rsc-universal 0.1.1

package/skills/medium-publishing/references/cross-post-and-canonical.md

@@ -0,0 +1,65 @@
+# Cross-posting & canonical — deep reference
+
+Detail for the import tool, manual canonical, and the static-HTML workaround for stubborn imports. The SKILL.md body covers the happy path; come here when the import fails or you need the exact UI walkthrough.
+
+## Import tool — step by step
+
+1. Go to **New story -> import icon**, or directly to `https://medium.com/p/import`.
+2. Paste the **original article URL** (the one on your own site).
+3. Click **Import**. Medium crawls the page and creates a draft.
+4. What it sets automatically:
+   - **Canonical link** -> the URL you pasted (your origin keeps SEO credit).
+   - **Publish date** -> backdated, read from the page's `article:published_time` meta tag.
+5. Review the draft for formatting drift (see the formatting gotchas in SKILL.md), set up to 5 tags, then **Publish**.
+
+The import is a one-time content copy plus metadata. Editing the Medium draft afterward does not sync back to your site, and vice versa.
+
+## Manual canonical link — exact UI path
+
+Use this when you drafted natively in Medium, or after a failed import where you pasted the body manually.
+
+1. Open the story in the editor (draft or published).
+2. Click **More options (•••)** in the top bar.
+3. **Story settings** -> the advanced/edit settings panel.
+4. Find **"Customize canonical link"**, paste the **origin URL**, save.
+5. Publish (or update if already published).
+
+This is per-story. There is no account-wide canonical default. The canonical must always point at the *original* location, never at the Medium story itself.
+
+Manual paste does **not** backdate. If the date matters, the import tool is the only path that backdates automatically; otherwise the Medium publish date is "now".
+
+## Static-HTML workaround for stubborn imports
+
+When the importer can't crawl your real page (JS-only rendering, bot wall, login wall, timeout), give it a clean page it *can* read:
+
+1. Export the article to a single static HTML file — content in the initial server markup (no client-side hydration required).
+2. Include both metadata tags the importer relies on:
+
+```html
+<link rel="canonical" href="https://mysite.com/original-post" />
+<meta property="article:published_time" content="2026-04-12T09:00:00Z" />
+```
+
+3. Host it at a public, anonymously reachable URL (a temporary path is fine).
+4. Run the import against **that** URL. Canonical and backdate come from the tags above, so they still point at your real origin and your real date.
+5. Once the Medium draft exists with correct metadata, you can take the temporary static page down.
+
+The canonical in the static page should still be your *real* origin URL, not the temporary static page — that's what search engines will credit.
+
+## HTTP error -> cause -> fix matrix
+
+| HTTP / symptom | Root cause | Fix |
+|---|---|---|
+| 400 | Malformed/redirecting URL, or required page metadata absent | Use the final canonical URL; confirm `<link rel="canonical">` and `article:published_time` present |
+| 404 | Page not reachable for the crawler (private, moved, geo/IP-blocked) | Make it publicly reachable, or use the static-HTML workaround |
+| 403 | Bot-blocking — WAF, Cloudflare challenge, login/paywall | Allowlist the crawler, serve a clean static copy, or paste natively + set canonical manually |
+| 500 | Origin server error during crawl | Retry; if persistent, serve a lightweight static version |
+| 504 | Crawl timeout — heavy JS, slow TTFB | Provide a fast static HTML version with content in initial markup |
+| Imports but body empty/garbled | Content rendered client-side only | Static HTML with the article text in the server-rendered markup |
+| Catch-all "couldn't import" | Any of the above combined | Fall back to native paste + manual "Customize canonical link" |
+
+## Verify a cross-post
+
+- View source / story settings on the published Medium story; confirm canonical resolves to the origin URL.
+- Confirm the displayed publish date matches the original.
+- Confirm <=5 tags, code blocks intact, no broken tables.

