@sellable/mcp 0.1.268 → 0.1.271

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sellable/mcp",
-  "version": "0.1.268",
+  "version": "0.1.271",
   "type": "module",
   "description": "Sellable MCP server for Claude Code and Codex campaign workflows",
   "main": "dist/index.js",

package/skills/create-post/SKILL.md

@@ -55,8 +55,8 @@ Hard fail patterns:
 - drafts that skip raw idea capture
 - drafts that skip premise development
 - hooks copied verbatim from another creator
-- drafts that clone a living creator's voice instead of adapting their public
-  hook/body mechanics into the user's voice
+- final drafts that claim to be authored by a living creator, copy outside
+  wording, or borrow another creator's personal proof/status as the user's proof
 - using `~/.sellable/configs/content/linkedin-posts-drafts.md` as the normal save target
   </role>
 
@@ -234,7 +234,7 @@ Use when the user gives an existing idea ID.
 1. Call `mcp__sellable__get_post_idea({ ideaId })`.
 2. Treat the idea's raw source as immutable source material.
 3. Run hook research.
-4. Run the configured thought-leader inspiration lab unless the user explicitly
+4. Run the configured thought-leader voice variant lab unless the user explicitly
    says to skip thought leaders or use only a named subset.
 5. Develop and select a premise card.
 6. Save hook research.
@@ -252,7 +252,7 @@ Use when the user gives a new rough idea, voice memo transcript, freestyle note,
 4. Distill only what the user actually said plus source-backed transcript
    worldview/proof ingredients. Label anything inferred from transcripts.
 5. Run hook research.
-6. Run the configured thought-leader inspiration lab unless the user explicitly
+6. Run the configured thought-leader voice variant lab unless the user explicitly
    says to skip thought leaders or use only a named subset.
 7. Develop and select a premise card.
 8. Save hook research.
@@ -318,7 +318,7 @@ Space benchmark research:
 
 The approved pack is capped at 20 gold standards. If adding new approved examples would exceed 20, ask which existing item to replace or skip the overflow. Candidate lists can be longer, but approved saved standards cannot exceed 20.
 
-## Thought Leader Inspiration Mode
+## Thought Leader Voice Variant Mode
 
 Use when the user names creators, thought leaders, profile URLs, or says to
 write variants using a specific person's recent best content as inspiration.
@@ -329,9 +329,41 @@ the configured active influencer list after the Transcript Worldview Packet and
 general hook research, unless the user explicitly says to skip thought leaders,
 skip external inspiration, or use only a named subset.
 
-This mode does not clone the person's voice. It extracts public, reusable
-mechanics from their recent posts and adapts those mechanics into the user's
-voice, proof, story, and product truth.
+This mode is a role-play variant lab for internal synthesis. Each background
+worker should assume that creator's public LinkedIn lens, taste, hook style,
+rhythm, and editorial instincts, then draft the post the creator would likely
+write from the user's idea and context. The worker may write in a voice-inspired
+style for the private variant, but it must not claim the creator authored it,
+copy source wording, copy personal proof, or invent facts about the user.
+
+The final orchestrator does not publish a collage of creator personas. It takes
+the best hooks, pressure, structure, rhythm, and proof moves from the role-played
+variants, then synthesizes the most compelling post for Christian/Sellable.
+
+Target fast flow:
+
+```text
++----------------------+      +----------------------+      +----------------------+
+| Raw idea + memory    | ---> | Active influencers   | ---> | Parallel workers     |
+| 3-5 min              |      | 1 min                |      | 6-10 min wall clock  |
++----------------------+      +----------------------+      +----------+-----------+
+                                                                  |
+                                                                  v
++----------------------+      +----------------------+      +----------------------+
+| Save compact draft   | <--- | Final synthesis      | <--- | Voice-modeled posts  |
+| 1 min                |      | 3-5 min              |      | representative length|
++----------------------+      +----------------------+      +----------------------+
+```
+
+Default output budget:
+
+- main orchestrator receipt: 300-700 words
+- each influencer worker: representative-length post plus compact notes
+- each influencer post variant: match that creator's normal length, rhythm,
+  vocabulary, formatting, and close for this kind of idea
+- final synthesized post: normal LinkedIn post length unless the user asks short
+- do not return long research reports unless the user explicitly asks for them;
+  spend tokens on the actual voice-modeled post
 
