MWMS AI Agent Skill Library Framework

System: MWMS
Document Type: Framework
Authority Level: MCR Source Of Truth
Status: Draft For MCR
Primary Location: MCR
Future Operational Destination: HeadOffice Brain, MWMS Brain, Brain Room, AI Manager, AI Employee Router, Task Executor Systems, Course Absorption System, Newsletter Intelligence, Opportunity System, AI Business Systems Brain
Parent Page: HeadOffice
Owner: Martyn
Developer Boundary: Do Not Touch M’s Active Build Areas Unless Specifically Assigned
Source Of Truth: MCR

Purpose

The purpose of this document is to define the MWMS AI Agent Skill Library Framework.

This framework explains how MWMS stores, governs, reuses, improves, and applies procedural skills for AI Employees.

MWMS must not rely on one-off prompts every time an AI Employee performs work.

As the MWMS AI workforce grows, repeated tasks must become reusable skills.

A skill is not just knowledge.

A skill is a repeatable way of doing work.

For example, the Course Absorption Agent should not need to relearn every session how to:

evaluate whether a course is worth absorbing
separate useful system value from generic content
avoid duplicate MCR pages
map insights to the correct Brain
produce full page output
recommend absorb, park, ignore, or reject
protect M’s active development work

That operating procedure should become a stored AI skill.

The AI Agent Skill Library gives MWMS a way to turn repeated work into reusable procedural memory.

Scope

This framework applies to all reusable AI Employee skills across MWMS.

This includes skills used by:

HeadOffice Brain
Brain Room
AI Manager
AI Employee Router
Task Executor systems
Dev Console
Newsletter Intelligence
Course Absorption
Opportunity System
Affiliate Brain
Research Brain
Experimentation Brain
Finance Brain
Content Brain
Ads Brain
Sales Brain
Conversion Brain
Operations Brain
AI Business Systems Brain
future AIBS client systems

This framework applies when MWMS needs to define repeatable procedures for AI Employees, workflows, reports, validation checks, tool usage, handoffs, task conversion, research, offer evaluation, developer support, or client-facing AI systems.

Core Definition

An AI Agent Skill is a reusable procedural playbook that teaches an AI Employee how to perform a specific type of work correctly.

A skill defines:

what task the AI Employee is performing
when the skill should be used
what input is required
what context is required
what steps must be followed
what output must be produced
what standards apply
what must be avoided
what validation is required
where the output should go
what failure triggers exist
what outcome should be created

A skill is different from a tool.

A tool gives the AI Employee access or capability.

A skill tells the AI Employee how to use its capability properly.

Core Principle

The core principle of this framework is:

Tools give AI Employees hands. Skills teach AI Employees how to work.

A powerful tool without a skill can create bad work faster.

A skill without the right tool may still be useful in manual or drafting mode.

The strongest MWMS AI Employees will combine:

clear role card
defined capability stack
approved tool permissions
correct context pack
reusable skill
validation checklist
handoff package
outcome measurement

The skill library helps MWMS move from repeated prompting to repeatable operating procedure.

Skill Library Layers

The MWMS AI Agent Skill Library is organized into eight layers:

Skill Identity
Skill Trigger
Skill Input
Skill Procedure
Skill Output
Skill Validation
Skill Handoff
Skill Improvement

1. Skill Identity

Each AI skill must have a clear identity.

A skill identity defines:

skill name
owning Brain
assigned AI Employee
purpose
workflow supported
related standards
status
version

Good skill names:

Course Value Extraction Skill
Newsletter Signal Filtering Skill
Brain Room Task Conversion Skill
Offer Evidence Separation Skill
Developer Handoff Precision Skill
MCR Duplicate Risk Check Skill
Dashboard Signal Readiness Skill
AIBS Client Report Drafting Skill

Bad skill names:

AI Skill
General Thinking
Better Output
Marketing Helper
Smart Workflow

Skill Identity Rule:

Every skill must describe the work it performs.

2. Skill Trigger

A skill must define when it should be used.