package/skills/medium-publishing/references/legacy-api.md

@@ -0,0 +1,100 @@
+# Legacy Medium API — pre-2025-token holders only
+
+> **Closed to new integrations.** As of **2025-01-01** Medium issues no new integration tokens and accepts no new integrations. Only tokens minted before that date still authenticate. Do not attempt to register a new token — it will be refused. This reference is useful **only** if you already hold a pre-2025 token. For everyone else, the publishing path is the web editor + import tool (see `cross-post-and-canonical.md`).
+
+## Auth
+
+All requests use a Bearer integration token:
+
+```
+Authorization: Bearer <MEDIUM_INTEGRATION_TOKEN>
+Content-Type: application/json
+Accept: application/json
+```
+
+Base URL: `https://api.medium.com/v1`.
+
+## Resolve the user id
+
+You need your `userId` before posting. Get it from `/me`:
+
+```bash
+curl -s -H "Authorization: Bearer $MEDIUM_TOKEN" \
+  https://api.medium.com/v1/me
+```
+
+```json
+{
+  "data": {
+    "id": "5303d74c64f66366f00cb9b2a94f3251bf5",
+    "username": "yourhandle",
+    "name": "Your Name",
+    "url": "https://medium.com/@yourhandle",
+    "imageUrl": "https://cdn-images-1.medium.com/..."
+  }
+}
+```
+
+## Create a post on your profile
+
+`POST /v1/users/{userId}/posts`
+
+```bash
+curl -s -X POST \
+  -H "Authorization: Bearer $MEDIUM_TOKEN" \
+  -H "Content-Type: application/json" \
+  https://api.medium.com/v1/users/<userId>/posts \
+  -d '{
+    "title": "My title",
+    "contentFormat": "markdown",
+    "content": "# My title\n\nBody in markdown...",
+    "canonicalUrl": "https://mysite.com/original-post",
+    "tags": ["react", "javascript", "web-development"],
+    "publishStatus": "draft",
+    "license": "all-rights-reserved"
+  }'
+```
+
+### Request fields
+
+| Field | Type | Notes |
+|---|---|---|
+| `title` | string | Story title (also used for the SEO title) |
+| `contentFormat` | string | `html` or `markdown`. Markdown here is more permissive than the web editor |
+| `content` | string | The body, in the chosen format |
+| `canonicalUrl` | string | Set to the **origin** URL for a cross-post — preserves your site's SEO credit |
+| `tags` | string[] | Up to 5; extras are dropped |
+| `publishStatus` | string | `public` \| `draft` \| `unlisted` |
+| `license` | string | e.g. `all-rights-reserved`, `cc-40-by`, `public-domain` |
+| `notifyFollowers` | boolean | Whether followers are notified |
+
+### Response (success)
+
+```json
+{
+  "data": {
+    "id": "e6f36a",
+    "title": "My title",
+    "authorId": "5303d74c64f66366f00cb9b2a94f3251bf5",
+    "url": "https://medium.com/@yourhandle/my-title-e6f36a",
+    "canonicalUrl": "https://mysite.com/original-post",
+    "publishStatus": "draft",
+    "tags": ["react", "javascript", "web-development"],
+    "license": "all-rights-reserved"
+  }
+}
+```
+
+## Create a post in a publication
+
+`POST /v1/publications/{publicationId}/posts` — same body shape.
+
+A `publishStatus: "draft"` under a publication remains **pending** an editor's action; it is not live until an editor publishes it, and you must be an accepted writer for that publication.
+
+## Why automation plans usually fail in 2026
+
+- New tokens: refused (closed since 2025-01-01).
+- Make: the Medium app is legacy — only pre-existing connections work.
+- n8n: Medium credentials still listed but cannot be newly configured.
+
+If the user has no pre-2025 token, route them to the import-tool path. Do not build a pipeline that assumes a new token.

