What Shall We Build Next?

Describe

Describe your task

Refine

Refine the plan

SubAgents

Review all agents

Deploy

Deploy your agent

A) SUBAGENT SUMMARY:  
Generate a production-ready Python script (reminder_service.py) that reads the client-payments workbook, emails personalised payment reminders with one-click acknowledgement links, and logs the “sent” status in the same workbook.

B) FINAL TASK OUTPUT:  
reminder_service.py — UTF-8 text, Python ≥3.9, ~150-300 LOC, ready-to-run from CLI (`python reminder_service.py`) and containing:  
• pandas/openpyxl logic to open & update {payments-file}  
• secure SMTP delivery using env-var values defined in {smtp-config}  
• per-row UUID generation → {{base-url}}/acknowledge?id=<uuid>  
• graceful retry/back-off, logging, and PEP-8 compliant docstrings.

C) SUBAGENT INPUT:  
{payments-file}  – path / URL of the Excel workbook  
{smtp-config}    – JSON / dict of required env-var names (SMTP_HOST …)  
{base-url}       – absolute URL stem that the acknowledgement endpoint will live under

E) SUBAGENT TASK SUMMARY (skill chain)  
{payments-file}+{smtp-config}+{base-url}  
 ▶ #223  “Draft detailed technical spec & pseudocode for reminder_service.py that fulfils all stated requirements (workbook columns, env-vars, retry rules, link pattern). Output as plain text.” ➜ [spec-text]  
 ▶ #223  “Using the above spec-text, generate the fully-functional Python code for reminder_service.py. Return only the code block.” ➜ reminder_service.py (v1)  
 ▶ #223  “Self-review reminder_service.py for syntax, PEP-8, and runtime errors; add missing imports, type hints, docstrings, and inline comments; then output the polished file.” ➜ reminder_service.py  (final)

F) SILOS  
Silo 1 – Specification Builder: first #223 call → [spec-text]  
Silo 2 – Code Generator: second #223 call → reminder_service.py (v1)  
Silo 3 – QA & Polish: third #223 call → reminder_service.py (final)

SubAgent #1 - Diagram

Expand Diagram

A) SUB-AGENT SUMMARY  
Generate a polished, responsive HTML e-mail template—containing the required merge-placeholders and a single call-to-action button—that the Core Service (reminder_service.py) can populate and send to every client.

B) FINAL TASK OUTPUT  
reminder_email.html – one self-contained HTML file (≈3-10 KB) with:  
• Inline CSS only (no external assets)  
• Place-holders {{ClientName}}, {{AmountDue}}, {{DueDate}}, {{Response_Link}}  
• CTA button styled in-line that points to {{Response_Link}} (which will resolve to {base-url}/acknowledge?id=<uuid>)  
• Accessible HTML (ALT text, minimum 14 px font, mobile-first ≤600 px width)  
• UTF-8, CRLF line endings

C) SUB-AGENT INPUT  
{base-url} – the public domain (schema+host) that will prefix the acknowledge endpoint link

E) SUB-AGENT TASK SUMMARY  
{base-url} → #190 (write HTML) → #223 (QC / amend) → reminder_email.html

Step-by-step chain  
1. Build prompt (internally) that instructs the writer to:  
   • use a friendly, professional tone  
   • create responsive table-based layout max-width 600 px  
   • insert placeholders {{ClientName}}, {{AmountDue}}, {{DueDate}} in the copy  
   • create CTA button “Confirm Receipt” whose href="{{Response_Link}}" where that value will be constructed later as {base-url}/acknowledge?id={{UUID}}  
   • keep all CSS inline; include dark-mode safe colours  
2. Invoke Skill #190 – “Write or rewrite text based on instructions” with the above prompt and {base-url} so the writer can embed an example of the fully-qualified link.  
   Output: raw HTML text.  
3. Invoke Skill #223 – “Powerful LLM Prompt-to-Text Response” to perform an automated quality-control pass:  
   • validate presence and spelling of all four placeholders  
   • ensure HTML has <!DOCTYPE html>, <html lang="en">, <meta charset="utf-8">  
   • minify whitespace while keeping readability  
   This step returns the final HTML.  
4. Save returned HTML as reminder_email.html (no further changes).

F) SILOS  
No additional silos—single linear flow sufficient for one file generation.

The sub-agent therefore consumes {base-url} and emits a production-ready reminder_email.html file that the later agents (and the Core Service) can copy straight into outbound e-mails.

SubAgent #2 - Diagram

Expand Diagram

A) SUBAGENT SUMMARY  
Generate a lightweight public HTTP endpoint that records a client’s click into the same Excel workbook and package all the code & dependency metadata needed for deployment.

