Documentation Index
Fetch the complete documentation index at: https://na-36-merge-docs-v2-dev-draft-into-docs-v2-clean-20260525.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
Content Writing Pipeline
How a v2/ documentation page goes from “needed” to “published”. Two pipelines run in sequence:
- The project pipeline (this page) – 9 stages, applies to a tab or a section of the docs. Takes a tab from “audience defined” to “all pages shipped.”
- The per-page review pipeline (Checks Framework) – 11 phases, applies to a single page once written. Takes a page from “drafted” to “approved for publication.”
The project pipeline writes the pages. The review pipeline ships them. Neither phase ships without the prior phase’s human sign-off recorded in workspace/plan/active/CONTENT-WRITING/decisions/decision-registry.md.
Where this pipeline canonical lives. The full project execution plan (997 lines) and per-stage tooling live under workspace/plan/active/_MY_PROCESS/ + workspace/plan/active/CONTENT-WRITING/. This page is the published reference. Locked structural decisions (D-NAV-01 currently) live in the decision-registry. Per-tab gate status lives in tab-status.md. As of 2026-05-23, no tab has IA Approved – all 5 are stuck at Phase 2 or earlier.
Page taxonomy
Every routable v2/ page declares 10 frontmatter fields. The enumerations below are canonical.
pageType (7 primary)
| Type | Job |
|---|
navigation | Route the reader. Portals, tab landings, section roots. No instructional content; routing only. |
concept | Explain how a mechanism works. Exit state: reader can describe or reason about X (not “do X”). |
tutorial | End-to-end guided learning. Prerequisites + ordered verifiable steps + outcome statement. |
guide | Opinionated guidance for a specific outcome. Takes a position; no step-by-step procedures as primary content. |
instruction | Step-by-step procedure. Prerequisites + numbered steps + verification step + next steps. |
reference | Structured lookup. Tables, ParamField, ResponseField. Highest factual standard. |
resource | Curated discovery. Tools, directories, showcases, community channels. |
pageVariant (refines pageType)
| pageType | Allowed variants |
|---|
navigation | portal, landing |
concept | overview |
tutorial | overview |
guide | overview |
instruction | overview, quickstart, setup |
reference | overview, specification, compendium, changelog |
resource | overview, knowledge-hub |
overview is a cross-type variant for entry/orientation pages within a section. knowledge-hub is the resource variant for external/community content (no instructional content mixed in).
audience (7 tokens)
| Token | Who | Arriving question |
|---|
founder | Evaluating Livepeer for their product/business | ”Is Livepeer right for my product?” |
builder | Building products via APIs/SDKs/hosted gateways | ”How do I add Livepeer to my app?” |
developer | Building custom pipelines, BYOC, protocol extensions | ”How do I build something on the network?” |
gateway | Running gateway infrastructure | ”How do I set up and run a gateway?” |
orchestrator | Running compute nodes | ”How do I set up, run, and earn with a node?” |
delegator | Staking LPT | ”How do I stake and participate?” |
community | Ecosystem, governance, newcomers, researchers | ”How do I get involved?” |
founder → builder → developer → gateway is the documented graduation path. Per-persona refinements (e.g. builder [studio, ai], orchestrator [operator, commercial]) are an optional array field; full enumeration in the content-pipeline-framework.
purpose (15 intent-based values)
orient · explain · learn · choose · evaluate · start · build · configure · operate · troubleshoot · verify · integrate · optimise · reference · update
Deprecated aliases that still appear in some files but should not be authored: landing → use orient. overview → use orient. orientation → use orient. setup → use configure. how_to → use build or configure. operations → use operate. tutorial → use learn. evaluation → use choose. decision → use choose. faq, glossary → use reference. changelog → use update. troubleshooting → use troubleshoot.