 Default workflow:
 
@@ -345,42 +377,63 @@ Default workflow:
    tracked people or memory, ask for the profile URL or handle before running
    person-specific research.
 3. When the host supports background agents, launch one bounded
-   `thought-leader-worker` per person. If not, process each person sequentially.
-   Each configured person must receive full person-specific research; do not
-   collapse the configured list into generic space research.
+   `influencer-voice-worker` per person. If not, process each person
+   sequentially. Each configured person must receive a person-specific voice
+   variant; do not collapse the configured list into generic space research.
 4. Each worker fetches that person's recent posts with
    `mcp__sellable__fetch_linkedin_posts`, and fetches profile/follower context
    with `mcp__sellable__fetch_linkedin_profile` when available.
-5. Each worker scores that person's recent posts by topic fit, hook strength,
-   body payoff, repeatability, engagement quality, follower-adjusted signal
-   when follower counts are available, and adaptation fit for the user's idea.
-6. Each worker returns a compact inspiration packet:
-   - best recent posts, max 5
-   - rejected posts, max 5
-   - weighted score and normalization confidence
-   - first-3-line hook previews
-   - hook mechanics
-   - body mechanics
-   - sentence/rhythm moves
-   - proof or status that cannot transfer
-   - allowed adaptation moves
-   - one adapted hook/body direction in the user's voice
-7. The orchestrator compares all creator packets, chooses 1-3 useful
-   inspiration adapters, and creates post variants from the current selected
-   premise and outline.
-8. The orchestrator must label each variant by `inspiration_adapter`, not
-   "written in <person>'s voice."
+5. Each worker samples enough recent posts to understand the person's current
+   public voice. Score internally for topic fit, hook strength, body payoff,
+   repeated patterns, engagement quality, and follower-adjusted signal when
+   follower counts are available. Do not return the full scoring table.
+6. Each worker must build a `voice_model` before drafting:
+   - representative length range for similar posts
+   - average paragraph/line length and spacing rhythm
+   - hook posture and opening moves
+   - common vocabulary, recurring phrases, and taboo phrases
+   - proof style: story, numbers, teardown, checklist, confession, challenge
+   - formatting: bullets, numbering, questions, equations, parentheticals
+   - close style: command, reflection, CTA, punchline, open loop
+   - factual boundaries: what must come from Christian/Sellable context
+7. Each worker writes the complete post that this creator would likely write
+   from the user's idea and context. Match the inferred length/rhythm/style; do
+   not compress to a generic short draft.
+8. Each worker returns a compact voice variant:
+   - person and profile URL
+   - worker status
+   - recent post count sampled
+   - voice model, max 8 bullets
+   - complete post variant in that creator's inferred voice and representative
+     length
+   - hook options, max 3
+   - what to steal for the final Christian/Sellable post, max 5 bullets
+   - what to drop or avoid, max 5 bullets
+   - proof gaps and factual risks
+   - source URLs sampled, max 3
+9. The orchestrator compares all creator variants, chooses the strongest hooks,
+   structures, proof moves, and rhythm, then writes one synthesized final post.
+10. The orchestrator must label worker outputs as `voice_variant_from: <person>`
+   and final synthesis as `final_voice: Christian/Sellable`.
+11. In the final user-visible response, include a compact
+   `BACKGROUND_AGENT_DRAFT_REVIEW` section before the final post whenever this
+   lab ran. It must review every active person's private variant, state the
+   strongest move, what to drop, and what was selected for synthesis. Do not
+   hide this behind a boolean receipt when the user is validating the workflow.
 
