Generate business and technical artifacts

donald-company-workspace/canonical.json

@@ -0,0 +1 @@
+{}

donald-company-workspace/prompts/careers_page.txt

@@ -0,0 +1,13 @@
+INPUT_JSON: @canonical.json
+
+TASK: Generate a careers page (Markdown) targeted at senior ML engineers, simulation experts, and control systems engineers.
+
+REQUIREMENTS:
+- Include: mission statement, top 5 reasons to join, 3 sample role descriptions with responsibilities and KPIs, early team culture bullets, visa/remote policy, perks relevant to senior hires.
+- Tone: ambitious, technical, respectful.
+- Output in Markdown.
+- Preserve facts from canonical.json; if missing, mark [MISSING_DATA: ...].
+
+MODEL_SETTINGS: temperature=0.1, max_tokens=900
+
+RETURN: Markdown careers page.

donald-company-workspace/prompts/investor_memo.txt

@@ -0,0 +1,12 @@
+INPUT_JSON: @canonical.json
+
+TASK: Produce an 850–1000 word investor memo for VCs focused on enterprise AI and industrial automation.
+
+CONSTRAINTS:
+- Tone: crisp, evidence-based, founder-friendly.
+- Structure: 1) One-line thesis, 2) Market & TAM, 3) Product & tech defensibility (explain WDBX briefly), 4) Business model & revenue mechanics, 5) Pilot traction & 90-day plan (use the priority_checklist_90_days exactly), 6) Top 5 risks & mitigations, 7) Ask (funding amount and use of proceeds).
+- Preserve all numeric claims and the 90-day checklist exactly as in the input JSON.
+- If any pilot metric is missing, return exactly the text: [MISSING_DATA: <list of missing required metrics>] and do not invent numbers.
+
+MODEL_SETTINGS: temperature=0.1, max_tokens=1000, top_p=0.95
+RETURN: Plain text investor memo.

donald-company-workspace/prompts/job_posting.txt

@@ -0,0 +1,12 @@
+INPUT_JSON: @canonical.json
+
+TASK: Create a senior ML engineer job posting (500–700 words) with requirements, responsibilities, interview stages, and sample KPIs.
+
+CONSTRAINTS:
+- Include skills: simulation/FEA familiarity, time-series sensor analytics, model interpretability, PLC/robotics integration preferred.
+- Include remote/hybrid options and compensation band guidance: "competitive; equity available".
+- If any required detail is missing, mark [MISSING_DATA: ...].
+
+MODEL_SETTINGS: temperature=0.1, max_tokens=700
+
+RETURN: Plain text job posting.

donald-company-workspace/prompts/rewrite_instruction.txt

@@ -0,0 +1,12 @@
+INPUT_JSON: @canonical.json
+
+TASK: Rewrite the provided draft to improve clarity, ensure tone matches the target audience, and strictly preserve all numeric claims and facts from INPUT_JSON. Replace any invented or unknown metrics with [MISSING_DATA: <metric_name>].
+
+REQUIREMENTS:
+- Do not change any numbers found in INPUT_JSON.
+- If a required detail is not present in INPUT_JSON, insert [MISSING_DATA: ...] verbatim.
+- Keep structure and headings; tighten phrasing for concision.
+
+MODEL_SETTINGS: temperature=0.0, max_tokens=800
+
+RETURN: Revised text only.

donald-company-workspace/prompts/slide_deck.txt

@@ -0,0 +1,9 @@
+INPUT_JSON: @canonical.json
+
+TASK: Produce a slide deck outline (10 slides). Each slide: title, 3–6 bullets, one-sentence speaker note.
+
+CONSTRAINTS: Focus on persuading Series A VC: traction, team, tech defensibility, unit economics, roadmap, and ask.
+
+MODEL_SETTINGS: temperature=0.1, max_tokens=800
+
+RETURN: Markdown outline with 10 slides.

donald-company-workspace/prompts/technical_blog.txt

@@ -0,0 +1,13 @@
+INPUT_JSON: @canonical.json
+
+TASK: Write a 1200-word technical blog post titled "WDBX & Abi: Causal Backtrace for Manufacturing."
+
+REQUIREMENTS:
+- Explain WDBX conceptually, how it integrates with LCM and digital twins.
+- Provide example workflow design→simulation→pilot→feedback and 3 engineering challenges + mitigation strategies.
+- Use headings and short code-style pseudocode where helpful.
+- If any claim lacks data, mark [MISSING_DATA: ...].
+
+MODEL_SETTINGS: temperature=0.1, max_tokens=1400
+
+RETURN: Markdown technical blog post.

donald-company-workspace/scripts/verify_numbers.js

@@ -0,0 +1,27 @@
+// Node script: verify that numbers in canonical.json appear in generated outputs
+const fs = require('fs');
+const path = require('path');
+
+function read(file) {
+  const p = path.resolve(file);
+  return fs.readFileSync(p, 'utf8');
+}
+
+try {
+  const canonical = JSON.parse(read('donald-company-workspace/canonical.json'));
+  const outputs = read('donald-company-workspace/outputs/investor_memo.txt');
+
+  const searchNums = (obj, set = new Set()) => {
+    if (typeof obj === 'number') set.add(String(obj));
+    else if (Array.isArray(obj)) obj.forEach(x => searchNums(x, set));
+    else if (obj && typeof obj === 'object') Object.values(obj).forEach(x => searchNums(x, set));
+    return set;
+  };
+
+  const nums = [...searchNums(canonical)];
+  const missing = nums.filter(n => !outputs.includes(n));
+  console.log('Numbers missing in investor memo:', missing);
+} catch (err) {
+  console.error('Verification failed:', err.message);
+  process.exitCode = 1;
+}