File size: 8,779 Bytes
36afb17
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
<!doctype html>
<html lang="de">
  <head>
    <meta charset="utf-8">
    <meta name="viewport" content="width=device-width, initial-scale=1">
    <title>PromptBoost V2 Adapter</title>
    <meta
      name="description"
      content="PromptBoost V2 operationalisiert einen User-Prompt als harness-neutralen, eval-faehigen Prompt-Entwurf."
    >
    <link rel="stylesheet" href="/src/styles.css">
  </head>
  <body>
    <header class="topbar">
      <a class="brand" href="#top" aria-label="PromptBoost V2 Start">
        <span class="brand-mark" aria-hidden="true">PB</span>
        <span>
          <strong>PromptBoost V2</strong>
          <small>User Intent -> Harness Draft</small>
        </span>
      </a>
      <nav class="topnav" aria-label="Hauptnavigation">
        <a href="#adapter">Adapter</a>
        <a href="#score">Score</a>
        <a href="#eval-lab">Eval Lab</a>
        <a href="#export">Export</a>
        <a href="#sources">Quellen</a>
      </nav>
    </header>

    <main id="top">
      <section id="adapter" class="workspace" aria-labelledby="app-title">
        <div class="workspace-head">
          <div>
            <p class="eyebrow">Phase 3 Adapter</p>
            <h1 id="app-title">Aus User-Intent wird ein Harness-Prompt-Entwurf.</h1>
          </div>
          <p>
            V2 trennt beobachteten User-Intent, Adapter-Scaffolding, blanke Harness Policy,
            Model Adapter und Eval Plan. Der Adapter veraendert nicht die Aufgabe, sondern
            operationalisiert sie als pruefbaren, portablen Entwurf.
          </p>
        </div>

        <div class="app-grid">
          <section class="panel input-panel" aria-labelledby="input-title">
            <div class="panel-head">
              <div>
                <p class="panel-kicker">Input</p>
                <h2 id="input-title">User Prompt</h2>
              </div>
              <span class="badge stable">Intent bleibt stabil</span>
            </div>

            <textarea
              id="rawPrompt"
              spellcheck="false"
              aria-label="Raw User Prompt"
            >Fasse ein langes PDF quellentreu in Markdown zusammen und erstelle ein PDF-Render-Skript.</textarea>

            <div class="control-grid">
              <label>
                <span>Task Type</span>
                <select id="taskType"></select>
              </label>
              <label>
                <span>Model Adapter</span>
                <select id="adapterProfile"></select>
              </label>
              <label>
                <span>Harness Policy</span>
                <select id="harnessPolicy"></select>
              </label>
            </div>

            <div class="button-row">
              <button id="generateButton" class="primary-button" type="button">Boost Prompt</button>
              <button id="copyPromptButton" type="button">Copy Prompt</button>
              <button id="downloadJsonButton" type="button">Export JSON</button>
            </div>
          </section>

          <section class="panel prompt-panel" aria-labelledby="prompt-title">
            <div class="panel-head">
              <div>
                <p class="panel-kicker">Output</p>
                <h2 id="prompt-title">Harness Prompt Draft</h2>
              </div>
              <span id="overallBadge" class="badge">Score</span>
            </div>
            <pre id="boostedPrompt" class="output-box" aria-label="Boosted Harness Prompt"></pre>
          </section>

          <section class="panel spec-panel" aria-labelledby="spec-title">
            <div class="panel-head">
              <div>
                <p class="panel-kicker">Compiler View</p>
                <h2 id="spec-title">Task Spec</h2>
              </div>
              <span id="wordBadge" class="badge">0 words</span>
            </div>
            <div id="taskSpecCards" class="spec-cards"></div>
          </section>
        </div>
      </section>

      <section id="score" class="section-band" aria-labelledby="score-title">
        <div class="section-head">
          <div>
            <p class="eyebrow">Local Validation</p>
            <h2 id="score-title">Der Boost bekommt nur einen Heuristik-Score.</h2>
          </div>
          <p>
            Der lokale Score beweist keine Ueberlegenheit und misst keine Ergebnisqualitaet.
            Er schaetzt nur Prompt-Shape-Signale wie Struktur, Eingabe-Adequacy, Trennung von
            Intent und Scaffolding sowie sichtbare Verifikationshinweise.
          </p>
        </div>
        <div id="scoreGrid" class="score-grid"></div>
      </section>