package/skills/medium-strategy/SKILL.md

@@ -0,0 +1,161 @@
+---
+name: medium-strategy
+description: "Use when planning Medium as a growth + money SYSTEM over months, not one post: enrolling in and optimizing for the Partner Program, deciding solo profile vs starting a publication vs joining or pitching one, engineering for Boost and the three distribution tiers, choosing tags/topics for reach, and a canonical cross-posting plan — read from and written back to 02-DOCS/wiki/medium/ so each session compounds. Triggers: 'how does the Medium Partner Program pay me', 'why does my story get views but no distribution', 'how do I get Boosted', 'should I start my own publication or pitch the big ones', 'repurpose my blog onto Medium without tanking my SEO', 'what tags maximize reach', 'qué tags uso en Medium', 'com guanyar diners a Medium'. NOT writing the article's hook/structure/prose (that is medium-writing), NOT the mechanical format-and-submit-one-story step (that is medium-publishing)."
+tags: [medium, partner-program, monetization, distribution, publications, cross-posting]
+recommends: [medium-writing, medium-publishing, content-engine, seo-geo, decision-records]
+origin: risco
+---
+
+# Medium Strategy — The Account as a Payout System, Not the Next Post
+
+*You operate at account + monetization altitude.* You decide how a Medium presence becomes a **growth and payout system** over months: how it earns from the Partner Program, where stories live (your profile, your publication, someone else's), how they win distribution, and what gets repurposed from owned content. You do not draft the article or press the publish button — those are single-story jobs you route out. Mixing altitudes is the fastest way to give generic advice: a portfolio question answered with per-post tactics produces noise.
+
+Your spine is a read/write loop with the account's own history. You **read** accumulated learnings before deciding and **write** the decision back, so next quarter compounds instead of restarting from a generic blog post.
+
+## Altitude boundary — route single-story work out
+
+| If the ask is… | Route to | Why it is not you |
+|---|---|---|
+| The article's hook, structure, story, prose | `medium-writing` | You decide where it goes and why; writing produces the words. |
+| Format one draft, toggle canonical, schedule, submit to a pub's queue | `medium-publishing` | You set the cross-post + publication plan; publishing presses the buttons per story. |
+| A calendar/repurposing plan across all channels | [`../content-engine/SKILL.md`](../content-engine/SKILL.md) | Multi-platform; you are Medium-only. |
+| Push the same post to many networks at once | [`../social-publisher/SKILL.md`](../social-publisher/SKILL.md) | Distribution fan-out, not Medium monetization. |
+| SEO/canonical mechanics on the *owned* site the cross-post points back to | [`../seo-geo/SKILL.md`](../seo-geo/SKILL.md) | Owned-site SEO; you only set the canonical-import decision. |
+| Building a subscriber email list | [`../newsletter/SKILL.md`](../newsletter/SKILL.md) | Medium is not your list. |
+
+## 1. Read the wiki FIRST — before any decision
+
+Before you recommend anything, read `02-DOCS/wiki/medium/`. **Why:** Medium strategy is iterative; last quarter's Boost hits, accepted-pub list, and tag wins are the only non-generic data you have. Skipping this means re-deriving from blog advice the writer could have Googled.
+
+Look for, in order:
+- **Decision log** — prior calls on solo vs publication, which pubs were pitched/accepted/dropped, the tag sets tried.
+- **Boost hit/miss log** — which stories were nominated, which got the extra reach, what they had in common.
+- **Member-reading-time learnings** — which topics/formats drove member *dwell time* (the thing that pays), not just views.
+- **Referral-source notes** — external vs on-platform reads (external reads now pay extra — see §3), and which off-Medium channels sent them.
+- **Dead experiments** — pubs that never replied, tags that pulled nothing. Do not re-run them.
+
+If the directory is empty or absent, **bootstrap it**: create `02-DOCS/wiki/medium/` with a `decisions/` subfolder and a `what-worked.md`, then proceed — this session becomes record #1. Templates and layout: [`references/wiki-records.md`](references/wiki-records.md).
+
+## 2. The money model — get this right or everything downstream is vanity
+
+Partner Program earnings are driven by **member reading time** (read + listen time from paying members, combined). Claps, highlights, and responses are secondary engagement signals, not the payout. Non-members and ads pay you nothing. Minimum payout threshold is **$10**.
+
+- Bad: "Get this to 50k views and 2k claps." Views from logged-out readers and claps from non-members do not move the payout.
+- Good: "Get members to *finish* it — depth, pacing, a reason to keep reading — because member dwell time is the meter."
+
+The 2025-2026 overhaul changed *which dwell time pays most*. These dates are the spine connecting tags, titles, and cross-posting to money:
+
+| Effective | Change | Strategic implication |
+|---|---|---|
+| 2025-10-01 | A member read originating from **outside** Medium earns a **+5% bonus**. | Off-Medium referral (newsletter, social, your blog) is now a payout lever, not just traffic. |
+| 2025-11-01 | **~15% of the Partner Program budget** is allocated to stories members discover via **search engines** (e.g. Google). | SEO-discoverable titles + evergreen topics now monetize directly. |
+| 2026-02 | When a non-member hits your paywall and **joins to read your story**, you earn a small one-time amount. | A genuinely worth-paying-for story can convert readers into members you get credited for. |
+| 2026-06-01 | A qualifying-publication **editor who assigns themselves to a story earns 25%** of that story's earnings; the writer keeps **100%** of their original earnings. | Editors are now financially motivated to recruit and edit good writers — pubs court you (see §6). |
+
+Conclusion to carry forward: **external + SEO referral now pay extra.** That is why the cross-posting plan (§8) and tag/title choices (§7) are monetization levers, not decoration.
+
+## 3. Enrollment — the bar is low; the old gate is gone
+
+The 100-follower requirement was removed in Aug 2023. Current bar — check each before advising "you're ready":
+- [ ] 18 or older.
+- [ ] Published at least **~6 stories**.
+- [ ] Account active **~3 months**.
+- [ ] File taxes / have a bank account in an **eligible country** (80+ supported).
+- [ ] A paid Medium membership is **NOT** required to enroll.
+
+If they clear this, the answer to "can I monetize?" is yes — move to the money model and distribution, not "build your following first."
+
+## 4. Distribution mechanics — three tiers
+
+A story does not just "go to your followers." Medium routes it through three tiers:
+
+- **Network Distribution** — to your followers + readers who follow the story's **tags/topics**.
+- **General Distribution** — broader algorithmic feeds, for any story that passes the distribution guidelines.
+- **Boost** — a curated tier that gives a story extra reach across the platform.
+
+Tags are a primary input the algorithm uses to match a story to interested readers across *all* tiers — which is why §7 treats them as a lever.
+
