Project Ops

Agentic Coding — Historical Track Audit

Project operations source: Agentic Coding — Historical Track Audit.

This is historical evidence, preserved on 2026-09-09 from source checkpoint 5ec1c54. The matching Astro checkpoint is 11cef23. The original inventory below is a planning record, not current curriculum readiness or a commitment to build every idea.

The current Track Guide and shared Track Map carry the current composition: Build a Safe Text-to-Speech Skill as the signature entry, shared preparation, and three developing build directions.

Next Artifact Pass — Evidence and Unresolved Questions

  • The signature module remains Draft / Drafting, Builder level, owned by Mike. Its primary lab remains Alpha, also owned by Mike. The Track remains Draft / Drafting and Unassigned; choosing a signature does not assign Track ownership or promote readiness.
  • The lab README records the missing runnable safe repository setup and referenced Voice Auditions project as reasons for Alpha. Resolve those core inputs before a full learner pilot. Existing worksheets, a checklist, prompt recipes, Activity, and standalone Notebook do not by themselves close those gaps.
  • The Activity and Notebook require a personal provider account, live MP3, and human listening review for full completion. The facilitator guide accepts a documented mock result as learning evidence. Preserve meaningful account-free planning, testing, and review participation, but reconcile full-completion expectations before claiming an equivalent mock route.
  • Timing also needs reconciliation: the lab lists 90–120 minutes, a 45–60 minute dry-run portion, and 2–3 hours including the static audition-site brief. That brief is required by its Done contract. The facilitator guide proposes 105 minutes. The Track shows the extended estimate instead of promising full completion in the shorter window; none of these estimates substitutes for pilot timing.
  • Accessibility, secrets-practice, current provider documentation, spend controls, and learner-pilot review remain open. Supply transcripts and determine an accessible audio-assessment path. This Track pass does not verify provider currency or authorize live synthesis.
  • Prompting and Context Engineering-or-equivalent safe-file practice are the signature module’s stated dependencies. Both modules remain Sketch / Planned and primary to AI Literacy, without reciprocal Agentic Coding Track metadata. Their own dependencies are unchanged.
  • Safety remains Draft / Drafting, primary to AI Literacy and related to Curiosity. It is a proposed shared-support relationship here, not a reassignment or a substitute for repository and credential practice.
  • The lower-friction beginner coding environment has not been selected. Game, bounded-automation, and sharing/maintenance directions need packages and tested boundaries. Their named ideas below are not evidence of canonical source pages.
  • Publication, deployment, and production remain separate human decisions. Voice cloning is idea-only and would need separate consent, rights, disclosure, and governance review. Signature and route walkthrough demonstrations remain Coming Soon.
  • Use a bounded planning/mock review to gather timing and friction if full-pilot inputs are not ready. Do not rewrite module or lab instructions as part of this Track composition pass.

Original Track Guide Snapshot

Agentic Coding — Original Outline

Status: Working track outline, not finished curriculum.

This track teaches learners how to use AI to build games, automations, prototypes, and small tools. It should not pretend everyone must become a software engineer. Instead, it should show how people can participate in building by describing goals, reviewing behavior, testing outputs, and learning enough structure to collaborate with coding agents responsibly.

The invitation is practical and creative. Some learners will write code. Some will design workflows. Some will test. Some will make games. Some will automate repetitive work. The shared skill is learning how to turn intention into working behavior without surrendering judgment.

Audience: Builders, hobbyists, students, automation-minded staff, game makers, and future technical contributors.

Purpose: Help learners use AI coding tools safely and productively for small, reviewable projects.

Likely Modules

  • What Is Code When AI Helps Write It?
  • Building a Tiny Game
  • Building a Small Automation
  • Planning Before Coding
  • Build a Safe Text-to-Speech Skill
  • Testing What the Agent Built
  • Publishing Your Work
  • Git and GitHub for Learners
  • Deploying Version 2

Starter Lab Ideas

  • Build a small browser game from a clear spec.
  • Automate a low-risk personal workflow.
  • Ask an AI coding agent to explain the files it changed.
  • Test a small app against acceptance criteria.
  • Publish a static project page.
  • Plan, build, and test a text-to-speech skill with a personal low-spend provider account.

Additional Reading

No additional reading is listed yet.

Dependencies

  • Prompting basics.
  • Agentic literacy.
  • Safe file and data habits.
  • Basic version-control concepts.
  • Testing and review habits.

MVP Holes

  • Define the lowest-friction coding environment for learners.
  • Decide how much Git/GitHub is required before publishing.
  • Set rules for not using private data in learner automations.
  • Define the boundary between fun coding labs and production automation.

Source: ops/track-audits/agentic-coding.md