API changes and talks about scoring models now

.claude-plugin/marketplace.json

@@ -6,21 +6,21 @@
         "email": "[email protected]"
     },
     "metadata": {
-        "description": "Claude Code plugins for scoring, evaluating, and improving content with resonance models on u22a8.ai",
+        "description": "Claude Code plugins for scoring, evaluating, and improving content with scoring models on u22a8.ai",
         "pluginRoot": "./plugins"
     },
     "plugins": [
         {
             "name": "u22a8",
             "source": "./plugins/u22a8",
-            "description": "Score and compare content against information resonance models on u22a8.ai",
+            "description": "Score and compare content against scoring models on u22a8.ai",
             "version": "0.1.0",
             "category": "content-quality",
             "tags": [
                 "scoring",
                 "writing",
                 "content-quality",
-                "resonance-models"
+                "scoring-models"
             ]
         }
     ]

README.md

@@ -1,8 +1,8 @@
 # U+22A8 plugins
 
-Claude Code plugin marketplace for [u22a8.ai](https://u22a8.ai) — score and improve content with resonance models.
+Claude Code plugin marketplace for [u22a8.ai](https://u22a8.ai) — score and improve content with scoring models.
 
-> A resonance model is a standard of judgment, learned from examples, that can score any content — turning subjective quality into something measurable.
+> A scoring model is a standard of judgment, learned from examples, that can score any content — turning subjective quality into something measurable.
 
 ## Quick start █░░
 
@@ -18,7 +18,7 @@ Claude Code plugin marketplace for [u22a8.ai](https://u22a8.ai) — score and im
 
 ### u22a8
 
-Score and improve content against resonance models on u22a8.ai.
+Score and improve content against scoring models on u22a8.ai.
 
 Two skills:
 

actions/score-docs/README.md

@@ -38,7 +38,7 @@ That's it. The action posts a score comment on every PR that changes a markdown
     # Model to score against. Default: u22a8.compelling-readme
     # Options: u22a8.compelling-readme, u22a8.technical-writing,
     #          u22a8.developer-landing-page, u22a8.puns
-    profile: u22a8.technical-writing
+    model: u22a8.technical-writing
 
     # Flag files that score below this threshold (0-100).
     # Informational only — does not block merge.
@@ -95,7 +95,7 @@ For tighter iteration — score a draft, improve it, rescore — the [U+22A8 Cla
 
 Most "AI quality" tools use an LLM to judge output. That means different scores on the same content on different runs. You can't set a threshold you can trust.
 
-U+22A8 scores are deterministic — same content, same profile, same score every time. That's what makes a threshold meaningful in a CI context.
+U+22A8 scores are deterministic — same content, same model, same score every time. That's what makes a threshold meaningful in a CI context.
 
 U+22A8 also outperforms LLM-as-judge on correlation with human judgment, at a fraction of the cost and latency. [Details at u22a8.ai.](https://u22a8.ai)
 

actions/score-docs/action.yml

@@ -1,13 +1,13 @@
 name: "U+22A8 Score Docs"
-description: "Score markdown files in PRs against U+22A8 resonance models. Posts per-trait quality scores as a PR comment."
+description: "Score markdown files in PRs against U+22A8 scoring models. Posts per-trait quality scores as a PR comment."
 author: "u22a8.ai"
 
 branding:
     icon: "check-circle"
     color: "green"
 
 inputs:
-    profile:
+    model:
         description: "Model handle to score against"
         required: false
         default: "u22a8.compelling-readme"
@@ -40,7 +40,7 @@ runs:
           shell: bash
           env:
               GH_TOKEN: ${{ inputs.github-token || github.token }}
-              U22A8_PROFILE: ${{ inputs.profile }}
+              U22A8_MODEL: ${{ inputs.model }}
               U22A8_THRESHOLD: ${{ inputs.threshold }}
               U22A8_FILES: ${{ inputs.files }}
               U22A8_API: "https://u22a8.ai"