 Hard rules:
 
-- Do not impersonate, clone, or claim to write as the thought leader.
+- The role-play variant is internal synthesis material, not a claim of authorship
+  by the creator.
 - Do not copy outside wording, proof, jokes, personal stories, or status.
+- Do not invent user proof just because it would fit the creator's style.
 - Do not let a large-audience creator win only because of reach. Use engagement
   per 1k followers when follower counts are available, and mark confidence when
   follower data is missing.
-- Keep Christian/Sellable voice as the final voice layer.
+- Keep Christian/Sellable truth and proof as the final factual layer.
 - If a thought leader has no recent full-text posts or the useful posts are
-  mostly lead magnets, save the packet as `weak` or `reject`.
+  mostly lead magnets, return a weak variant and explain what could still be
+  useful.
 
 ## Visible Whole-Flow Debug Mode
 
@@ -533,16 +586,16 @@ Visible Flow Trace
 - score:
 - combined body plan:
 
-14. Thought Leader Inspiration Lab
+14. Thought Leader Voice Variant Lab
 - specified thought leaders:
 - worker status per person:
-- recent posts reviewed:
-- weighted winners:
+- recent posts sampled:
+- voice model:
+- complete voice variants:
 - follower-adjusted signal when available:
-- reusable hook/body mechanics:
-- forbidden borrowing:
-- adapted variants in user's voice:
-- selected inspiration adapter:
+- hook/body moves worth stealing:
+- proof gaps and factual risks:
+- selected synthesis ingredients:
 
 15. Draft
 - draft body:
@@ -656,7 +709,7 @@ The research worker must return a compact packet only:
 - hook-to-body promise maps that show how each hook tells the body
 - body structures and exact body language moves
 - preview measurements
-- thought leader inspiration packets from the configured active influencer list
+- thought leader voice variants from the configured active influencer list
   unless the user explicitly opted out or supplied a named subset
 - track-person and gold-standard recommendations
 - blocked states or confidence gaps
@@ -705,7 +758,7 @@ Record provenance:
 - source hook preview measurements and whether they came from full text or a search preview
 - selected hook patterns
 - audience tension snapshot and selected angle
-- thought leader inspiration packets and selected inspiration adapters from the
+- thought leader voice variants and selected synthesis ingredients from the
   configured active influencer list unless explicitly skipped
 - premise cards and selected premise
 - exact phrase patterns and sentence shapes
@@ -818,15 +871,14 @@ Body structures learned:
    - exact language moves:
    - adapted Sellable body move:
 
-Thought leader inspiration packets:
+Thought leader voice variants:
 1. person + profile URL
    - worker status:
-   - recent posts reviewed:
-   - weighted winners:
-   - follower-adjusted signal:
-   - hook/body mechanics:
-   - forbidden borrowing:
-   - adapted variant direction in Christian/Sellable voice:
+   - recent posts sampled:
+   - voice model:
+   - full role-played post variant:
+   - steal/drop notes:
+   - factual risk:
 
 Rejected examples:
 - author/source:
@@ -1033,7 +1085,7 @@ selecting it. A selected hook may carry a `warn` only when the warning is about
 intentional blank-line rhythm or a slight line-length overage; include a compact
 fallback in the validation receipt.
 
-## Step 2.5: Thought Leader Inspiration Lab
+## Step 2.5: Thought Leader Voice Variant Lab
 
 Run this step by default for draft-producing create-post flows using the
 configured active people in `discovery/influencers.md`. Skip it only when the
@@ -1041,40 +1093,104 @@ user explicitly says to skip thought leaders, skip external inspiration, or use
 only general hook research. If the user supplied a `thought_leader_list`, use
 that named subset instead of the full configured list.
 
