AI Glossary
Plain-language definitions for models, prompts, context, agents, tools, retrieval, and evaluations.
Core concept
- Audience
- Learner, Practitioner
- Level
- Beginner
- Prerequisites
- None
- Owner
- Project42 Editorial
CurrentReviewed
Open →Project 42 Field Guide
Answers for the work in front of you.
Practical references with visible sources and verification dates. Learn a concept, check a workflow, or compare providers without digging through a full course.
Showing 50 of 50 resources
Prompt Checklist
A compact preflight for tasks, context, constraints, output, and verification.
Core concept
- Audience
- Learner, Practitioner
- Level
- Beginner
- Prerequisites
- None
- Owner
- Project42 Editorial
CurrentReviewed
Open →Agent Safety Checklist
Permission, data, verification, logging, and rollback questions before an agent acts.
Core concept
- Audience
- Developer, Operator
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Provider Decision Card
A reusable comparison card for quality, cost, data, tools, and operations.
AnthropicOpenAIGoogle
- Audience
- Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Daily AI Task Starter
A reusable prompt and working loop for turning ordinary work into a clear, verifiable AI task.
Core concept
- Audience
- Learner, Practitioner
- Level
- Beginner
- Prerequisites
- None
- Owner
- Project42 Editorial
CurrentReviewed
Open →Context Budget Playbook
Keep long AI sessions focused by loading the smallest useful working set and refreshing stale context deliberately.
Core conceptAnthropicOpenAI
- Audience
- Practitioner, Developer
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →AI Output Verification Ladder
Choose a level of checking that matches the claim, uncertainty, and consequence of the work.
Core concept
- Audience
- Learner, Practitioner, Developer
- Level
- Beginner
- Prerequisites
- None
- Owner
- Project42 Editorial
CurrentReviewed
Open →AI-Assisted Code Review Checklist
Review AI-assisted changes for correctness, security, evidence, maintainability, and scope without treating model output as approval.
Core conceptAnthropicOpenAIGoogle
- Audience
- Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 2 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →AI Coding-Agent Permission Boundaries
Define least-privilege tool access, trust boundaries, approval gates, and postconditions for coding-agent work.
Core conceptAnthropicOpenAIGoogle
- Audience
- Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 2 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →AI Coding-Agent Work Plan Template
Turn a requested outcome into a bounded, dependency-aware plan with acceptance evidence and explicit approval gates.
Core conceptAnthropicOpenAIGoogle
- Audience
- Practitioner, Developer, Operator, Leader
- Level
- Beginner
- Prerequisites
- 2 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Implementation Evidence Loop
Implement coding-agent changes in small increments that continuously prove scope, behavior, safety, and recoverability.
Core concept
- Audience
- Practitioner, Developer, Operator
- Level
- Intermediate
- Prerequisites
- 2 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Repository Orientation Checklist
Build a verified map of a repository before asking an AI coding agent to plan or modify it.
Core conceptAnthropicOpenAIGoogle
- Audience
- Practitioner, Developer, Operator
- Level
- Beginner
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Test and Debug Handoff
Transfer a failing or partially verified coding-agent change with a reproducible state, bounded hypothesis, and clear next action.
Core concept
- Audience
- Practitioner, Developer, Operator
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Bounded AI Red-Team Exercise
Plan and run authorized adversarial tests against the complete AI application while protecting people, data, systems, and downstream tools.
Core concept
- Audience
- Developer, Operator, Leader
- Level
- Advanced
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →AI Evaluation Plan Charter
Turn an AI use case into a versioned evaluation plan with measurable outcomes, representative slices, blocking thresholds, accountable owners, and a repeatable decision process.
Core concept
- Audience
- Practitioner, Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 2 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Evidence-Based AI Evaluation Rubric
Design observable criteria, anchored score levels, calibrated graders, and blocking safety rules that make qualitative AI judgments reproducible.
Core concept
- Audience
- Practitioner, Developer, Educator, Leader
- Level
- Intermediate
- Prerequisites
- 2 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Human-Controlled AI Release Gate
Convert evaluation, safety, security, privacy, operational, and rollback evidence into an accountable promote, hold, or reject decision.