Skill triggers may include:

a course file is uploaded
a newsletter enters the review workflow
a Brain Room message contains a task request
an affiliate offer needs evaluation
a developer issue needs instructions for M
a report needs validation
a handoff is required
a dashboard item is being prepared
a future client report is being drafted

Trigger clarity prevents the wrong skill from being applied to the wrong task.

Skill Trigger Rule:

A skill should activate because the task requires it, not because the skill exists.

3. Skill Input

A skill must define what input it can process.

Input may include:

course transcript
course PDF
newsletter email
Brain Room message
offer page
research source
finance numbers
experiment result
screenshot
WordPress page list
Supabase row
developer file
client document
pasted user instruction

The skill must also define:

required input
optional input
incomplete input handling
messy input handling
source grounding requirement
input normalization requirement

Skill Input Rule:

A skill should not operate on input it is not designed to handle.

4. Skill Procedure

The procedure is the heart of the skill.

It defines the steps the AI Employee must follow.

A procedure may include:

normalize the input
identify source of truth
classify the task
identify owning Brain
extract relevant signal
ignore irrelevant content
separate evidence from assumptions
apply relevant MWMS standards
produce required output
validate the output
prepare handoff
capture learning
log outcome or failure

Skill procedures should be simple enough to follow but detailed enough to prevent drift.

Skill Procedure Rule:

A skill must tell the AI Employee how to perform the work, not just what the work is.

5. Skill Output

A skill must define its expected output.

Possible outputs include:

course absorption report
full MCR page output
newsletter intelligence record
Agentic Work Unit
AI Employee Role Card
Capability Stack
Tool Permission Record
Context Pack
validation report
handoff package
failure log
outcome log
developer brief
offer evaluation report
dashboard card
client report draft

Skill Output Rule:

A skill is incomplete unless the output format is defined.

6. Skill Validation

Every important skill must include validation rules.

Validation may check:

task alignment
completeness
source grounding
specificity
correct Brain routing
correct output format
risk level
duplication risk
developer boundary
tool permission boundary
compliance concerns
human review requirement
outcome usefulness

Skill Validation Rule:

A skill should not only produce work. It should help make the work checkable.

7. Skill Handoff

A skill must define where the output goes next.

Possible handoff destinations include:

MCR
HeadOffice review
Brain Room
AI Manager
Newsletter Queue Review
Routed Actions
HeadOffice Dashboard
Research Brain
Finance Brain
Experimentation Brain
Ads Brain
Content Brain
M developer handoff
Parking System
Archive
Human Review Queue
AIBS Client Review

Skill Handoff Rule:

A skill must produce output with a destination, not output that floats in space.

8. Skill Improvement

Skills must improve over time.

Skill improvement may happen when:

output repeatedly fails validation
user corrects the AI Employee
a workflow changes
a new MWMS rule is created
a better template is created
tool permissions change
a repeated failure mode appears
a stronger prompt or procedure is discovered
client delivery requirements change

Skill Improvement Rule:

Skills must evolve through Kaizen, not remain static instructions.

Skill Types

MWMS should classify AI skills by type.

1. Intake Skills

Used when information first enters MWMS.

Examples:

Messy Input Intake Skill
Source Completeness Check Skill
Input Classification Skill
Brain Ownership Detection Skill

Purpose:

To make sure raw input becomes usable input.

2. Extraction Skills

Used to pull useful content from source material.

Examples:

Course Framework Extraction Skill
Newsletter Signal Extraction Skill
Offer Claim Extraction Skill
Research Evidence Extraction Skill
Developer Issue Extraction Skill

Purpose:

To separate useful signal from noise.

3. Evaluation Skills

Used to judge quality, risk, suitability, or value.

Examples:

Course Absorption Value Skill
Offer Test Suitability Skill
Finance Risk Review Skill
Experiment Signal Quality Skill
Dashboard Readiness Skill

Purpose:

To help MWMS decide whether something should move forward.

4. Creation Skills

Used to create structured outputs.

Examples:

Full Page Output Creation Skill
Agentic Work Unit Creation Skill
Developer Brief Creation Skill
Handoff Package Creation Skill
Client Report Drafting Skill

Purpose:

To turn source material into usable work products.

5. Validation Skills

Used to check outputs before use.