-If the host supports background agents, start one bounded
-`thought-leader-worker` per specified person. Each worker owns only that
-person's recent content analysis. The orchestrator owns comparison, adaptation,
-and final voice safety.
+This is a draft-readiness gate, not an optional research note. Before final
+prose, resolve the active configured influencer list, record the expected person
+count, and produce one voice variant for every active person. A draft
+cannot be `ready` when the configured list was not loaded, when an active person
+is missing a variant, or when the lab is summarized as generic hook research. If
+a profile fetch or post fetch succeeds but produces weak material, keep a
+`weak` variant with the attempted fetches and confidence gaps. If a profile URL
+cannot be resolved or the research tool fails, return `retry-needed` or save
+only `needs_revision` unless the user explicitly opted out of the lab.
+
+When the host supports background agents, fire one bounded
+`influencer-voice-worker` per specified person. This is the default execution
+mode. Do not process the configured influencer list sequentially just to keep
+the main orchestrator simple. Sequential processing is allowed only when the
+host has no background-agent mechanism; record that fallback in the receipt.
+Each worker owns only that person's voice role-play variant. The orchestrator
+owns comparison, synthesis, factual safety, and final Christian/Sellable proof.
 
 Each worker must return:
 
 - person name and profile URL or handle
-- recent post count reviewed
-- best recent posts, max 5
-- rejected posts, max 5
+- worker status
+- recent post count sampled
 - follower count when available
-- engagement per 1k followers when available
-- weighted winner score and normalization confidence
-- first-3-line hook previews
-- hook mechanics
-- body mechanics
-- rhythm and sentence moves
-- proof/status that cannot transfer
-- allowed adaptation moves
-- one adapted direction for the user's selected premise
-
-The orchestrator then creates 3-5 inspiration variants from the selected
-premise and outline. Variants must be labeled as:
+- normalization confidence
+- voice model, max 8 bullets, including representative length, rhythm,
+  vocabulary, formatting, proof style, and close style
+- complete role-played post variant in the creator's inferred voice and
+  representative length
+- hook options, max 3
+- best lines or moves worth stealing, max 5 bullets
+- what to drop or avoid, max 5 bullets
+- proof gaps and factual risks
+- source URLs sampled, max 3
+
+Do not return full research tables, long post autopsies, or full source text
+unless the user explicitly asks for the research report. The normal worker job is
+to draft the variant.
+
+The orchestrator then synthesizes the final post from the voice variants.
+Variant labels must be:
 
 ```text
-inspiration_adapter: <person or source mechanic>
-voice_layer: Christian/Sellable
-borrowed: hook mechanism | body sequence | proof order | rhythm | close move
-not_borrowed: source wording | source proof | source personal story | persona
+voice_variant_from: <person>
+role_play_basis: public posts sampled + configured Reason lane
+post_variant: <complete variant>
+steal_for_final: hook | structure | proof order | rhythm | close move
+drop_for_final: copied wording | borrowed personal proof | fake user fact
 ```
 
-Do not write "in <person>'s voice." Write "using <person>'s public hook/body
-mechanics as inspiration, in Christian/Sellable voice."
+It is acceptable for the worker to write the private variant "in <person>'s
+voice" for synthesis. It is not acceptable to publish the variant as that
+person, copy their wording, or borrow their personal proof/status. The final
+draft should be the most compelling synthesis, grounded in Christian/Sellable
+context.
+
+Before drafting, record:
+
+- `activeInfluencerSource`: `discovery/influencers.md` or named subset
+- `expectedActiveInfluencerCount`
+- `actualVoiceVariantCount`
+- `missingVoiceVariants`
+- `weakVoiceVariants`
+- `selectedSynthesisIngredients`
+- `voiceVariantsGeneratedBeforeFinalProse: yes | no`
+- `workerExecutionMode: background_agents | sequential_fallback`
+- `backgroundWorkerIds` or worker labels when available
+
+If `actualVoiceVariantCount` is lower than
+`expectedActiveInfluencerCount`, do not save the draft as `ready`.
+
+When returning the completed draft, show the synthesis review in this shape
+before the final post:
+
+```text
+BACKGROUND_AGENT_DRAFT_REVIEW
+workerExecutionMode: background_agents | sequential_fallback
+expectedActiveInfluencerCount: <number>
+actualVoiceVariantCount: <number>
+
+1. <person>
+- worker_status:
+- draft_review:
+- strongest_move:
+- weakness_or_drop:
+- synthesis_decision:
+
+SYNTHESIS_INGREDIENTS
+- hook:
+- structure:
+- rhythm:
+- proof_order:
+- close:
+- dropped:
+```
+
+The review must cover every active configured person, not only the winning
+variant. Then return `FINAL_POST` in Christian/Sellable voice.
 