+**Clearing General** = follow the distribution guidelines (original, non-spammy, not overtly promotional, well-formed). This is table stakes; most decent stories pass.
+
+**Earning Boost** is a two-step, publication-independent path:
+1. A community curator **or** a Boost-eligible publication editor **nominates** the story.
+2. Medium's internal curation team **validates** it against distribution standards before it gets the extra reach.
+
+To qualify it must be **constructive, original, insightful, well-crafted, and memorable**. You do NOT need to be in a publication to be Boosted — but being in a Boost-eligible pub gives you a nomination path you otherwise lack. Depth, disqualifiers, and the eligible-pub landscape: [`references/distribution-and-boost.md`](references/distribution-and-boost.md).
+
+Note the common confusion to correct on sight: **views ≠ distribution.** A story can rack up views from a shared link while getting near-zero algorithmic distribution. The fix is tags + Boost-worthiness, not "post more."
+
+## 5. Publications — solo vs your own vs joining vs pitching
+
+Acceptance into pubs is **craft-gated, not follower-gated** — even a brand-new writer gets in if the story fits the guidelines. Decide deliberately:
+
+| Path | Choose when | Tradeoff |
+|---|---|---|
+| **Solo profile** | Early, or your brand IS the draw, or you cross-post owned content with canonical (§8). | No editor nomination path to Boost; you carry all distribution yourself. |
+| **Your own publication** | You publish consistently and want a branded home + the ability to nominate. | You become the editor; growth is on you; setup + curation overhead. |
+| **Join an existing pub** | You want reach + an editor nomination path now, and have stories fitting its niche. | You publish on their terms and follow their guidelines/queue. |
+| **Pitch an editor** | A specific big pub matches your topic and you can write to its bar. | Slower; requires originals (see tension in §8). |
+
+How to pitch, in order: (1) **read and follow the pub's submission guidelines** — skipping this is the #1 rejection cause; (2) the story must be **100% original / never published anywhere** — Medium rejects duplicate content even when it's your own; (3) it must **not be overtly self-promotional**. The 2026-06 editor-earns-25% change means editors are now paid to find and edit good writers, so a well-targeted pitch lands better than before — lead with fit, not flattery.
+
+## 6. Tags & titles as distribution levers
+
+Up to **5 tags** per story. Use a deliberate mix, not five popular ones:
+- **2-3 high-volume tags** for reach (e.g. `Artificial Intelligence`, `Productivity`).
+- **2-3 long-tail tags** for precise, lower-competition match (e.g. `AI Coding Tools`, `LLM Evaluation`).
+
+- Bad: 5 broad tags → `Technology, AI, Programming, Software, Coding` — you compete with everyone and match no one precisely.
+- Good: `Artificial Intelligence, Programming` (reach) + `AI Coding Tools, Developer Productivity, Claude Code` (precise) — Network distribution finds the exact interested readers.
+
+Titles now carry money: with ~15% of the budget on search-discovered reads (§3), an SEO-discoverable, search-intent title earns from the search budget. Write the title for what a member would *type into Google*, not only for a clever in-feed hook.
+
+## 7. Cross-posting plan — protect owned SEO
+
+To repurpose owned content (blog, newsletter) onto Medium without competing with yourself, use Medium's official **"import a story"** flow: it sets `rel=canonical` back to your original URL, so Google credits the owned site and Medium does not rank as a duplicate.
+
+The tension to resolve every time: **most publications require originals** (§5), so the canonical-import play is for **your own profile/publication**, NOT for pitching pub editors. The decision rule:
+
+- Repurposing owned content for reach + member reads → **canonical-import to your own profile/pub.**
+- Targeting a publication → **write native (original) for that pub.** Do not import a published piece into a pub that requires originals; it gets rejected.
+
+Pair this with §3: external referral now pays +5%, so a canonical import you also drive traffic *to* from your list earns the bonus on top.
+
+## 8. Review / double-down loop
+
+On a monthly/quarterly review:
+1. Read `what-worked.md` and the Boost log.
+2. Pick the **next bet** — double down on the tag set / pub / format that drove member reading time; name the metric it bets on.
+3. **Drop dead pub relationships** — pubs that never replied or never distributed.
+4. Log all of it (next section). A review you don't write down is a review you'll redo.
+
+## 9. Write the decision back to the wiki
+
+Every session ends with a dated record under `02-DOCS/wiki/medium/decisions/`. Minimum fields — the bet-on metric is mandatory, because a decision with no metric is an opinion, not a checkable bet:
+
+```markdown
+# 2026-06-02 — Solo vs publication for AI-tools series
+
+Decision: Stay solo profile + canonical-import the blog series; pitch one Boost-eligible AI pub with a native original.
+Rationale: 400 followers, no editor nomination path; the blog has evergreen search traffic worth importing (external +5%, search 15%).
+Bets on metric: member reading time from external/search referral; 1 Boost nomination this quarter.
+```
+
+After a review, append a `what-worked.md` entry tied to specific stories/pubs/tags (e.g. "`AI Coding Tools` long-tail + Boost → 3.2h member reading time"). Full templates, layout, and the verify.sh contract: [`references/wiki-records.md`](references/wiki-records.md). Sanity-check the wiki with `scripts/verify.sh 02-DOCS/wiki/medium/`.
+
+## Anti-patterns
+
+| Anti-pattern | Why it fails | Do instead |
+|---|---|---|
+| Chasing views and claps | Neither pays; **member reading time** does. | Optimize for member dwell time and finish rate. |
+| Treating views as distribution | A shared link can spike views with zero algorithmic reach. | Fix tags + Boost-worthiness, not post volume. |
+| Importing an already-published piece into a pub that requires originals | Medium rejects duplicate content, even your own. | Native original for pubs; canonical-import only to your own profile/pub. |
+| Five broad tags | You compete with everyone, match no one. | 2-3 popular + 2-3 long-tail. |
+| Pitching without reading the pub's guidelines | The #1 rejection cause. | Read guidelines first; lead with fit. |
+| Ignoring the external/search referral bonus | Leaves the +5% and the 15% search budget on the table. | Drive off-Medium traffic; write search-intent titles. |
+| "Build your following first" before enrolling | The 100-follower gate is gone since 2023. | If they clear the ~6 stories / ~3 months / eligible-country bar, enroll now. |
+| Never writing decisions to the wiki | Next quarter restarts from generic advice. | End every session with a dated decision record. |
+
+## References
+
+- [`references/distribution-and-boost.md`](references/distribution-and-boost.md) — the three tiers in depth, the Boost two-step path, the 5 curation criteria expanded, common distribution disqualifiers, the Boost-eligible publication landscape.
+- [`references/wiki-records.md`](references/wiki-records.md) — decision-record + what-worked templates, the `02-DOCS/wiki/medium/` layout, and exactly what verify.sh checks.