Examples:

MCR Page Validation Skill
Developer Instruction Validation Skill
Newsletter Signal Validation Skill
Offer Evaluation Validation Skill
Client Report Validation Skill

Purpose:

To stop weak or unsafe outputs before they become operational truth.

6. Routing Skills

Used to send work to the right Brain, workflow, person, or queue.

Examples:

Brain Routing Skill
Research Handoff Skill
Finance Review Routing Skill
M Developer Handoff Skill
Parking System Routing Skill

Purpose:

To prevent work from becoming lost, misrouted, or ownerless.

7. Tool Use Skills

Used when an AI Employee operates with a tool or plugin.

Examples:

Gmail Newsletter Read Skill
Supabase Internal Row Write Skill
WordPress Page List Review Skill
File Review Skill
Dashboard Record Preparation Skill

Purpose:

To ensure tools are used safely inside approved boundaries.

8. Reporting Skills

Used to create decision-ready reports.

Examples:

Course Absorption Report Skill
Newsletter Intelligence Report Skill
Offer Evaluation Report Skill
Developer Support Report Skill
AIBS Client Report Skill

Purpose:

To make reports useful, not passive summaries.

9. Failure Handling Skills

Used when something goes wrong.

Examples:

Failed Output Classification Skill
Escalation Decision Skill
Failure Log Creation Skill
Containment Action Skill
Kaizen Lesson Capture Skill

Purpose:

To convert failure into system improvement.

10. Outcome Measurement Skills

Used to judge whether work mattered.

Examples:

Outcome Scoring Skill
Risk Reduction Capture Skill
Business Value Summary Skill
AI Employee Usefulness Review Skill
Workflow Value Review Skill

Purpose:

To make MWMS outcome-driven instead of output-driven.

AI Agent Skill Record Template

Each skill should eventually be recorded using the following structure.

Skill Name

Field:
Skill Name:

Skill Type

Field:
Skill Type:

Recommended values:

Intake Skill
Extraction Skill
Evaluation Skill
Creation Skill
Validation Skill
Routing Skill
Tool Use Skill
Reporting Skill
Failure Handling Skill
Outcome Measurement Skill

Owning Brain

Field:
Owning Brain:

Assigned AI Employee

Field:
Assigned AI Employee:

Skill Purpose

Field:
Skill Purpose:

When To Use This Skill

Field:
When To Use This Skill:

Required Input

Field:
Required Input:

Required Context

Field:
Required Context:

Related Standards

Field:
Related Standards:

Skill Procedure

Field:
Skill Procedure:

Recommended structure:

Capture source.
Normalize input.
Identify task type.
Apply relevant MWMS rules.
Produce required output.
Validate output.
Prepare handoff.
Capture learning or outcome.

Required Output

Field:
Required Output:

Validation Requirement

Field:
Validation Requirement:

Human Review Requirement

Field:
Human Review Requirement:

Tool Permission Boundary

Field:
Tool Permission Boundary:

Forbidden Actions

Field:
Forbidden Actions:

Handoff Destination

Field:
Handoff Destination:

Failure Triggers

Field:
Failure Triggers:

Expected Outcome

Field:
Expected Outcome:

Skill Status

Field:
Skill Status:

Recommended values:

Proposed
Draft
Manual Use
Proven Manual Use
Assisted Use
Controlled Automation Candidate
Parked
Deprecated
Retired

Skill Version

Field:
Skill Version:

Last Reviewed

Field:
Last Reviewed:

Quick Use Version

Use this shorter format when drafting a new skill.

Skill Name:
Skill Type:
Owning Brain:
Assigned AI Employee:
Skill Purpose:
When To Use This Skill:
Required Input:
Required Context:
Related Standards:
Skill Procedure:
Required Output:
Validation Requirement:
Human Review Requirement:
Tool Permission Boundary:
Forbidden Actions:
Handoff Destination:
Failure Triggers:
Expected Outcome:
Skill Status:
Skill Version:
Last Reviewed:

Example 1: Course Absorption Value Extraction Skill

Skill Name:
Course Absorption Value Extraction Skill

Skill Type:
Extraction Skill / Evaluation Skill

Owning Brain:
HeadOffice Brain

Assigned AI Employee:
Course Absorption Agent

Skill Purpose:
Evaluate uploaded course material and extract only reusable MWMS system value.