B) FINAL TASK OUTPUT  
[endpoint-and-reqs] → a three-file bundle  
• acknowledge_endpoint.py (UTF-8 Python 3 script, ≈120-180 LOC)  
• requirements.txt (plain-text, 4-8 lines)  
• 00_README_ENDPOINT_SUMMARY.txt (plain-text, ≤30 lines, lists the two filenames, a 1-sentence purpose for each, and the SHA-256 hash of each file)

C) SUBAGENT INPUT  
{payments-file} → absolute or relative path to the existing .xlsx workbook  
{base-url}   → fully-qualified URL the email button will hit, e.g. https://pay.example.com

E) SUBAGENT TASK SUMMARY  
{payments-file}, {base-url}  
 │  
 └─► #223 Powerful LLM Prompt-to-Text Response  ➜ “CREATE acknowledge_endpoint.py”  
  Prompt details:  
  • Use Flask (single-file)  
  • Route GET /acknowledge?id=<uuid>  
  • Load workbook with pandas/openpyxl; locate row where UUID column == id; write “Clicked” in Response Status and utcnow() in Response Timestamp  
  • Use filelock.FileLock(<payments-file>.lock) to prevent race conditions  
  • Return minimal «Thank-you, your response was registered» HTML  
  → OUTPUT: acknowledge_endpoint.py (saved)  
 │  
 └─► #223 Powerful LLM Prompt-to-Text Response  ➜ “CREATE requirements.txt”  
  Prompt includes: “List only what is needed for acknowledge_endpoint.py to run.”  
  → OUTPUT: requirements.txt (saved)  
 │  
 └─► #223 Powerful LLM Prompt-to-Text Response  ➜ “GENERATE summary bundle”  
  Prompt supplies the literal contents of acknowledge_endpoint.py and requirements.txt and instructs LLM to:  
  • Calculate SHA-256 hashes for each file content  
  • Produce a text block with:  
   – Filename  
   – One-sentence description  
   – SHA-256 hash  
  → OUTPUT: 00_README_ENDPOINT_SUMMARY.txt (saved)  
 │  
 └─► Collect the three files as final bundle → [endpoint-and-reqs]

F) SILOS

SILO 1 – Endpoint Code Generation  
 Input: {payments-file}, {base-url}  
 Action: #223 → acknowledge_endpoint.py

SILO 2 – Dependency Declaration  
 Input: acknowledge_endpoint.py  
 Action: #223 → requirements.txt

SILO 3 – Integrity & Documentation  
 Input: acknowledge_endpoint.py + requirements.txt  
 Action: #223 → 00_README_ENDPOINT_SUMMARY.txt

The three silos run sequentially; the final bundle is returned to the parent workflow as [endpoint-and-reqs].

SubAgent #3 - Diagram

Expand Flow

A) SUBAGENT SUMMARY  
“Runbook Author” creates a production-ready Markdown runbook (DEPLOY.md) that explains how to install the codebase, set environment variables, run the reminder-sending script, expose the acknowledgement endpoint, automate with cron / gunicorn / Docker, and apply basic security & backup practices.

B) FINAL TASK OUTPUT  
• DEPLOY.md – UTF-8 Markdown, ≤ 10 KB, containing:  
  – Title & overview  
  – File map (reminder_service.py, reminder_email.html, acknowledge_endpoint.py, requirements.txt, payments.xlsx)  
  – Prerequisites (Python 3.10+, pip, optional Docker)  
  – One-line install: pip install ‑r requirements.txt  
  – Environment variables table (SMTP_HOST, SMTP_PORT, SMTP_USER, SMTP_PASS, DOMAIN_BASE_URL, PAYMENTS_FILE)  
  – CLI usage examples  
  – Cron snippet (daily 08:00)  
  – Gunicorn + systemd service OR docker-compose.yml example  
  – Security/back-up notes  
  – Troubleshooting section

C) SUBAGENT INPUT  
{payments-file} – absolute/relative path or S3 URL of the Excel workbook  
{smtp-config} – JSON/YAML or env-var list of the SMTP settings  
{base-url} – public domain or ngrok URL where /acknowledge endpoint will live  
[reminder-service-script] – text of reminder_service.py  
[email-template] – text of reminder_email.html  
[endpoint-and-reqs] – bundle containing acknowledge_endpoint.py, requirements.txt, & their summary

E) SUBAGENT TASK SUMMARY  
Inputs → Skill Chain → Output

1. Collect all inputs (filenames, env vars, URLs).  
2. #223 Powerful LLM Prompt-to-Text Response  
   • Prompt: “Draft a bullet list of deployment instructions for this project. Include prerequisites, env vars with short descriptions, install commands, running scripts manually, cron example, gunicorn command, docker-compose snippet, security & backup tips. Use the following information … {inject all inputs}.”  
   • Output: “raw_runbook.txt” – well-structured bullets & code snippets (plain text).

3. #190 Write or Rewrite Text Based on Instructions  
   • Prompt: “Rewrite raw_runbook.txt into polished Markdown. Add H2/H3 headings, fenced code blocks, ordered lists. File name must be ‘DEPLOY.md’. Begin with a one-sentence executive summary.”  
   • Input: raw_runbook.txt  
   • Output: DEPLOY.md (final clean Markdown file).

F) SILOS

Silo 1 – “Content Drafting”  
   Inputs → #223 → raw_runbook.txt

Silo 2 – “Markdown Polishing”  
   raw_runbook.txt → #190 → DEPLOY.md  ( [deployment-runbook] )