Core concept
- Audience
- Developer, Operator, Leader
- Level
- Advanced
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Representative AI Evaluation Dataset
Build a privacy-safe, versioned set of normal, boundary, adversarial, and incident-derived cases that measures the application people will actually use.
Core concept
- Audience
- Practitioner, Developer, Operator, Educator
- Level
- Intermediate
- Prerequisites
- 2 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Agent Handoff Evidence Contract
Transfer work between agents or people with explicit ownership, minimized context, verified state, acceptance evidence, and safe continuation rules.
Core conceptAnthropicOpenAIGoogle
- Audience
- Practitioner, Developer, Operator
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →MCP Primitives and Lifecycle Reference
Use MCP hosts, clients, servers, capabilities, transports, tools, resources, and prompts without confusing protocol roles or trust boundaries.
Core conceptAnthropicOpenAIGoogle
- Audience
- Learner, Practitioner, Developer, Operator
- Level
- Beginner
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →MCP Server Trust Review
Review an MCP server's provenance, code path, identity, data access, tools, authorization, and containment before connection.
Core conceptAnthropic
- Audience
- Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 2 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Agent Orchestration Pattern Decision Guide
Choose direct tools, deterministic workflows, manager-agent delegation, handoffs, parallel work, or human routing from control and evidence needs.
Core conceptAnthropicOpenAIGoogle
- Audience
- Practitioner, Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Agent Tool Contract Design Template
Design model-callable tools with a narrow purpose, explicit schemas, enforced authorization, predictable errors, and verifiable side effects.
Core conceptAnthropicOpenAIGoogle
- Audience
- Developer, Operator
- Level
- Intermediate
- Prerequisites
- 2 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Context Refresh Runbook
Refresh long-running AI work without carrying forward superseded decisions, stale facts, unsafe instructions, or unresolved side effects.
Core conceptAnthropicOpenAIGoogle
- Audience
- Practitioner, Developer, Operator
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Context Selection Checklist
Select the smallest trustworthy working set by relevance, authority, freshness, provenance, sensitivity, and contradiction value.
Core concept
- Audience
- Practitioner, Developer, Operator
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Prompt Pattern Decision Guide
Choose the smallest prompting pattern that matches the failure you need to prevent, then evaluate it against representative cases.
Core conceptAnthropicOpenAIGoogle
- Audience
- Learner, Practitioner, Developer
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Reusable Prompt Template Pack
Copy safe, provider-neutral templates for summarizing, transforming, comparing, researching, and reviewing work with visible evidence.
Core concept
- Audience
- Learner, Practitioner, Educator
- Level
- Beginner
- Prerequisites
- None
- Owner
- Project42 Editorial
CurrentReviewed
Open →Structured Output Contract
Design portable output contracts, use provider schema features where supported, and validate shape, meaning, authority, and evidence before downstream use.
Core conceptAnthropicOpenAIGoogle
- Audience
- Practitioner, Developer
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Task Framing Contract
Turn an unclear request into an outcome, authority boundary, evidence plan, and explicit stopping condition before asking an AI system to work.
Core concept
- Audience
- Learner, Practitioner, Developer
- Level
- Beginner
- Prerequisites
- None
- Owner
- Project42 Editorial
CurrentReviewed
Open →Anthropic Evaluation and Error Triage
Turn a Claude workflow into measurable tests, preserve request evidence, classify API failures, and release or recover through explicit gates.
Anthropic
- Audience
- Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Anthropic Messages API Request Workflow
Send a bounded Claude Messages API request with environment-based credentials, explicit configuration, response inspection, and recoverable failure handling.
Anthropic
- Audience
- Developer, Operator
- Level
- Beginner
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Anthropic Strict Tools and Structured Output
Choose between Claude JSON output and strict tool use, define constrained schemas, validate semantics, and keep tool execution under application control.
Anthropic
- Audience
- Developer, Operator
- Level
- Intermediate
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Cross-Provider Evaluation and Migration Workflow
Compare or migrate Anthropic, OpenAI, and Google workflows through one application-owned contract, provider-specific adapters, representative evaluation, staged cutover, and tested rollback.