package/skills/medium-strategy/evals/README.md

@@ -0,0 +1,3 @@
+# Evals — medium-strategy
+
+These cases are run by the repo's skill-eval harness, not by hand. `should_trigger` and `should_not_trigger` check **routing**: given each prompt, does the harness select medium-strategy (and for the negatives, does it route to the named real sibling instead). `capability` checks a generated response against the `must_include` rubric for the 90-day-plan scenario — that the answer reads the wiki first, gets the member-reading-time payout model and a 2025-2026 change right, makes the solo-vs-publication call, and writes a dated decision record back. No network or Medium account is needed; the cases are self-contained. Run them through the eval runner as configured for this repo (the same path used for the other `skills/*/evals/cases.yaml`).

package/skills/medium-strategy/evals/cases.yaml

@@ -0,0 +1,50 @@
+skill: medium-strategy
+
+should_trigger:
+  - prompt: "How does the Medium Partner Program actually pay me?"
+    why: Core monetization question — the member-reading-time payout model is this skill's spine.
+  - prompt: "Should I start my own Medium publication or pitch the big ones?"
+    why: The solo-vs-own-pub-vs-pitch decision is an account-altitude call this skill owns.
+  - prompt: "My Medium stories get views but almost no distribution — how do I get Boosted?"
+    why: Distribution/Boost mechanics; non-obvious because the writer conflates views with distribution, which this skill corrects.
+  - prompt: "I want to repurpose my blog posts onto Medium without tanking my own SEO."
+    why: Canonical-import cross-posting plan; non-obvious framing (the SEO risk is the real question, not formatting).
+  - prompt: "com puc fer que els meus articles a Medium arribin a més gent i guanyin diners"
+    why: Catalan trigger covering growth + monetization at the account level.
+  - prompt: "What tags should I put on my Medium AI article to maximize reach?"
+    why: Tags as a distribution lever (popular + long-tail mix), an account-level reach decision.
+  - prompt: "Quarterly review of my Medium account — what's working and where should I double down?"
+    why: The review/double-down loop reading the wiki — distinctly portfolio altitude, not per-post.
+
+should_not_trigger:
+  - prompt: "Write the intro hook and structure for my Medium essay on burnout."
+    route_to: medium-writing
+    why: Per-article craft — producing the words, not deciding where the piece goes.
+  - prompt: "Format this draft and submit it to the Better Programming queue."
+    route_to: medium-publishing
+    why: The mechanical per-story publish/submit step, not the portfolio plan.
+  - prompt: "Build me a content calendar across Medium, LinkedIn, and my newsletter."
+    route_to: content-engine
+    why: Multi-platform calendar/repurposing; this skill is Medium-only.
+  - prompt: "Set the canonical tag and SEO meta on my own blog post."
+    route_to: seo-geo
+    why: Owned-site SEO mechanics; this skill only sets the canonical-import decision, not the site config.
+  - prompt: "Push this same post out to all my social channels at once."
+    route_to: social-publisher
+    why: Multi-network distribution fan-out, not Medium monetization strategy.
+
+capability:
+  - scenario: >
+      A writer with 12 published Medium stories, ~400 followers, getting views but
+      almost no distribution and earning ~$8/month, asks for a 90-day plan to grow
+      and monetize on Medium.
+    must_include:
+      - Reads 02-DOCS/wiki/medium/ first (or bootstraps it if empty/absent) before recommending.
+      - Explains earnings are driven by member reading time (not views/claps) and notes the $10 payout threshold.
+      - Names at least one 2025-2026 change (external +5%, 15% search budget, paywall-join bonus, or editor 25%) WITH its strategic implication.
+      - Gives a solo-vs-publication recommendation with rationale (incl. the Boost nomination-path tradeoff).
+      - A concrete Boost/distribution play — the two-step curator nomination and the 5 quality criteria.
+      - A tag/title strategy — 2-3 popular + 2-3 long-tail tags, and a search-intent title for the search budget.
+      - A cross-posting decision — canonical import to own profile vs native originals for pub pitches.
+      - Corrects the views-equals-distribution confusion.
+      - Writes a dated decision record back to 02-DOCS/wiki/medium/ with a decision, rationale, and a bet-on metric.