factual · conceptual · procedural · analytical · evaluative · structural · change · narrative · technical
Information type is section-level, identified by the author at runtime per section. It is not a frontmatter field. The veracity standard that applies to each section is derived from its information type:
| Info type | Veracity standard |
|---|
factual, technical, change | very high |
procedural | high |
structural, analytical, evaluative | medium |
conceptual, narrative | lower |
Other frontmatter fields
| Field | Values | Notes |
|---|
veracityStatus | verified · unverified · stale | Set after Phase 6 (veracity pass). |
complexity | beginner · intermediate · advanced | |
lifecycleStage | discover · evaluate · setup · build · operate · troubleshoot · optimise | |
industry | array max 2: technical · finance · economics · business · marketing · governance · broadcast · AI · livepeer | First entry dominates. |
niche | array: web3 · protocol · tokenomics · AI · video · streaming · hardware · infrastructure | |
lastVerified | YYYY-MM-DD | Updated on review pass. |
status | draft · current · published · deprecated · retired | current only after all REVIEW markers cleared. |
The full required minimum frontmatter for every routable page is documented in docs-guide/standards/frontmatter.mdx and enforced (advisory) by the per-page review pipeline.
Voice and copy
UK English throughout. The full canonical rules – banned words, banned phrases, banned constructions, opening order, paragraph discipline, per-audience extensions, heading rules, domain terminology – live in docs-guide/standards/voice-and-copy.mdx.
Working source with full per-audience prohibited-phrase lists: workspace/plan/active/CONTENT-WRITING/Prompts/voice-rules.md.
The 9-stage project pipeline
This pipeline applies to a whole tab (or a section of one). Each stage has a human gate; the next stage cannot start until the prior gate is opened in tab-status.md.
Canonical source: workspace/plan/active/_MY_PROCESS/my-process.mdx. Full per-stage inventory (with usefulness scores per item) lives in _MY_PROCESS/index.md.
| # | Stage | What it produces | Primary tools | Gate artefact |
|---|
| 1 | Personas, Audience & Journey | Primary personas + arriving questions + JTBDs + section structure per tab | Prompts/Prompts-By-Phase/1-audience-design/audience-design-v5.md + pack-guide.md + per-tab veracity-sources.md/glossary-{tab}.md | Phase 1 audience-design output per tab |
| 2 | IA section structuring | Tab-level IA: page hierarchy, predecessor/successor links, persona paths | Prompts/Prompts-By-Phase/2-structure-audit/structure-audit.md + content-scan.md + content-pipeline-tab-map skill | v2/{tab}/_workspace/tab-map.mdx (NOT YET CREATED for any tab – see Gaps) |
| 3 | Current content mapping + gaps | Content inventory per tab, dispositions (KEEP / MOVE / MERGE / SPLIT / REWRITE / DROP), gap report, terminology lock | Prompts/Prompts-By-Phase/2-content-scan/content-scan.md; sub-phases 3.a research, 3.b veracity, 3.c terminology extractor | context-packs/{tab}-content-scan.md (exists for 3 of 5 tabs) |
| 4 | IA pages structuring + journeys | Per-page brief: purpose, sections, info type per section, frontmatter, journey position | Prompts/Prompts-By-Phase/3-content-pass/content-pass.md (REVIEW/WRITE/REWRITE/AUDIT modes) + Frameworks/Frameworks.mdx taxonomy | Approved per-page brief in _workspace/canonical/REVIEW-REGISTRY.md |
| 5 | Find information + place | Content sources mapped to page + section, gap pages identified | content-pipeline-tab-map skill; docs-coverage-and-route-integrity-audit skill; docs-quality-and-freshness-audit skill | Per-tab gap report |
| 6 | Copy writing + content fill | Page draft with content (markdown layer, no MDX components yet) | page-authoring skill; content-pipeline-pass-a skill; Prompts/Prompts-By-Phase/3-content-pass/content-pass.md WRITE mode; docs-copy skill | Page draft saved to _workspace/drafts/ |
| 7 | Review copy against purpose + voice | Quality verdict (PASS / NEEDS WORK / REWRITE REQUIRED) + prioritised fix list | Checks Framework Phases 0–6 (per-page review pipeline) | Review packet from docs-review-packet-generation |
| 8 | Layout + style | Production MDX file with components, frontmatter, template, render-validated | content-pipeline-pass-b skill; Prompts/Prompts-By-Phase/4-layout-pass/layout-pass.md; component-layout-governance skill | Page MDX rendered locally without error |
| 9 | Human review + style refinement | Approved, published page | Checks Framework Phase 10 (human sign-off); rubric-static-review skill; repo-audit-orchestrator skill | Page in v2/ with status: current; gate ✅ in tab-status.md |
Stage 7 = Phases 0–6 of the per-page review pipeline. Stage 8 = Phase 7 of the per-page review pipeline. Stage 9 = Phase 10. The two pipelines interlock at these points.