 ## Step 3: Draft
 
@@ -1095,11 +1211,16 @@ Draft from:
 - story/proof files
 - post writing rules
 - 1-3 relevant approved gold standards when available
-- thought leader inspiration variants from the configured active influencer
+- thought leader voice variants from the configured active influencer
   list or supplied named subset
 
 If a claim cannot be traced to the raw idea, core memory, or user answer in the current session, remove it or ask.
 
+Do not write final draft prose until the Thought Leader Voice Variant Lab has
+either completed for every active configured influencer or the validation
+receipt has an explicit user opt-out reason. A light source-research summary
+does not satisfy this gate.
+
 ## Step 4: Validation
 
 Use `references/post-validation.md`.
@@ -1116,7 +1237,8 @@ Every saved draft needs a validation receipt with:
 - selected hook and why
 - pre-draft narrative outline
 - selected source template and no-copy adaptation rationale
-- thought leader inspiration lab, including skipped/opt-out reason when omitted
+- thought leader voice variant lab, including skipped/opt-out reason when
+  omitted and expected-vs-actual variant counts for the active configured list
 - post positioning breakdown
 - viral-post outline
 - hook-to-body promise map
@@ -1226,7 +1348,9 @@ selected_premise: <premise or none>
 selected_hook: <hook>
 pre_draft_narrative_outline: <compact I/A/i outline summary or none>
 selected_source_template: <template name/source or none>
-thought_leader_inspiration: <adapters used or none>
+thought_leader_voice_variants: <variants and synthesis ingredients used or none>
+background_agent_draft_review: <per-person review and synthesis decision when the lab ran>
+final_post: <complete final post in Christian/Sellable voice>
 visible_flow_trace: <required when user asked for whole-flow/debug/step-by-step mode; include checkpoint statuses and quality break>
 validation_summary:
   premise_value: pass | needs_user_input | needs_revision

package/skills/create-post/references/hook-research-playbook.md

@@ -7,7 +7,7 @@ readers click "see more," decompose each post into a reusable narrative and
 positioning template, and then adapt those templates to the user's real story
 without copying source wording.
 
-V2 research has five outputs:
+V2 research has six outputs:
 
 1. weighted source winners: the best recent posts to learn from
 2. hook autopsies: exact preview measurements, open loops, and tension created
@@ -16,7 +16,7 @@ V2 research has five outputs:
 4. viral-post outlines: the reusable narrative structure of each source post
 5. post positioning breakdown templates: line-level positioning and narrative
    technique maps that can be adapted into the user's draft
-6. thought leader inspiration packets from the configured active influencer
+6. thought leader voice variants from the configured active influencer
    list, unless the user explicitly skipped thought leaders or supplied a
    named subset
 
@@ -56,7 +56,7 @@ Worker owns:
 - line-to-template conversion
 - body-structure extraction
 - body-expression input extraction for the later draft lab
-- per-creator inspiration packet assembly when the user supplied thought
+- per-creator voice variant assembly when the user supplied thought
   leaders
 - rejected-example notes
 - track-person and gold-standard recommendations
@@ -88,7 +88,7 @@ Research packet:
 - preview measurements
 - hook-to-body promise maps
 - body expression inputs
-- thought leader inspiration packets
+- thought leader voice variants
 - confidence gaps
 - save recommendations
 ```
@@ -698,9 +698,9 @@ with Christian's verified campaign-source ladder. Keep the hook's "unexpected
 best source" tension. Do not borrow the source's dinner/community proof.
 ```
 
-## Thought Leader Inspiration Workers
+## Thought Leader Voice Variant Workers
 
-Run a separate inspiration analysis per person for the configured active
+Run a separate voice-variant worker per person for the configured active
 thought-leader list by default during draft-producing create-post flows, unless
 the user explicitly says to skip thought leaders, skip external inspiration, or
 use only a named subset. Also do this when the user supplies a list of thought
@@ -710,9 +710,31 @@ memory-backed `discovery/influencers.md`; include only rows where
 person's lane/adaptation brief. This is different from general search: the
 source set is the person's own recent content.
 