actions/score-docs/examples/u22a8-score.yml

@@ -1,6 +1,6 @@
 # .github/workflows/u22a8-score.yml
 #
-# Score markdown files on every PR against a U+22A8 resonance model.
+# Score markdown files on every PR against a U+22A8 scoring model.
 # Posts a comment with per-trait quality scores.
 #
 # More info: https://u22a8.ai
@@ -21,5 +21,5 @@ jobs:
             - uses: actions/checkout@v4
             - uses: u22a8/plugins/actions/score-docs@main
               # with:
-              #   profile: u22a8.compelling-readme    # default
+              #   model: u22a8.compelling-readme     # default
               #   threshold: 60                      # flag files below this score

actions/score-docs/score-docs.sh

@@ -2,12 +2,12 @@
 set -euo pipefail
 
 # U+22A8 Score Docs — GitHub Action
-# Scores markdown files changed in a PR against a U+22A8 resonance model.
+# Scores markdown files changed in a PR against a U+22A8 scoring model.
 # Posts (or updates) a PR comment with per-trait scores.
 
 MARKER="<!-- u22a8-score -->"
 API="${U22A8_API:-https://u22a8.ai}"
-PROFILE="${U22A8_PROFILE:-u22a8.compelling-readme}"
+MODEL="${U22A8_MODEL:-u22a8.compelling-readme}"
 THRESHOLD="${U22A8_THRESHOLD:-}"
 REPO="${GITHUB_REPOSITORY:-}"
 
@@ -49,7 +49,7 @@ if [[ ${#MD_FILES[@]} -eq 0 ]]; then
     exit 0
 fi
 
-echo "Found ${#MD_FILES[@]} markdown file(s) to score against ${PROFILE}."
+echo "Found ${#MD_FILES[@]} markdown file(s) to score against ${MODEL}."
 
 # Score each file and build comment body
 COMMENT_BODY="${MARKER}"$'\n'
@@ -64,7 +64,7 @@ for file in "${MD_FILES[@]}"; do
     CONTENT=$(cat "$file")
 
     # Call U+22A8 API with tracking headers
-    RESPONSE=$(curl -sf "${API}/p/${PROFILE}" \
+    RESPONSE=$(curl -sf "${API}/m/${MODEL}" \
         -H "Content-Type: application/json" \
         -H "Accept: application/json" \
         -H "User-Agent: u22a8-score-docs/1.0 (GitHub Action)" \
@@ -84,7 +84,7 @@ for file in "${MD_FILES[@]}"; do
     fi
 
     # Get plain-text formatted output for the comment
-    PLAIN_RESPONSE=$(curl -sf "${API}/p/${PROFILE}" \
+    PLAIN_RESPONSE=$(curl -sf "${API}/m/${MODEL}" \
         -H "Content-Type: application/json" \
         -H "Accept: text/plain" \
         -H "User-Agent: u22a8-score-docs/1.0 (GitHub Action)" \
@@ -94,7 +94,7 @@ for file in "${MD_FILES[@]}"; do
         2>/dev/null) || PLAIN_RESPONSE=""
 
     # Build file section
-    COMMENT_BODY+="\`${file}\` scored against \`${PROFILE}\` — **${COMPOSITE}/100**"$'\n\n'
+    COMMENT_BODY+="\`${file}\` scored against \`${MODEL}\` — **${COMPOSITE}/100**"$'\n\n'
 
     if [[ -n "$PLAIN_RESPONSE" ]]; then
         COMMENT_BODY+='```'$'\n'

plugins/u22a8/.claude-plugin/plugin.json

@@ -1,12 +1,12 @@
 {
     "name": "u22a8",
-    "description": "Score and compare content against resonance models on u22a8.ai",
+    "description": "Score and compare content against scoring models on u22a8.ai",
     "version": "0.1.0",
     "author": {
         "name": "u22a8.ai"
     },
     "homepage": "https://u22a8.ai",
     "repository": "https://github.com/u22a8/plugins",
     "license": "Apache-2.0",
-    "keywords": ["scoring", "writing", "content-quality", "resonance-models"]
+    "keywords": ["scoring", "writing", "content-quality", "scoring-models"]
 }

plugins/u22a8/skills/evaluate/SKILL.md

@@ -1,22 +1,22 @@
 ---
 name: evaluate
-description: Score content against resonance model traits. Use when the user wants to measure how content performs on specific traits.
+description: Score content against scoring model traits. Use when the user wants to measure how content performs on specific traits.
 user-invocable: true
 allowed-tools: Read, Glob, Bash(u22a8 MCP tools)
 argument-hint: [file-or-text]
 ---
 
 # Evaluate Content
 
-A resonance model is a standard of judgment, learned from examples, that can score any content. Each model has traits — distinct dimensions of quality — and scores content 0–100 per trait. Scores are deterministic and instant.
+A scoring model is a standard of judgment, learned from examples, that can score any content. Each model has traits — distinct dimensions of quality — and scores content 0–100 per trait. Scores are deterministic and instant.
 