Locked decisions
| ID | Decision | Applies to | Date | Source |
|---|
| D-NAV-01 | pageType: navigation disambiguation page is a locked cross-tab pattern – one file per tab, shared entry point for all paths, all downstream sections reference it. | All tabs with multiple setup paths (Orchestrators confirmed; applies to Gateways, Developers, any other multi-path tab) | 2026-03-23 | decision-registry.md |
No phase may consume a decision that is not in this registry. Decisions made in chat that are not written here do not exist. Proposals and open items go in blocking-items.md, not the registry.
Current state (as of 2026-05-23)
Per tab-status.md:
| Tab | IA Approved | Terminology Locked | Content Scan Done | Structure Approved | Content Pass Open | Veracity Done | Layout Done |
|---|
| Gateways | ⬜ | ⬜ | ✅ | ⬜ | ⬜ | ⬜ | ⬜ |
| Orchestrators | ⬜ | ⬜ | 🔄 | ⬜ | ⬜ | ⬜ | ⬜ |
| Developers | ⬜ | ⬜ | 🔄 | ⬜ | ⬜ | ⬜ | ⬜ |
| About | ⬜ | ⬜ | ✅ | ⬜ | ⬜ | ⬜ | ⬜ |
| Delegators | ⬜ | ⬜ | ✅ | ⬜ | ⬜ | ⬜ | ⬜ |
Legend: ⬜ not started · 🔄 in progress · ✅ human-approved.
Zero tabs have IA Approved. The project pipeline is stalled at Phase 2 or earlier across the entire docs estate. A tab is only “open” for the next phase when its gate is explicitly unlocked by human sign-off recorded in decision-registry.md. Agents may not proceed past a locked gate.
Supporting infrastructure
The full inventory lives in _MY_PROCESS/index.md – every item per stage with usefulness score, status, and known issues. Summary by category:
AI skills (live as of 2026-05-23)
| Skill | Stage | Status | Lives at |
|---|
page-authoring | 6 (write) | tested, mature | ai-tools/ai-skills/page-authoring/SKILL.md |
content-pipeline-pass-a | 6 (write) + 7 (review content layer) | tested, mature | ai-tools/ai-skills/content-pipeline-pass-a/SKILL.md |
content-pipeline-pass-b | 8 (layout) | tested, mature | ai-tools/ai-skills/content-pipeline-pass-b/SKILL.md |
content-pipeline-tab-map | 2 (IA structuring) + 5 (placement) | known-good | ai-tools/ai-skills/content-pipeline-tab-map/SKILL.md |
docs-copy | 6 (write) | mature; 5-mode routing | ai-tools/ai-skills/docs-copy/SKILL.md |
rubric-static-review | 7 (review) + 9 (final review) | mature | ai-tools/ai-skills/rubric-static-review/SKILL.md |
docs-review-packet-generation | 7 (review) | tested | ai-tools/ai-skills/docs-review-packet-generation/SKILL.md |
docs-review-fix-execution | 7 (review) | tested | ai-tools/ai-skills/docs-review-fix-execution/SKILL.md |
docs-quality-and-freshness-audit | 5 (gaps) + 7 (re-check) | script-backed, production | ai-tools/ai-skills/docs-quality-and-freshness-audit/SKILL.md |
docs-coverage-and-route-integrity-audit | 5 (gaps) | known-good | ai-tools/ai-skills/docs-coverage-and-route-integrity-audit/SKILL.md |
style-and-language-homogenizer-en-gb | 7 (voice) | script-backed, production | ai-tools/ai-skills/style-and-language-homogenizer-en-gb/SKILL.md |
component-layout-governance | 8 (layout) | script-backed, production | ai-tools/ai-skills/component-layout-governance/SKILL.md |
repo-audit-orchestrator | 9 (final review) | mature | ai-tools/ai-skills/repo-audit-orchestrator/SKILL.md |
product-thinking | 9 (final review) | tested | ai-tools/ai-skills/product-thinking/SKILL.md |
Note (2026-05-23): the parent audit of all 34 local SKILL.md files found 18 still carry status: draft with empty test logs (see SLICE-02-ai-tools.md for the live status per skill).