Core conceptAnthropicOpenAIGoogle
- Audience
- Developer, Operator, Leader
- Level
- Advanced
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Cross-Provider Runtime Configuration Contract
Separate application intent from Anthropic, OpenAI, and Google credentials, models, API versions, state, capabilities, and operational limits without pretending their APIs are equivalent.
Core conceptAnthropicOpenAIGoogle
- Audience
- Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Google Gemini Evaluation and Error Triage
Build representative Gemini evaluation datasets, preserve interaction evidence, classify API failures, and release or recover through explicit gates.
Google
- Audience
- Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Google Gemini Interactions API Request Workflow
Send a bounded Gemini Interactions API request with environment-based credentials and model configuration, typed response inspection, and recoverable failure handling.
Google
- Audience
- Developer, Operator
- Level
- Beginner
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Google Gemini Functions and Structured Output
Choose between Gemini schema-constrained output and function calling, define narrow contracts, validate semantics, and keep execution under application control.
Google
- Audience
- Developer, Operator
- Level
- Intermediate
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →OpenAI Evaluation and Error Triage
Apply eval-driven development to an OpenAI workflow, preserve request evidence, classify failures, and release or recover through explicit gates.
OpenAI
- Audience
- Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →OpenAI Responses API Request Workflow
Send a bounded OpenAI Responses API request with environment-based credentials, explicit configuration, output inspection, and recoverable failure handling.
OpenAI
- Audience
- Developer, Operator
- Level
- Beginner
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →OpenAI Strict Functions and Structured Output
Choose between Responses structured text and strict function calls, define constrained schemas, validate semantics, and keep execution under application control.
OpenAI
- Audience
- Developer, Operator
- Level
- Intermediate
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Citation Support Checklist
Check that every citation resolves, identifies the intended source, and supports the nearby claim without scope, date, or attribution drift.
Core conceptAnthropicOpenAIGoogle
- Audience
- Learner, Practitioner, Educator
- Level
- Beginner
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Claim Decomposition Map
Break a broad answer into atomic factual, comparative, causal, predictive, and normative claims so each can receive the evidence and review it needs.
Core concept
- Audience
- Learner, Practitioner, Developer, Educator
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Consequence-Based Review Gate
Scale evidence, independence, expertise, approval, monitoring, and rollback to the realistic harm caused by an incorrect AI-assisted result.
Core concept
- Audience
- Practitioner, Developer, Operator, Leader
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Fact Verification Workflow
Verify material facts through bounded questions, current primary evidence, contradiction searches, reproducible checks, and explicit uncertainty.
Core concept
- Audience
- Learner, Practitioner, Developer, Operator
- Level
- Intermediate
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Source Authority Ladder
Choose evidence by ownership, directness, currency, scope, and independence instead of treating every search result as equally trustworthy.
Core concept
- Audience
- Learner, Practitioner, Developer, Educator
- Level
- Beginner
- Prerequisites
- 1 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Agent Tool Failure Recovery
Diagnose tool selection, schema, authorization, execution, result-matching, timeout, and side-effect failures without allowing an agent to retry blindly.
Core concept
- Audience
- Developer, Operator
- Level
- Advanced
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →AI API Failure Triage
Classify authentication, request, quota, timeout, rate-limit, provider, and application failures before retrying or escalating an AI API call.
Core concept
- Audience
- Developer, Operator
- Level
- Intermediate
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →AI Incident Triage
Stabilize an AI-enabled service, classify impact, preserve privacy-safe evidence, reconcile uncertain actions, and route decisions to accountable humans.
Core concept
- Audience
- Developer, Operator, Leader
- Level
- Advanced
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →AI Rollback and Incident Closeout
Restore a verified AI workflow, reconcile state and side effects, prove recovery, and close an incident only after corrective work has accountable evidence.
Core concept
- Audience
- Developer, Operator, Leader
- Level
- Advanced
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →Context and Quality Regression Triage
Isolate whether an AI quality regression comes from task distribution, prompts, context, retrieval, model behavior, tools, graders, or application changes.
Core concept
- Audience
- Practitioner, Developer, Operator
- Level
- Advanced
- Prerequisites
- 3 listed
- Owner
- Project42 Editorial
CurrentReviewed
Open →