+This is a hard drafting dependency. General hook research, space benchmark
+research, or a single combined "influencer takeaways" paragraph does not satisfy
+the thought-leader step. The research artifact must record the active
+configured list, expected active person count, actual voice variant count,
+missing variants, weak variants, and selected synthesis ingredients. Ready
+drafts require one voice variant per active configured person unless the user
+explicitly opted out or supplied a smaller named subset.
+
 Use one bounded background worker per person when the host supports background
-agents. If not, process the list sequentially. Each worker must keep its output
-compact and must not return a full corpus dump.
+agents. This is required for Codex/Claude hosts that expose background agents or
+Task workers. If the host does not expose a worker mechanism, process the list
+sequentially and record `workerExecutionMode: sequential_fallback` plus the
+reason. Each worker must keep its output compact and must not return a full
+corpus dump or long research report.
+
+Runtime budget:
+
+```text
+per worker target: representative-length post + compact notes
+post variant target: match the creator's normal length for this type of idea
+source samples returned: max 3 URLs
+hook options returned: max 3
+steal/drop notes: max 5 bullets each
+normal wall-clock: one background wave, 6-10 minutes total for 6-10 people
+```
 
 Worker steps:
 
@@ -722,7 +744,7 @@ Worker steps:
 2. Fetch recent posts with `mcp__sellable__fetch_linkedin_posts`.
 3. Fetch profile/follower context with `mcp__sellable__fetch_linkedin_profile`
    when available.
-4. Score the person's recent posts by:
+4. Sample and score the person's recent posts internally by:
    - topic fit to the user's current idea
    - hook preview strength
    - body payoff strength
@@ -732,54 +754,61 @@ Worker steps:
    - engagement per 1k followers when follower count is available
    - adaptation fit for Christian/Sellable voice
    - lead-magnet, giveaway, and status-only penalties
-5. Keep max 5 winners and max 5 rejected examples.
-6. Extract hook mechanics, body mechanics, sentence/rhythm moves, proof slots,
-   and close moves.
-7. Identify forbidden borrowing: source wording, source proof, jokes, personal
-   stories, audience status, and persona markers.
-8. Create one adapted direction for the user's selected premise in
-   Christian/Sellable voice.
-
-Each configured person must receive a full person-specific packet. Do not
+5. Build a `voice_model` before drafting. It must infer:
+   - representative post length range for similar ideas
+   - paragraph and line rhythm
+   - hook posture and opening moves
+   - vocabulary and recurring phrases
+   - formatting habits: bullets, numbering, equations, parentheticals, spacing
+   - proof style: story, metric, teardown, confession, field guide, challenge
+   - close style: command, reflection, CTA, punchline, or open loop
+   - factual boundaries that must come from Christian/Sellable
+6. Assume and role-play the person's public LinkedIn voice for a private
+   synthesis variant. This means adopting their likely length, vocabulary,
+   hook tension, pacing, section shape, directness, humor level, proof style,
+   formatting, and editorial instincts.
+7. Write the complete post that this creator would likely write from the user's
+   selected premise and context. Do not compress it into a short generic draft.
+8. Identify what the final synthesis should steal and what it must drop.
+9. Keep factual claims grounded in the user's context. Do not invent user proof.
+
+Each configured person must receive a full person-specific variant. Do not
 summarize the group as a single "GTM influencers" research pass. The later
 drafting stage should be able to compare how Alex, Zayd, Tas, Anthony, or any
-other configured person would structurally pressure-test the same Christian
-idea without copying their voice, proof, jokes, or persona.
+other configured person would actually write the same Christian idea, then
+combine the strongest pieces.
+
+If a person's recent posts are weak or off-topic, still return a packet with
+`worker_status: weak`, the reviewed post count, a weaker voice variant if
+possible, confidence gaps, and a clear reason the person should not strongly
+influence the final synthesis. If the profile or posts cannot be fetched because
+of a tool failure, record `worker_status: retry_needed` and do not allow a
+`ready` draft.
 
