PUBLIC AGENT HISTORYFICTIONAL DEMO

Watch a workshop agent learn from manuals and corrections

A fully fictional agent history with versioned source files, one important correction and review-only output.

7 public files9 visible events2 produced files0 real tools
AGENT7ACTIVE
FILES
Training repository

Open every file it learned from.

This is fabricated public-safe content. No private document is hidden behind these summaries.

AAGENTS.mdActive · v2

Codex agent instructions

# Workshop Apprentice

## Training mode

Learn from approved files, show uncertainty, and ask the teacher concise
questions when meaning, scope, conflicts, or exceptions are unclear.

## Learned specialization

Use approved manuals and corrections to prepare safe troubleshooting checklists for human review.

## Rules

- Read `rules/training-rules.yaml` before using the training material.
- Consult durable teacher explanations in `memory/teacher-notes/`.
- Cite the files used and never invent missing guidance.
- Never publish, send, purchase, or make an external change without human approval.
Frules/training-rules.yamlActive · v1

Training rules

mode: training
knowledge_policy:
  use_only_approved_files: true
  cite_files_used: true
  ask_when_uncertain: true
quick_replies:
  yes_no_when_binary: true
  values:
  - 'Yes'
  - 'No'
human_approval_required_for:
- publishing
- sending
- purchasing
- external changes
Fmemory/README.mdActive · v1

Cognitive memory map

# Cognitive memory

This folder stores durable teacher explanations and lessons. Every memory is a
visible file that can be reviewed, versioned, exported, and used by Codex.
Fknowledge/m17-overload.mdActive · v1

Approved M17 manual

The M17 overload section requires recording the exact fault code, confirming safe state, and checking load, ventilation and connection evidence before forming a conclusion.
Fprocedures/overload-triage.mdActive · v1

Overload triage procedure

Confirm a safe state. Record the exact code. Inspect approved load and ventilation evidence. List missing measurements. Prepare findings for a qualified technician.
Fcorrections/reset-is-not-repair.yamlActive · v1

A reset is not a repair

Never recommend repeated resets after a protection trip. Preserve fault evidence, identify the cause and require qualified approval before restart.
Ftests/intermittent-stop.yamlActive · v1

Unexpected intermittent stop

For an unfamiliar intermittent stop, ask for the missing fault record, cite approved files, separate observation from assumption and stop before recommending action.
Questions the agent asked

Uncertainty became visible.

These fabricated examples show quick answers and teacher explanations becoming inspectable memory.

Yes or no

Should I treat “Approved M17 manual” as authoritative guidance?

Yes

knowledge/m17-overload.md
Explanation

When should I use “Approved M17 manual”, and what exceptions should I remember?

Use it for relevant draft work, but stop when evidence is missing or a human decision is required.

knowledge/m17-overload.md
Agent’s desk

Files it produced.

Every file shows its review state and the sources used. These copies were fabricated for the demonstration.

Publishedoutputs/service-note-draft.md · v1

Service note draft

Draft finding: the available evidence is consistent with an overload event, but the cause is not established. A qualified technician must review the missing measurements before restart.
Used:Approved M17 manualOverload triage procedure
Publishedoutputs/overload-checklist.md · v1

Motor overload evidence checklist

A review-ready checklist covering safe state, fault code, load evidence, ventilation, connections, missing measurements and technician approval.
Used:Approved M17 manualOverload triage procedure
SAFE SANDBOX TEST

Can you find something it has not learned?

Use a simulated situation only. The demo must find a taught method, stop for human approval or admit that its files do not contain the answer.

Do not enter personal, customer, confidential, or live operational information.
Transparent prototype

The public sandbox uses deterministic matching against reviewed fictional summaries. It demonstrates visible training state; it does not pretend a model was fine-tuned or connected to real tools.