From 2b2c187e3e362cc1c4bd4f81d5ec4a0d3ea09e23 Mon Sep 17 00:00:00 2001 From: Kyle Sexton <153232337+kyle-sexton@users.noreply.github.com> Date: Thu, 30 Jul 2026 12:26:28 -0400 Subject: [PATCH] chore(knowledge): retire models-explained blog doc-queue entry after slice completion The claude-models-explained blog slice completed the full pipeline: rendered-channel fetch (raw-md confirmed absent) with firecrawl extraction, 7-unit digest fan-out executing the run-7 pairing cross-link contract (a)-(g), dual verification with two correction rounds (dominant class: assertion-specific vendor-attestation under-application), REVERIFY2: PASS from both verifiers with no degraded fallback, and interview handoff (8 CAs + 8 OQs + pairing results, all 32 digest questions mapped). The queue entry leaves the profile per the remove-on-completion rule, taking its inlined pairing contract with it (executed; results live in the slice's handoff). The vendor-blog attestation rule the context-engineering slice's handoff recommended (OQ-3) is codified in the profile, matching the form both verifiers enforced on this slice: assertion-specific `vendor-claimed (blog, fetch)` markers with an illustrative assertion-class list. Knowledge plugin 0.10.8 -> 0.10.9 with changelog. Co-Authored-By: Claude Fable 5 --- plugins/knowledge/.claude-plugin/plugin.json | 2 +- plugins/knowledge/CHANGELOG.md | 19 ++++++++++++ .../context/anthropic-docs-profile.md | 29 +++++++------------ 3 files changed, 31 insertions(+), 19 deletions(-) diff --git a/plugins/knowledge/.claude-plugin/plugin.json b/plugins/knowledge/.claude-plugin/plugin.json index faab55ab7..101d6ce21 100644 --- a/plugins/knowledge/.claude-plugin/plugin.json +++ b/plugins/knowledge/.claude-plugin/plugin.json @@ -1,7 +1,7 @@ { "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json", "name": "knowledge", - "version": "0.10.8", + "version": "0.10.9", "description": "Ingest external knowledge into durable, synthesized artifacts. Ships a book-distillation pipeline (PDF/EPUB into concept-organized, author-attributed skill reference files), a YouTube pipeline (watch, transcript, link harvest, and repo-applicability synthesis), a course-digest pipeline (extract and synthesize online video courses — Dometrain, Teachable — into repo-applicable recommendations), and a docpage-digest pipeline (single online documentation page into a verified knowledge slice with dual verification — one cross-vendor verifier — and an interview handoff), plus a re-runnable setup action; a configurable library directory governs where synthesized artifacts land in the consuming repo.", "author": { "name": "Melodic Software", diff --git a/plugins/knowledge/CHANGELOG.md b/plugins/knowledge/CHANGELOG.md index d217c790e..caeada24c 100644 --- a/plugins/knowledge/CHANGELOG.md +++ b/plugins/knowledge/CHANGELOG.md @@ -4,6 +4,25 @@ All notable changes to the `knowledge` plugin are recorded here. The `version` i `.claude-plugin/plugin.json` is the delivery vehicle — a consumer receives a change only after that version increases. +## [0.10.9] + +### Changed + +- **`docpage-digest` Anthropic profile: models-explained blog queue entry removed.** The + claude-models-explained blog slice completed — rendered-channel fetch (raw-md confirmed + absent, matching the profile's blog-post channel note) with firecrawl extraction, through + interview handoff, dual verification (two correction rounds, re-verified REVERIFY2: PASS by + both verifiers, no degraded fallback) — so its entry leaves the doc queue per the queue's + remove-on-completion rule, taking its inlined seven-item pairing cross-link contract with it + (the contract was executed by the slice; its results live in the slice's handoff). +- **`docpage-digest` Anthropic profile: vendor-blog attestation rule.** Blog-only assertions + (behavioral, performance, figure/percentage, comparative, positioning — an illustrative, not + exhaustive, list) carry an assertion-specific `vendor-claimed (blog, fetch)` + marker beside their vocabulary tag — never satisfied by related-property citations, never + co-occurring with a same-assertion live-doc citation, never deferred to the interview. Closes + the rule gap the context-engineering blog slice's handoff flagged (its OQ-3), with the shape + enforced end-to-end by both verifiers on this slice. + ## [0.10.8] ### Changed diff --git a/plugins/knowledge/skills/docpage-digest/context/anthropic-docs-profile.md b/plugins/knowledge/skills/docpage-digest/context/anthropic-docs-profile.md index d2ef35d26..f263486d9 100644 --- a/plugins/knowledge/skills/docpage-digest/context/anthropic-docs-profile.md +++ b/plugins/knowledge/skills/docpage-digest/context/anthropic-docs-profile.md @@ -48,6 +48,17 @@ the tag asserts: to `mixed`, and third-party APIs (e.g. the GitHub API) count as API surfaces — no vendor exemption. (From the best-practices slice: the third-party-API ruling is its cross-vendor finding; the own-basis rule was applied there and ratified by both re-verifications.) +- **Vendor-blog attestation:** a `claude.com/blog` page is marketing-adjacent vendor voice, not + reference documentation. Any assertion of fact that exists ONLY in the blog (no harness or + platform doc states the same assertion) — behavioral, performance, figure/percentage, + comparative, frequency, methodological/definitional, positioning, or any other class; the + list is illustrative, not exhaustive — additionally carries + `vendor-claimed (blog, fetch)` beside its vocabulary tag — assertion-specific + (related-property citations never exempt it), never co-occurring with a live-doc citation for + the same assertion, and never deferred to the interview. The marker is an attestation note + that composes with the tag and, where applicability itself is inferred, with + `unverified-inference`. (Shape recommended by the context-engineering blog slice's handoff, + exercised end-to-end and enforced by both verifiers on the models-explained slice.) ## Digest-agent model matching @@ -86,24 +97,6 @@ Supplementary references: Blog posts: -- - — pairs with the completed choosing-a-model doc digest (routing vet executed 2026-07-29, - melodic-software/claude-code-plugins#1697). Cross-link contract for this run (portable copy - of that slice's pairing package; the fuller original also lives in the slice's interview - handoff, work-root slug `platform-claude-com-docs-en-abo-38ebff0f`, where that machine-local - memory-tier slice still exists): compare the blog against the choosing-a-model doc on - (a) the capabilities/speed/cost triad framing, also cross-linking the other model-choice - blog ("Choosing a Claude model and effort level in Claude Code") that - code.claude.com/docs/en/model-config.md links for harness model-fit guidance; - (b) whether Effort is a third consideration or a fourth factor (the considerations-count - split); (c) the selection-factors frame, the effort-vs-model-switch lever, and any - per-model effort recommendations; (d) the two-approach framing, Opus 5 / Fable 5 / - Mythos 5 positioning language, and pricing figures; (e) per-model fit sections vs the - doc's matrix rows (source lines 71–74); (f) the upgrade/switch decision's eval-first - four steps; (g) overlap with the models-overview cards, including Sonnet 5's "The best - combination of speed and intelligence" tagline. Divergences between the channels escalate - as the blog slice's own open questions; agreements are recorded once, pointer each way, - never restated in both. - — pairs with any verification-loop or loop-engineering work the consuming setup already tracks