Prompts
| Prompt | Phase | Status |
|---|
audience-design-v5.md | Project pipeline Phase 1 | Tested, known-good |
structure-audit.md + content-scan.md | Project pipeline Phase 2 + 3 | Tested |
content-pass.md (REVIEW/WRITE/REWRITE/AUDIT modes) | Project pipeline Phase 4 + 6 + 7 | REVIEW mode pilot-complete on Gateways; AUDIT mode awaits test runs |
layout-pass.md | Project pipeline Phase 8 | Exists; test status unknown |
veracity-pass.md | Per-page review Phase 6 | DRAFT – requires 1 review agent pass + 3 test runs + human sign-off before production use |
Per-phase pack guides (pre-flight + running order + failure modes) live alongside each prompt at workspace/plan/active/CONTENT-WRITING/Prompts/Prompts-By-Phase/{N-phase}/pack-guide.md.
Canonical frameworks
| File | What it provides |
|---|
docs-guide/standards/frontmatter.mdx | Required + optional frontmatter fields per authority tier |
docs-guide/standards/voice-and-copy.mdx | Voice + copy + heading rules + per-audience extensions |
docs-guide/standards/naming-conventions.mdx | File + folder + workflow + script naming |
docs-guide/frameworks/page-taxonomy-framework.mdx | Retired 2026-05-23 – content consolidated into this page. The retired file is a thin redirect. |
docs-guide/frameworks/page-composition-framework.mdx | Retired 2026-05-23 – content consolidated into Checks Framework Phase 7. |
workspace/plan/active/CONTENT-WRITING/Frameworks/content-pipeline-framework.md | Full taxonomy spec + decisions log (997 lines) |
v2/orchestrators/_workspace/canonical/Frameworks.mdx | Canonical taxonomy reference + section naming rubric + voice + content rules (LOCKED) |
workspace/plan/active/CONTENT-WRITING/Frameworks/veracity.md | Veracity standards per information type (LOCKED 2026-03-23) |
workspace/plan/active/CONTENT-WRITING/Frameworks/veracity-library.md | 45-source authoritative library |
Automated scripts (production-ready)
| Script | Purpose | Stage |
|---|
operations/scripts/audits/content/quality/docs-quality-and-freshness-audit.js | TODO/TBD markers, thin content, freshness risk | 5 (gaps) + 7 (re-check) |
operations/scripts/audits/content/style/style-and-language-homogenizer-en-gb.js | UK English + style consistency | 7 (voice) |
operations/scripts/validators/components/library/component-layout-governance.js | Page composition vs component layout contracts | 8 (layout) |
operations/scripts/validators/content/structure/check-docs-path-sync.js | Internal link resolution | 7 + 8 |
operations/tests/run-all.js | Full test suite | 8 |
Known gaps (2026-05-23, verified)
The pipeline is real but incomplete. Five blockers prevent any tab moving past Phase 2 today.