When To Use This Skill:
Use when the user uploads course files, transcripts, lesson descriptions, PDFs, or course blocks for MWMS absorption.

Required Input:

course file or transcript
lesson title if available
current course absorption context
existing MCR page awareness where available

Required Context:

Course Absorption System v2
MCR source-of-truth rule
anti-duplication rule
full page output rule
current MWMS Brain/Blueprint priorities
M developer boundary

Related Standards:

MWMS Messy Input Normalization Record
MWMS Agentic Reporting Template
MWMS AI Output Validation Checklist
MWMS AI Agent Context Pack Template
MWMS AI Agent Outcome Log Record

Skill Procedure:

Identify the course topic.
Check whether the material is likely to improve MWMS.
Ignore generic content, hype, tool walkthrough fluff, and duplicated ideas.
Extract reusable frameworks, workflows, standards, prompts, operating rules, or strategic insights.
Map useful insights to relevant Brains.
Decide absorb, update existing page, create new page, park, or reject.
If strong enough, prepare full page output.
Include employee creation suggestions if useful.
Include trends, concerns, and what to ignore.
Require human review before MCR save.

Required Output:
Course absorption report or full MCR page output.

Validation Requirement:
Operational Validation.

Human Review Requirement:
Required before MCR save.

Tool Permission Boundary:
Provided Input Only unless verification is required.

Forbidden Actions:

do not absorb weak material
do not create duplicate pages
do not claim unsupported source content
do not interfere with M’s active build
do not create developer work unless specifically requested

Handoff Destination:
MCR, Blueprint background, Parking System, or relevant Brain.

Failure Triggers:

source incomplete
duplicate page risk
weak content
unclear Brain mapping
technical claims requiring M review

Expected Outcome:
MWMS Brain/Blueprint improves, or weak material is rejected.

Skill Status:
Manual Use

Skill Version:
v1.0

Last Reviewed:
YYYY-MM-DD

Example 2: Developer Handoff Precision Skill

Skill Name:
Developer Handoff Precision Skill

Skill Type:
Creation Skill / Validation Skill / Handoff Skill

Owning Brain:
HeadOffice Brain

Assigned AI Employee:
Developer Support Agent

Skill Purpose:
Prepare exact, evidence-based developer instructions for M.

When To Use This Skill:
Use when the user needs code, plugin, WordPress, Supabase, Make.com, or system implementation instructions for M.

Required Input:

exact user request
screenshot if relevant
file content if relevant
current save point
site/domain
known boundaries

Required Context:

exact instruction requirement
full file output rule
current visible evidence
what not to touch
M’s active build boundaries
current save point

Related Standards:

MWMS AI Output Validation Checklist
MWMS AI Employee Handoff Package Template
MWMS AI Agent Context Pack Template
MWMS AI Tool Permission Record Template
MWMS AI Agent Failure Log Record

Skill Procedure:

Identify the exact system, site, screen, file, or issue.
Confirm what evidence is current.
Do not guess missing file paths or hidden state.
Define the current problem.
Define the required change.
Include exact file path where available.
Include exact insertion or replacement location where possible.
Include what not to touch.
Include test steps.
Include expected result.
Escalate if M would need to guess.

Required Output:
Developer handoff brief or full file output.

Validation Requirement:
High Risk Validation.

Human Review Requirement:
Always required.

Tool Permission Boundary:
Provided Input Only unless approved read access is given.

Forbidden Actions:

do not alter live code
do not guess file paths
do not give vague search instructions
do not touch unrelated systems
do not override current screenshot or file evidence

Handoff Destination:
M, Dev Console, Brain Room, or HeadOffice review.

Failure Triggers:

missing file path
missing file content
unclear current state
live system risk
request affects M’s active build without approval

Expected Outcome:
M can act safely without guessing.