 Score content against specific traits. Measurement only — no edits.
 
 ## Core Flow
 
 1. **Identify content to score.** Read from a file path ($ARGUMENTS), a URL, a selection, or conversation context. URLs (http/https) can be passed directly to `score` — they are fetched and text-extracted automatically. If unclear, ask the user what content to evaluate.
 
-2. **Determine the model.** If the user specifies a model, use it. Otherwise, call `list_profiles` to see what's available and either pick the most relevant one based on context (e.g. a README → `u22a8.compelling-readme`) or present the options and let the user choose.
+2. **Determine the model.** If the user specifies a model, use it. Otherwise, call `list_models` to see what's available and either pick the most relevant one based on context (e.g. a README → `u22a8.compelling-readme`) or present the options and let the user choose.
 
 3. **Identify traits.** Use `list_traits` to show the model's traits. Suggest relevant ones based on context, but confirm the user's choice. Do NOT silently evaluate all traits unless the user explicitly asks for "all".
 

plugins/u22a8/skills/improve/SKILL.md

@@ -1,20 +1,20 @@
 ---
 name: improve
-description: Evaluate content, identify weaknesses, make targeted edits, and re-evaluate to confirm improvement. Use when the user wants to improve content against resonance model traits.
+description: Evaluate content, identify weaknesses, make targeted edits, and re-evaluate to confirm improvement. Use when the user wants to improve content against scoring model traits.
 user-invocable: true
 allowed-tools: Read, Write, Edit, Glob, Bash(u22a8 MCP tools)
 argument-hint: [file-or-text]
 ---
 
 # Improve Content
 
-A resonance model is a standard of judgment, learned from examples, that can score any content. Each model has traits — distinct dimensions of quality — and scores content 0–100 per trait. Scores are deterministic and instant.
+A scoring model is a standard of judgment, learned from examples, that can score any content. Each model has traits — distinct dimensions of quality — and scores content 0–100 per trait. Scores are deterministic and instant.
 
 Evaluate content, analyze weaknesses, make targeted edits, and re-evaluate to confirm improvement.
 
 ## Core Flow
 
-1. **Identify content and model.** Read content from file ($ARGUMENTS), URL, or conversation. URLs passed to `score` are fetched automatically. If the user specifies a model, use it. Otherwise, call `list_profiles` and pick the most relevant one based on context, or let the user choose. Then confirm which traits to optimize for using `list_traits`.
+1. **Identify content and model.** Read content from file ($ARGUMENTS), URL, or conversation. URLs passed to `score` are fetched automatically. If the user specifies a model, use it. Otherwise, call `list_models` and pick the most relevant one based on context, or let the user choose. Then confirm which traits to optimize for using `list_traits`.
 
 2. **Baseline evaluation.** Call `score` with the current content. Record per-trait scores, zones, and headroom. Present the baseline — highlight which traits have the most headroom (furthest from "solid" threshold).
 
@@ -71,7 +71,7 @@ Evaluate content, analyze weaknesses, make targeted edits, and re-evaluate to co
 
 User: "Improve my blog post for flow and storytelling"
 
-1. No model specified and content is a blog post — call `list_profiles`, pick the most fitting model or ask the user
+1. No model specified and content is a blog post — call `list_models`, pick the most fitting model or ask the user
 2. Baseline: Flow=45 (Developing, headroom 12), Storytelling=62 (Solid), Composite=53
 3. Analyze: "Flow has the most headroom. The third paragraph jumps abruptly from personal anecdote to technical explanation without transition."
 4. Edit: Add transitional sentences, restructure the opening to establish the narrative thread