      <section id="eval-lab" class="section-band" aria-labelledby="eval-title">
        <div class="section-head">
          <div>
            <p class="eyebrow">Eval Harness</p>
            <h2 id="eval-title">Profilvergleich ueber Beispielprompts</h2>
          </div>
          <p>
            Die Tabelle vergleicht lokal nur Prompt-Profile nach Heuristik-Score. Sie ist kein
            Modellbenchmark; echte Modelllaeufe koennen spaeter ueber das PromptRunner-Interface
            und einen JudgeRunner angeschlossen werden.
          </p>
        </div>
        <div class="table-wrap">
          <table>
            <thead>
              <tr>
                <th>Case</th>
                <th>Task Type</th>
                <th>Best Profile by Heuristic</th>
                <th>Heuristic</th>
                <th>Key Signal</th>
              </tr>
            </thead>
            <tbody id="evalRows"></tbody>
          </table>
        </div>
      </section>

      <section id="export" class="section-band export-band" aria-labelledby="export-title">
        <div class="section-head">
          <div>
            <p class="eyebrow">Portable Artifact</p>
            <h2 id="export-title">Portables JSON-Artefakt fuer das groessere Agentur-Projekt</h2>
          </div>
          <p>
            Dieses Objekt ist ein portables Export-Artefakt: beobachteter Task Spec,
            Adapter-Scaffolding, Harness Policy, Adapterprofil, Prompt Draft,
            Heuristik-Score und Eval Plan in einer transportierbaren Struktur.
          </p>
        </div>
        <pre id="jsonExport" class="output-box json-box" aria-label="PromptBoost JSON Export"></pre>
      </section>

      <section id="sources" class="section-band" aria-labelledby="sources-title">
        <div class="section-head">
          <div>
            <p class="eyebrow">Quellenrahmen</p>
            <h2 id="sources-title">Dokumentation als Prior, Evals als Urteil</h2>
          </div>
          <p>
            Die Adapter unterscheiden Doku-Priors von expliziten Adapter-Hypothesen.
            Modelllaeufe muessen diese Hypothesen spaeter empirisch bestaetigen oder verwerfen.
          </p>
        </div>
        <div class="table-wrap">
          <table>
            <thead>
              <tr>
                <th>Familie</th>
                <th>Prior / Quellenanker</th>
              </tr>
            </thead>
            <tbody>
              <tr>
                <td>OpenAI / GPT 路 strong_prior</td>
                <td><a href="https://developers.openai.com/api/docs/guides/prompt-engineering" rel="noreferrer">Prompt engineering</a> 路 <a href="https://developers.openai.com/api/docs/models/gpt-5.4" rel="noreferrer">GPT-5.4</a></td>
              </tr>
              <tr>
                <td>Claude 路 strong_prior</td>
                <td><a href="https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/claude-prompting-best-practices" rel="noreferrer">Prompting best practices</a> 路 <a href="https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/long-context-tips" rel="noreferrer">Long context</a></td>
              </tr>
              <tr>
                <td>Gemini 路 strong_prior</td>
                <td><a href="https://ai.google.dev/gemini-api/docs/prompting-strategies" rel="noreferrer">Prompt design strategies</a></td>
              </tr>
              <tr>
                <td>Kimi 路 moderate_prior</td>
                <td><a href="https://platform.moonshot.ai/docs/guide/prompt-best-practice" rel="noreferrer">Prompt best practices</a> 路 <a href="https://platform.moonshot.ai/docs/guide/kimi-k2-5-quickstart" rel="noreferrer">K2.5 quickstart</a></td>
              </tr>
              <tr>
                <td>GLM / Qwen / DeepSeek / Mistral / MiniMax / Llama</td>
                <td>Siehe Adapter-Evidenz im JSON Export; moderate_prior oder hypothesis bedeutet: Quelle reicht fuer Startannahmen, nicht fuer absolute Regeln.</td>
              </tr>
            </tbody>
          </table>
        </div>
      </section>
    </main>

    <script type="module" src="/src/app.ts"></script>
  </body>
</html>