Skill Status:
Manual Use

Skill Version:
v1.0

Last Reviewed:
YYYY-MM-DD

Example 3: Newsletter Signal Filtering Skill

Skill Name:
Newsletter Signal Filtering Skill

Skill Type:
Extraction Skill / Evaluation Skill / Routing Skill

Owning Brain:
HeadOffice Brain

Assigned AI Employee:
Newsletter Signal Extraction Agent

Skill Purpose:
Separate business-relevant newsletter signals from generic AI news and route useful signals to the correct Brain or review queue.

When To Use This Skill:
Use when newsletters enter the HeadOffice Newsletter Intelligence workflow.

Required Input:

newsletter subject
sender
date
body or cleaned body
snippet
metadata where available

Required Context:

Newsletter Intelligence Operating Protocol
Brain Routing Rule
business relevance filter
dashboard readiness rules
action categories
current MWMS priorities

Related Standards:

MWMS Messy Input Normalization Record
MWMS Agentic Reporting Template
MWMS AI Output Validation Checklist
MWMS AI Employee Handoff Package Template
MWMS AI Agent Outcome Log Record

Skill Procedure:

Remove newsletter noise, sponsor clutter, footer content, and generic filler.
Identify concrete business signals.
Ignore generic AI hype.
Classify signal type.
Assign primary Brain.
Identify supporting Brains.
Define action type.
Set priority and urgency conservatively.
Recommend ACT NOW, TEST, MONITOR, PARK, or REJECT.
Route only after validation.

Required Output:
Newsletter intelligence record or report.

Validation Requirement:
Operational Validation.

Human Review Requirement:
Required before downstream action.

Tool Permission Boundary:
Read Only / Controlled Write only inside approved workflow.

Forbidden Actions:

do not create downstream tasks without review
do not mark generic news urgent
do not invent current facts
do not route weak signals to dashboards

Handoff Destination:
Newsletter Queue Review, HeadOffice Dashboard candidate, Routed Actions, Parking System.

Failure Triggers:

incomplete email body
unclear signal
compliance or paid traffic risk
current verification required
low confidence

Expected Outcome:
Useful external intelligence becomes visible and actionable while noise is rejected.

Skill Status:
Assisted Use

Skill Version:
v1.0

Last Reviewed:
YYYY-MM-DD

Example 4: MCR Duplicate Risk Check Skill

Skill Name:
MCR Duplicate Risk Check Skill

Skill Type:
Validation Skill / Governance Skill

Owning Brain:
HeadOffice Brain

Assigned AI Employee:
HeadOffice Validation Agent

Skill Purpose:
Prevent duplicate pages, wrong parent placement, and unnecessary page creation inside MCR.

When To Use This Skill:
Use before creating, replacing, renaming, or deleting MCR pages.

Required Input:

proposed page title
existing page list or search result
parent page
page content if duplicate exists
current WordPress evidence

Required Context:

MCR source-of-truth rule
Page Naming Standard
Document Structure Standard
correct parent map
duplicate cleanup rule

Related Standards:

MWMS Page Naming Standard
MWMS Document Structure Standard
MWMS AI Output Validation Checklist
MWMS Messy Input Normalization Record
MWMS AI Agent Failure Log Record

Skill Procedure:

Search exact page title.
Confirm whether page already exists.
Compare parent page first.
If both copies are identical and parent is same, keep either one and trash duplicate.
If parents differ, correct parent wins first.
If wrong-parent version has newer content, merge useful content into correct-parent page before trashing.
If content differs, compare value before deletion.
Do not recommend deletion without checking content identity or parent placement.
Confirm final page appears once only.

Required Output:
Keep/trash/merge recommendation.

Validation Requirement:
Operational Validation.

Human Review Requirement:
Required before trashing.

Tool Permission Boundary:
Provided Input Only / WordPress read evidence if available.

Forbidden Actions:

do not assume newer date means better keeper
do not delete without parent/content check
do not create new page if replacement is required
do not ignore visible WordPress evidence

Handoff Destination:
MCR page cleanup action or HeadOffice review.

Failure Triggers:

duplicate content uncertain
parent unclear
both pages have unique content
user screenshot contradicts recommendation

Expected Outcome:
Cleaner MCR structure and fewer duplicate pages.