package/skills/medium-strategy/references/distribution-and-boost.md

@@ -0,0 +1,65 @@
+# Distribution & Boost — the depth behind SKILL.md §4
+
+Load this when a writer's core problem is reach: "views but no distribution," "how do I get Boosted," or "what disqualifies a story."
+
+## The three tiers, in detail
+
+Medium does not push every story equally. A story is routed through up to three tiers, gated at each step.
+
+### Network Distribution
+The baseline. A published story reaches:
+- People who **follow you**.
+- People who **follow the story's tags/topics**.
+
+This is why a story with no relevant tags reaches almost no one beyond direct followers — there is no topic surface for the network to match it to. Tags are the lever here (SKILL.md §6).
+
+### General Distribution
+Broader algorithmic feeds (the home feed, topic pages, "for you" surfaces). A story is eligible if it **passes the distribution guidelines** — original, not spam, not overtly promotional, well-formed, follows the rules. Most competently written stories clear this. Failing it is usually a content-quality or self-promotion problem, not bad luck.
+
+### Boost
+A **curated** tier that grants extra reach across the platform — the closest thing to amplification. It is not automatic and not buyable.
+
+## The Boost two-step path
+
+1. **Nomination.** A story is nominated by either:
+   - a **community curator**, or
+   - an **editor of a Boost-eligible publication** (this is the concrete benefit of being in such a pub).
+2. **Validation.** Medium's internal **curation team** reviews the nominated story against distribution standards before it receives the boosted reach.
+
+Both steps must pass. A nomination is necessary, not sufficient.
+
+**Publication-independent:** you can be Boosted from your own solo profile — but you then depend on a community curator finding you, with no editor nomination path. That asymmetry is a real input to the solo-vs-pub decision (SKILL.md §5).
+
+## The 5 Boost qualification criteria, expanded
+
+A story must be all five. Use these as a pre-submission checklist:
+
+| Criterion | What it actually means | Failure smell |
+|---|---|---|
+| **Constructive** | Adds something usable — a method, an argument, a resolution. | Vague hot-take with no takeaway. |
+| **Original** | Your own thinking and words; not rehashed, not AI-slop, not previously published. | Reads like a summary of other posts. |
+| **Insightful** | A non-obvious angle or hard-won lesson, not a list everyone already knows. | "10 tips" with the obvious 10. |
+| **Well-crafted** | Clear structure, clean prose, good pacing, formatted to read. | Wall of text, broken flow. |
+| **Memorable** | A reader could describe it to someone a week later. | Forgettable; blends into the feed. |
+
+If a story misses one of these, route the fix to `medium-writing` — that is craft, not strategy.
+
+## Common distribution disqualifiers
+
+Stories get held out of General/Boost distribution most often for:
+- **Overtly self-promotional** content (the post exists to sell, not to help).
+- **Affiliate-link-heavy** or thin "content marketing" pieces.
+- **Previously published / duplicate** content without proper canonical handling.
+- **Misleading titles or clickbait** that the body doesn't deliver on.
+- **Guideline violations** (spam, low-effort, policy breaches).
+
+Diagnosing "no distribution": first check the story passes guidelines (General), then check tags (Network match), then assess Boost-worthiness against the 5 criteria. Do not jump to "post more."
+
+## Boost-eligible publication landscape
+
+Medium runs a **Boost Nomination Program** with 100+ participating publications whose editors can nominate. The strategic takeaway is not the specific list (it changes) but the pattern: **joining a Boost-eligible publication in your niche buys you a nomination path** you cannot get from a cold solo profile. When advising solo-vs-pub, weigh this concretely — if Boost reach is the goal and the writer has no curator relationships, an eligible pub is the faster path.
+
+## Sources
+- Medium Help Center, "Medium's Distribution Guidelines," help.medium.com/hc/en-us/articles/360006362473 (accessed 2026-06-02).
+- Medium Help Center, "Browse 100+ publications in the Boost Nomination Program," help.medium.com/hc/en-us/articles/23964242497559 (accessed 2026-06-02).
+- mediumcourse.com Boost guide (accessed 2026-06-02).

