	BL-001..BL-046 (workspace/plan/future/BACKLOG/master-tasks.md)
Active plans under workspace/plan/active/	26

---

Contributor operating loop

1. Read the relevant governance entry point before editing: policies/governance-index.mdx or docs-guide/index.mdx.
2. Identify the canonical source for the surface being changed (every surface declares it in operations/governance/config/*.json or its policy doc).
3. Change the source, generator, validator, or policy that owns the behaviour – never the derived output.
4. Regenerate derived outputs with the named generator command – never hand-edit.
5. Run the smallest relevant validator – usually lpd test --staged.
6. If a failure appears, fix the owning source or repair command. Bypasses are governance debt.
7. For route or nav work, treat docs.json as production infrastructure. Validate route impact via check-docs-path-sync.js + redirect-integrity hook.

---

What good looks like

Every major feature should satisfy this contract:

Requirement	Acceptance check
Discoverable	The feature has a feature page, framework, policy, catalog entry, or generated registry entry
Governed	The canonical source and allowed write paths are documented
Validated	A contributor can run a deterministic check locally or in CI
Repairable	The repair path is exact and does not depend on private knowledge
Bounded	Workspaces, archives, generated files, and public pages are separated by structure and selection helpers
Purgeable	Legacy paths have explicit migration, archive, or deletion criteria
Dated	Every claim is verifiable by re-running a named command at the date the doc was verified

Per the 2026-05-25 audit, 4 of 5 unified surfaces fully satisfy this contract in the canonical registry repo-governance-surfaces.json (bridge_mode "retired", legacy bridge inventory empty post-D-ACT-10). The remaining surface (github-workspace-governance) is at advisory state. The legacy 8-entry ownerless-governance-surfaces.json is superseded but kept for reference. The Gap Analysis is the queue for promoting the remaining advisory surface to autofix or blocking.

---

Related pages

Feature deep dives

• AI Features – public AI surfaces + agent runtime + skill system + adapters
• UI System – component library + templates + composables + design tokens + snippets
• Automations – 4-tier composable architecture + 11 workflows + 321 scripts + hook pipeline
• Data Integrations – 11 integration families incl. contracts (gold-standard) + OpenAPI
• Adaptive Architecture – ownerless contract + 5-stage control loop + remediator families
• Contributor Tools – lpd CLI + scoped preview + VS Code extensions + hooks

Framework deep dives

• Content Writing Pipeline – 9-stage project pipeline + taxonomies + voice + 5 known gaps
• Checks Framework – 11-phase per-page review + 9 universal Cat 1-9 + per-pageType + per-tab variation
• Component Framework Canonical – 6-category taxonomy + 7-tag JSDoc
• Script Framework – type/concern/niche + 11-tag JSDoc
• GitHub Actions Framework – workflow taxonomy + naming convention
• Styles Engineering Guide – --lp-* design tokens + theme contract
• AI Tools Governance – registry + 7-lane model + lifecycle states

Policy deep dives

• Ownerless Governance – the 4-part contract + promotion ladder
• Source of Truth – canonical boundaries per surface
• Infrastructure Principles – gate-layer rule ownership
• Generated Artefact + Hook Governance – 3 classes + commit/hook/CI policies
• Quality Gates – gate-layer map with named tools per tier
• Agent Governance Framework – adapter rules per D-DG-11
• Workspace Lifecycle – TTL rules (30/90 days) + retention enforcement
• Cleanup + Quarantine – conservative deletion discipline
• V2 Folder Governance – non-publishable lane contract
• Governance Index – entry point for all governed surfaces

Standards

• Voice + Copy – canonical voice + UK English + banned words + per-audience extensions + heading rules
• Frontmatter Spec – required + optional frontmatter fields per authority tier
• Naming Conventions – file + folder + script + workflow naming
• Authoring Standard – quality bar + publication gate

Cross-cutting

• Gap Analysis – implementation-ready backlog with P0/P1/P2 priorities (refreshed 2026-05-23)
• Decisions Registry – unified index of all locked decisions (32 total across 3 ID prefixes)
• docs-guide/index.mdx – docs-guide overview (the front door)
• workspace/plan/active/_MY_PROCESS/ – page-authoring + checks process canonical source
• workspace/thread-outputs/repo-consolidation-deep/ – 10 file-level slice audits + master consolidation map (~4,400 lines of evidence)