Skill Status:
Manual Use

Skill Version:
v1.0

Last Reviewed:
YYYY-MM-DD

Skill Library Governance Rules

Rule 1: Skills Are Not Random Prompts

A skill must be more structured than a prompt.

A prompt may ask for output.

A skill defines repeatable work.

Rule 2: Skills Must Belong To A Brain

Every skill must have an owning Brain.

This prevents skill sprawl.

Rule 3: Skills Must Connect To AI Employees

A skill should usually be assigned to an AI Employee.

If no AI Employee owns the skill, it should remain draft or parked.

Rule 4: Skills Must Have Boundaries

Every skill must define what not to do.

Skills without forbidden actions create drift.

Rule 5: Skills Must Include Validation

A skill must make output checkable.

If a skill cannot be validated, it is not ready for operational use.

Rule 6: Skills Must Not Replace Human Review

Skills can improve output quality.

They cannot remove required human review for high-risk work.

Rule 7: Skills Must Respect Tool Permissions

A skill cannot authorize tool access.

Tool access must come from the AI Tool Permission Record.

Rule 8: Skills Must Improve Through Kaizen

Skills should be reviewed after repeated use, failure, or workflow change.

Rule 9: Skills Must Be Reusable

A skill should only be created when the task is likely to repeat.

One-off instructions do not need to become formal skills.

Rule 10: Client Skills Must Be Isolated

Future AIBS client skills must be client-safe, permissioned, and context-isolated.

Do not mix client contexts or apply client-specific procedures across unrelated clients without review.

Skill Creation Criteria

Create a formal AI skill when:

The task repeats.
The task has a clear procedure.
The task belongs to a Brain.
The task supports a known AI Employee.
The output format can be defined.
The validation rules are known.
The handoff destination is clear.
The task creates a useful business outcome.
The skill prevents repeated prompting.
The skill improves consistency.

Do not create a formal skill when:

The task is one-off.
The task is vague.
No Brain owns it.
No AI Employee owns it.
The procedure is not known.
The output cannot be validated.
The skill duplicates an existing skill.
The workflow is not useful.
The skill would create unnecessary documentation.
The skill is just tool hype.

Skill Review Cycle

Per Use

Ask:

Did the skill produce the right output?
Did it follow the procedure?
Did it avoid forbidden actions?
Did it require human correction?
Did it create an outcome?

After Failure

Ask:

Did the skill miss a step?
Was input missing?
Was context missing?
Was validation too weak?
Were forbidden actions unclear?
Should the skill be updated?

Monthly Or Per Major Workflow Change

Ask:

Is the skill still useful?
Is it too broad?
Does it overlap another skill?
Should it be simplified?
Should it be split?
Should it be parked or retired?

Skill Statuses

Recommended statuses:

Proposed
Draft
Manual Use
Proven Manual Use
Assisted Use
Controlled Automation Candidate
Parked
Deprecated
Retired

Proposed

Idea exists but no procedure yet.

Draft

Skill structure exists but has not been proven.

Manual Use

Skill can be used manually with human review.

Proven Manual Use

Skill has been used repeatedly and works well.

Assisted Use

Skill may support assisted workflows, drafts, or queue creation.

Controlled Automation Candidate

Skill may later support automation but needs readiness review first.

Parked

Useful later but not now.

Deprecated

Superseded by better skill or framework.

Retired

No longer used.

Future Plugin Or UI Relevance

This framework may later support:

AI Skill Library page
AI Employee skill registry
AI Manager skill selector
Brain Room task skill assignment
Task Executor skill instructions
AI Employee Router skill matching
HeadOffice skill review dashboard
AIBS client skill package library

Possible future fields:

skill_id
skill_name
skill_type
owning_brain
assigned_ai_employee
skill_purpose
trigger_conditions
required_input
required_context
related_standards
skill_procedure
required_output
validation_requirement
human_review_requirement
tool_permission_boundary
forbidden_actions
handoff_destination
failure_triggers
expected_outcome
skill_status
skill_version
last_reviewed
created_at
updated_at

No technical build is authorized by this framework alone.

Governance Role

HeadOffice owns the MWMS AI Agent Skill Library Framework.