 Worker output:
 
 ```text
-Thought leader inspiration packet:
+Thought leader voice variant:
 person:
 profile_url:
-recent_posts_reviewed:
+worker_id_or_label:
+worker_status: pass | weak | retry_needed | blocked
+recent_posts_sampled:
 follower_count: <number | unavailable>
 normalization_confidence: high | medium | low
-weighted_winners:
-  - source_url:
-    engagement:
-    engagement_per_1k_followers:
-    score:
-    why_kept:
-    first_3_line_preview:
-    hook_mechanics:
-    body_mechanics:
-    sentence_or_rhythm_moves:
-    forbidden_borrowing:
-adapted_direction:
-  inspiration_adapter:
-  voice_layer: Christian/Sellable
-  hook_shape:
-  body_shape:
-  proof_slots_needed:
-  why_it_might_win:
-  why_it_might_fail:
-rejected_examples:
+voice_model: [max 8 bullets covering length, rhythm, vocabulary, formatting, proof style, close]
+hook_options: [max 3]
+post_variant: <complete post in the creator's inferred voice and representative length>
+steal_for_final: [max 5 bullets]
+drop_for_final: [max 5 bullets]
+source_urls_sampled: [max 3]
 confidence_gaps:
 ```
 
-Do not call the adapted direction "in <person>'s voice." It is an inspiration
-adapter that borrows public mechanics, then writes in the user's voice.
+The worker may call the private draft "in <person>'s voice" because that is the
+role-play job. The orchestrator must not claim the creator authored it, must not
+copy exact outside wording into the final, and must ground final claims in
+Christian/Sellable context.
 
 ## Blocked States
 

package/skills/create-post/references/post-file-contract.md

@@ -74,10 +74,13 @@ Hook research files must preserve:
 - rendered preview records for kept source hooks and adapted hook blocks,
   including literal mobile/desktop preview blocks and whether the first-screen
   promise is visible
-- thought leader inspiration packets from the configured active influencer list
-  or supplied named subset, including worker status, profile URL/handle, recent
-  posts reviewed, weighted winners, follower-adjusted signal when available,
-  allowed mechanics to adapt, and forbidden borrowing
+- thought leader voice variants from the configured active influencer list or
+  supplied named subset, including worker status, profile URL/handle, recent
+  posts sampled, inferred voice model, representative-length role-played post
+  variant, source URLs sampled, steal/drop notes, factual risk notes,
+  `workerExecutionMode`, expected active influencer count, actual variant count,
+  missing variants, weak variants, and background worker IDs or labels when
+  available
 - source-message outlines for keeper posts, preserving source paragraph order
   and branching each paragraph, line, or phrase into high-level goal, reader
   effect, reusable move, and adaptation guard without reproducing the full
@@ -104,9 +107,11 @@ Draft files must preserve:
 - draft body
 - source-message outline for any outside post being adapted, when the draft's
   body shape was learned from a specific source post
-- thought leader inspiration lab by default, including the selected inspiration
-  adapter, variants considered, Christian/Sellable voice layer,
-  copying/voice-clone risk, and skipped/opt-out reason when omitted
+- thought leader voice variant lab by default, including voice models, variants
+  considered, selected synthesis ingredients, final Christian/Sellable voice,
+  style-simulation risk, active configured influencer list, expected-vs-actual
+  variant counts, `workerExecutionMode`, background worker IDs or labels, and
+  skipped/opt-out reason when omitted
 - pre-draft narrative outline: hierarchical `I.`, `A.`, `i.` outline covering
   hook debt, thesis, reader, core mechanism, definitions, proof claims,
   working body patterns adapted, narrative beats, mobile scan path, body

package/skills/create-post/references/post-validation.md

@@ -334,31 +334,59 @@ If `selectedAngle` is missing, if the audience tension is only a generic label
 like "founders want growth," or if `credibleWhyUs` depends on borrowed proof
 from another creator, save as `needs_revision`.
 