Tab maps do not exist at the canonical location
The per-page review pipeline (content-pipeline-pass-a skill) requires a tab map at v2/{tab}/_workspace/tab-map.mdx before it can run with full adjacency context. No tab map exists at the canonical location for any tab. The Orchestrators tab map exists in workspace output (workspace/plan/active/CONTENTI-PIPLEINE/03-tab-map.md – note the misspelled folder name) but not at the canonical location.
Unblocks: project pipeline Phase 5 onwards for any tab.
Action: move or symlink the ORCHS tab map into v2/orchestrators/_workspace/tab-map.mdx; produce equivalents for the other 4 tabs.
veracity-pass.md is DRAFT and unvalidated
The veracity pass prompt at workspace/plan/active/CONTENT-WRITING/Prompts/Prompts-By-Phase/7-veracity-pass/veracity-pass.md is explicitly DRAFT in its header and in PROJECT-MANAGEMENT-CANONICAL.md. The pack guide requires 3 test runs with human sign-off before production. No test runs are recorded.
Unblocks: per-page review Phase 6 (veracity pass).
Action: run 3 test passes on pages with governance-controlled values, CLI flag claims, and mixed information types; record results; human sign-off.
Frontmatter validator path is wrong in multiple references
page-authoring/SKILL.md and master-checks.mdx reference tools/config/frontmatter-taxonomy.js and tools/lib/docs/frontmatter-taxonomy.js. Neither path appears in the current tools/ tree. The automated frontmatter validator either does not exist, exists under a different name, or has been retired without updating references.
Unblocks: automated frontmatter validation in the per-page review pipeline.
Action: locate the current frontmatter validator (if any) and update all references, or confirm absence and remove the references from prompts and skill files.
page-taxonomy-framework.mdx and page-composition-framework.mdx are scratch notes
Per the parallel slice audit (SLICE-08-snippets.md and prior session findings), both files sit under docs-guide/frameworks/ but are not real frameworks. page-composition-framework.mdx is a literal MDX template scaffold with placeholder description “Describe page-structure-template”. page-taxonomy-framework.mdx has no frontmatter and contains typo’d field labels.
Action: either rewrite both as real frameworks (sourcing from v2/orchestrators/_workspace/canonical/Frameworks.mdx) or move both to _workspace/ lane.
Three pageType enumerations are in circulation
The canonical 7-type schema (above) is correct. However: docs-guide/frameworks/page-taxonomy-framework.mdx (scratch notes) has different enums; workspace/plan/active/ORCHS/01-CORE-NEEDS-AND-STANDARDS.md frontmatter schema lists deprecated values (landing, faq, troubleshooting as pageType values, not pageVariants); v2/resources/documentation-guide/copy-style/authoring-guide.mdx (public) also lists a subset. Agents following the wrong source will produce wrong frontmatter.
Action: lock one canonical pageType enum on this page; have all other sources transclude or be marked deprecated.
- Checks Framework – the per-page review pipeline that operationalises Stage 7 + 8 + 9 above (11 phases, 9 universal check categories, per-pageType specific checks).
docs-guide/standards/voice-and-copy.mdx – canonical voice + copy + heading rules.
docs-guide/standards/frontmatter.mdx – frontmatter required minimum + per-tier rules.
docs-guide/standards/naming-conventions.mdx – file + folder + script + workflow naming.
docs-guide/policies/source-of-truth-policy.mdx – canonical-source contract for every governed surface.
workspace/plan/active/_MY_PROCESS/ – full 9-stage project pipeline source.
workspace/plan/active/CONTENT-WRITING/ – execution plan, decisions registry, tab status, prompts, frameworks (997 lines + 240 supporting files).
workspace/plan/active/master-checks.mdx – meta-checks framework (882 lines) referenced from the per-tab canonical/checks.mdx files.
v2/{tab}/_workspace/canonical/checks.mdx – per-tab Cat 1-9 quality spec (exists for gateways, orchestrators, about, delegators, developers1 as of 2026-05-23).