HeadOffice is responsible for:

deciding when a skill becomes formal
preventing duplicate skills
ensuring skills belong to Brains
ensuring skills align with AI Employee Role Cards
ensuring skills respect Capability Stacks
ensuring skills do not override Tool Permission Records
ensuring skills include validation and handoff rules
ensuring high-risk skills require human review
reviewing skill failures
retiring weak or duplicate skills
protecting M’s active build areas
protecting future AIBS client systems

Individual Brains may propose skills for their AI Employees, but HeadOffice governs cross-Brain, high-risk, tool-enabled, client-facing, and automation-related skills.

Relationship To Other MWMS Standards

This framework supports and must align with:

MWMS AI Agent Operations Core
MWMS AI Employee Role Card Standard
MWMS AI Employee Role Card Template
MWMS AI Employee Capability Stack Framework
MWMS AI Employee Capability Stack Template
MWMS AI Tool Permission And Access Framework
MWMS AI Tool Permission Record Template
MWMS AI Agent Memory And Context Framework
MWMS AI Agent Context Pack Template
MWMS Agentic Work Unit Standard
MWMS Agentic Work Unit Template
MWMS AI Workflow Pipeline Standard
MWMS AI Workflow Pipeline Checklist
MWMS AI Output Validation Standard
MWMS AI Output Validation Checklist
MWMS Messy Input Normalization Framework
MWMS Messy Input Normalization Record
MWMS Agentic Reporting Standard
MWMS Agentic Reporting Template
MWMS AI Employee Handoff Protocol
MWMS AI Employee Handoff Package Template
MWMS AI Agent Failure Handling And Escalation Protocol
MWMS AI Agent Failure Log Record
MWMS AI Agent Outcome Measurement Framework
MWMS AI Agent Outcome Log Record
MWMS AI Agent Deployment Readiness Checklist
MWMS AI Agent Deployment Readiness Review Template
MWMS AI Workforce Governance Model
MWMS Brain Routing Rule
MWMS Brain To Brain Request Protocol
MWMS Supabase Event Schema
AI Business Systems Brain Blueprint

This framework adds the procedural memory layer to the MWMS AI Agent Operations Core.

Drift Protection

This framework protects MWMS from the following forms of drift:

Repeating the same prompt work manually every session
Treating tools as skills
Treating one-off instructions as permanent procedures
Creating vague AI Employees with no reusable process
Letting AI Employees work without procedural boundaries
Allowing skill duplication across Brains
Allowing skills to bypass validation
Allowing skills to override tool permissions
Letting skills become outdated without review
Creating client-facing AI workflows without reusable safe procedures
Automating procedures before they are proven manually
Losing valuable workflow knowledge inside chat threads
Creating output consistency problems between AI Employees
Treating skill volume as progress
Letting AI workforce growth become ungoverned

Any repeated AI task without a defined skill should be reviewed for possible skill creation.

Architectural Intent

The architectural intent of the MWMS AI Agent Skill Library Framework is to give MWMS reusable procedural memory.

MWMS is not building a pile of prompts.

MWMS is building a governed AI workforce.

That workforce needs reusable skills.

The long-term goal is that every important AI Employee can answer:

What skills do I have?
When should I use each skill?
What input does the skill require?
What procedure should I follow?
What standards apply?
What output should I create?
What must I avoid?
How is the output validated?
Where does the result go?
What failure triggers stop the skill?
What outcome should the skill create?
How does the skill improve over time?

When MWMS can define and improve skills this way, AI Employees become more consistent, more useful, safer, easier to train, and easier to package into future AIBS client systems.

Change Log

v1.0 — Initial Draft

Created the MWMS AI Agent Skill Library Framework as the procedural memory layer for the MWMS AI Agent Operations Core.

This framework defines AI Agent Skills, skill identity, triggers, inputs, procedures, outputs, validation, handoffs, improvement rules, skill types, skill record template, governance rules, creation criteria, review cycle, statuses, future plugin/UI relevance, governance role, drift protection, and architectural intent.

It establishes that tools give AI Employees hands, while skills teach AI Employees how to work.