package/skills/medium-strategy/references/wiki-records.md

@@ -0,0 +1,60 @@
+# Wiki records — layout, templates, and the verify.sh contract
+
+This is the persistence layer that makes medium-strategy compound. Read SKILL.md §1 and §9 first; this file is the exact format.
+
+## Layout under `02-DOCS/wiki/medium/`
+
+```text
+02-DOCS/wiki/medium/
+├── decisions/                # one file per strategy decision (append, never rewrite history)
+│   ├── 2026-06-02-solo-vs-pub-ai-series.md
+│   └── 2026-09-01-q3-review-double-down.md
+├── what-worked.md            # running log of outcomes tied to stories/pubs/tags
+├── pubs.md                   # accepted / pitched / dropped publications + their guidelines notes
+└── tags.md                   # tag sets tried + which pulled member reading time
+```
+
+`decisions/` per-file is the primary form. A single running `decisions.md` at the root is also accepted by verify.sh if a writer prefers one log file.
+
+## Decision record template
+
+Every decision file MUST carry three signals: a **date**, a **Decision:** line, and a **named metric/bet**. The metric is mandatory — a decision with no metric is an opinion, not something you can check next quarter.
+
+```markdown
+# 2026-06-02 — Solo vs publication for the AI-tools series
+
+Decision: Stay solo profile; canonical-import the 6 evergreen blog posts; pitch one Boost-eligible AI publication with a new native original.
+Rationale: ~400 followers and no editor nomination path; the owned blog already has search traffic worth importing (external read +5%, search budget 15%). Pitching one pub buys a Boost nomination path without giving up the solo home.
+Bets on metric: member reading time from external + search referral over 90 days; ≥1 Boost nomination this quarter.
+Status: open
+```
+
+When the bet resolves, append a short outcome to the same file (`Status: won — Boost nomination 2026-07; +2.1h member reading time`) rather than editing the original — append-only history.
+
+## what-worked.md entry template
+
+```markdown
+## 2026-09-01 — Q3 review
+- `AI Coding Tools` (long-tail) + Boost → 3.2h member reading time on one story. Double down: keep the long-tail tag, pitch the same pub again.
+- Imported blog post w/ canonical + newsletter push → +5% external bonus visible, but low member conversion. Keep importing; stop expecting member reads from cold list.
+- Pitched "Big Generic Pub" twice, no reply → DEAD. Drop it.
+```
+
+## pubs.md / tags.md
+
+Keep these lightweight. `pubs.md`: pub name | status (accepted/pitched/dropped) | niche fit | a one-line guideline note. `tags.md`: tag | popular-or-long-tail | did it pull member reading time?
+
+## The verify.sh contract
+
+`scripts/verify.sh` is a **structural lint, not a craft judge.** Run it after a session:
+
+```bash
+./scripts/verify.sh 02-DOCS/wiki/medium/
+```
+
+It checks, read-only:
+1. The target directory exists.
+2. At least one decision record is present (`decisions/*.md` or a root `decisions.md`).
+3. Each record carries the three required signals: a date (`YYYY-MM-DD`), a `Decision:` line, and a named metric/bet (`Bets on metric:` / `metric` / `bet`).
+
+A missing metric/bet is a **hard failure** — by design. A missing or empty target (a fresh account with no wiki yet) is reported and exits **0** — no false failure. It judges structure only; whether the *strategy* is good is the capability eval's job.