-## Thought Leader Inspiration Audit
+## Thought Leader Voice Variant Audit
 
 For draft-producing create-post flows, validate the configured thought-leader
-inspiration lab by default. If it was skipped, record the explicit opt-out
-reason. When thought leaders or creators were used, validate that inspiration
-was used as structural adaptation, not persona cloning.
+voice variant lab by default. If it was skipped, record the explicit opt-out
+reason. When thought leaders or creators were used, validate that background
+workers produced complete role-play variants before final prose was synthesized.
+
+This audit is a ready-status gate. It is not enough to mention that creators
+were reviewed. The receipt must prove that the configured influencer workflow
+ran before prose was drafted.
 
 Record:
 
+- `activeInfluencerSource`: `discovery/influencers.md` | named subset | user
+  opt-out
+- `expectedActiveInfluencerCount`
+- `actualVoiceVariantCount`
+- `missingVoiceVariants`
+- `weakVoiceVariants`
+- `workerExecutionMode`: `background_agents` | `sequential_fallback`
+- `backgroundWorkerIds`: worker IDs or labels when available
 - `thoughtLeaderList`: names and profile URLs or handles
-- `workerPackets`: one packet per person, or `not_available` with reason
-- `weightedWinners`: best posts used, with follower-adjusted signal when
-  available
+- `voiceModels`: inferred length, rhythm, vocabulary, formatting, proof style,
+  and close style per person
+- `voiceVariants`: one representative-length role-played post per person, or
+  `not_available` with reason
 - `normalizationConfidence`: high | medium | low
-- `inspirationAdapters`: hook/body mechanics selected for adaptation
-- `voiceLayer`: must be `Christian/Sellable`
-- `forbiddenBorrowingAvoided`: source wording, proof, jokes, personal stories,
-  status, and persona markers avoided
-- `variantsConsidered`: adapted variants generated from the selected premise
-- `selectedVariant`: winning adapter and why
-- `copyingOrVoiceCloneRisk`: pass | needs_revision | blocked
-
-Save as `needs_revision` when a draft imitates a living person's voice, borrows
-their proof/status, copies outside wording, or lets a creator with a giant
-audience win without any follower-adjusted or confidence-adjusted signal.
+- `selectedSynthesisIngredients`: hooks, body moves, proof order, rhythm, close
+- `finalVoice`: usually `Christian/Sellable`
+- `forbiddenBorrowingAvoided`: exact outside wording, proof, jokes, personal
+  stories, status, and fake user facts avoided
+- `voiceVariantsGeneratedBeforeFinalProse`: yes | no
+- `selectedSynthesis`: what won and why
+- `styleSimulationRisk`: pass | needs_revision | blocked
+
+Save as `needs_revision` when a final draft copies exact outside wording,
+borrows another creator's proof/status as if it were the user's, invents user
+facts, or lets a creator with a giant audience win without any
+confidence-adjusted reason.
+
+Save as `needs_revision` when `actualVoiceVariantCount` is lower than
+`expectedActiveInfluencerCount`, when the active configured list was not loaded,
+when variants are replaced by a generic influencer summary, or when voice
+variants were generated after final prose instead of before final prose. Return
+`retry-needed` when a tool/package/search failure prevents fetching configured
+profiles or posts.
+Only mark `thought_leader_adaptation: not_used` when the user explicitly opted
+out before drafting.
+
+Save as `needs_revision` when the host supports background agents but the
+configured influencers were processed sequentially without a recorded fallback
+reason. The default expected value is `workerExecutionMode:
+background_agents`.
 
 ## LinkedIn Preview Audit