package/skills/medium-strategy/scripts/verify.sh

@@ -0,0 +1,118 @@
+#!/usr/bin/env bash
+#
+# verify.sh — structural lint for a medium-strategy wiki under 02-DOCS/wiki/medium/.
+#
+# WHAT IT DOES (read-only; never edits a file)
+#   Static, network-free checks on the decision records the skill writes back.
+#     1. The target directory exists.
+#     2. At least one decision record is present: a file under decisions/*.md
+#        OR a running decisions.md log at the target root.
+#     3. Each decision record carries the three required signals:
+#          - a date              (a YYYY-MM-DD anywhere in the file)
+#          - a Decision: line     (the concrete choice)
+#          - a named metric/bet   (a "Bets on metric:" / "metric" / "bet" field)
+#        A record missing the metric/bet field is a HARD FAILURE — a decision
+#        with no metric is an opinion, not a checkable bet.
+#
+#   Missing date or Decision line is also a hard failure. A missing/empty target
+#   (a fresh account with no wiki yet) is reported and exits 0 — no false failure.
+#   Pure bash + grep, no dependencies. It is a lint, not a strategy oracle.
+#
+# HOW TO RUN
+#   ./verify.sh 02-DOCS/wiki/medium/    # lint an account's strategy wiki
+#   ./verify.sh                         # no target -> nothing to check, exit 0
+#
+# EXIT CODES
+#   0  clean, or nothing to check
+#   1  a hard structural failure
+#   2  bad usage
+#
+# Runs on stock macOS bash 3.2.
+
+set -euo pipefail
+
+if [ -t 1 ]; then
+  RED=$'\033[31m'; GREEN=$'\033[32m'; YELLOW=$'\033[33m'; NC=$'\033[0m'
+else
+  RED=''; GREEN=''; YELLOW=''; NC=''
+fi
+
+ok_count=0; warn_count=0; fail_count=0
+ok()   { printf '%s[ ok ]%s %s\n' "$GREEN"  "$NC" "$*"; ok_count=$((ok_count + 1)); }
+warn() { printf '%s[warn]%s %s\n' "$YELLOW" "$NC" "$*"; warn_count=$((warn_count + 1)); }
+fail() { printf '%s[fail]%s %s\n' "$RED"    "$NC" "$*"; fail_count=$((fail_count + 1)); }
+
+usage() { sed -n '2,34p' "$0" | sed 's/^# \{0,1\}//'; }
+
+summary_exit() {
+  printf '\nok=%d warn=%d fail=%d\n' "$ok_count" "$warn_count" "$fail_count"
+  [ "$fail_count" -gt 0 ] && exit 1
+  exit 0
+}
+
+# --- arg parse --------------------------------------------------------------
+TARGET=""
+while [ $# -gt 0 ]; do
+  case "$1" in
+    -h|--help) usage; exit 0 ;;
+    -*) printf 'unknown flag: %s\n' "$1" >&2; usage >&2; exit 2 ;;
+    *) TARGET="$1"; shift ;;
+  esac
+done
+
+# No target, or missing dir -> nothing to check, clean exit (no false fail).
+if [ -z "$TARGET" ]; then
+  ok "no wiki path given — nothing to lint"
+  summary_exit
+fi
+if [ ! -d "$TARGET" ]; then
+  warn "directory not found: $TARGET — nothing to lint (fresh account?)"
+  summary_exit
+fi
+ok "wiki directory exists: $TARGET"
+
+# --- collect decision records ----------------------------------------------
+# Per-decision files under decisions/, plus an optional running decisions.md log.
+RECORDS=""
+if [ -d "$TARGET/decisions" ]; then
+  for f in "$TARGET"/decisions/*.md; do
+    [ -f "$f" ] && RECORDS="$RECORDS$f
+"
+  done
+fi
+[ -f "$TARGET/decisions.md" ] && RECORDS="$RECORDS$TARGET/decisions.md
+"
+
+# Trim blank lines / handle empty.
+RECORDS="$(printf '%s' "$RECORDS" | sed '/^$/d')"
+
+if [ -z "$RECORDS" ]; then
+  warn "no decision records yet (decisions/*.md or decisions.md) — nothing to lint"
+  summary_exit
+fi
+
+# --- per-record structural checks ------------------------------------------
+DATE_RE='[0-9]{4}-[0-9]{2}-[0-9]{2}'
+
+while IFS= read -r rec; do
+  [ -z "$rec" ] && continue
+  if [ ! -s "$rec" ]; then
+    fail "$rec: empty record"
+    continue
+  fi
+
+  miss=""
+  grep -Eq  "$DATE_RE"                       "$rec" || miss="$miss date"
+  grep -Eiq '^[[:space:]-]*Decision:'        "$rec" || miss="$miss decision-line"
+  grep -Eiq '(bets on metric:|metric|bet:)'  "$rec" || miss="$miss metric/bet"
+
+  if [ -n "$miss" ]; then
+    fail "$rec: missing required field(s):$miss"
+  else
+    ok "$rec: date + decision + metric present"
+  fi
+done <<EOF
+$RECORDS
+EOF
+
+summary_exit