Persona / Semantic
Actionable
Shifts output from theory into executable steps, verbs, and timelines.
When to use: Plans, tutorials, meeting takeaways, next-step recommendations.
Make this actionable by next Tuesday.
← Resource Library
The Trigger-Word Directory
All 128 prompt trigger words in one searchable directory. Copy the example phrase from any card.
128 words
Instruction Verbs
Adapt
Modifies content for a specific audience, channel, format, or use case.
When to use: Use when the source is useful but needs to fit a different context.
Adapt this section for a LinkedIn post.
Instruction Verbs
Analyze
Directs the AI to examine source material carefully before producing an answer.
When to use: Use when the model should inspect patterns, evidence, structure, or meaning before writing.
Read and analyze all provided transcripts before creating the summary.
Prompt Operations
Append
Adds new material to the end of an existing output without rewriting the earlier content.
When to use: Use when you want the AI to preserve what exists and add a new section, list item, paragraph, or example after it.
Append a section called “Risks to verify” to the end of the draft.
Skill Design
Ask me questions until you have enough information to successfully complete the task
A clarification safety phrase that prevents the AI from guessing when required context is missing.
When to use: Use at the end of complex prompts, reusable skills, or templates where missing information would lower quality or cause hallucination.
Ask me questions until you have enough information to successfully complete the task.
Instruction Verbs
Assess
Examines condition, readiness, risk, or fit.
When to use: Use for readiness checks, risk reviews, or quality checks.
Assess whether this draft is ready to publish.
Skill Design
Assumptions
States what the skill may treat as true when information is missing or implied.
When to use: Use when a skill needs safe defaults without inventing facts.
Assumptions: if no channel is provided, prepare the output for an internal Notion page.
Drift Prevention
Attention Anchoring
Repeats or places the most important constraint where the model is likely to weight it strongly.
When to use: Long prompts, production agents, strict style or safety constraints.
End every response with a status check against the core objective.
Data Analytics
Attribution
The method used to assign credit for an outcome to one or more touchpoints, channels, or actions.
When to use: Marketing performance analysis, funnel reporting, ROI conversations.
Explain the attribution logic before interpreting ROI.
Context Engineering
Audience Context
Information about who the output is for, including their knowledge level, role, concerns, and language.
When to use: Use when the same content needs to land differently for beginners, executives, technical users, or clients.
Audience context: small agency owners who understand analytics but are new to prompt engineering.
Instruction Verbs
Audit
Systematically checks something against standards, requirements, or expected behavior.
When to use: Use when you need a structured inspection rather than a general review.
Audit this prompt for ambiguity, missing constraints, and output risks.
Negative Constraint
Avoid Jargon
Forces simpler language and reduces technical or corporate vocabulary.
When to use: Broad audiences, student-facing content, executive summaries.
Avoid jargon. Use plain language.
Data Analytics
Baseline
The normal or expected level of performance used as the comparison point for future changes.
When to use: Before evaluating growth, decline, campaign lift, or anomaly claims.
Establish the baseline before interpreting this change.
Instruction Verbs
Benchmark
Compares something against a baseline, standard, peer set, or expected range.
When to use: Use when evaluating whether performance or quality is good, weak, or typical.
Benchmark this output against the examples in the style guide.
Instruction Verbs
Categorize
Sorts items into meaningful categories, usually based on labels or themes.
When to use: Use when the output needs a clear taxonomy or bucketed structure.
Categorize these terms by prompt function.
Data Analytics
Causation
A relationship where evidence supports that one factor directly contributed to an outcome.
When to use: Testing claims, experiment analysis, decision justification.
Separate correlation from causation in this analysis.
Reasoning
Chain-of-Thought
Allocates more reasoning effort before the final answer, reducing errors in complex tasks.
When to use: Complex logic, multi-step analysis, technical work, debugging.
Use chain-of-thought reasoning internally, then give me the answer.
Grounding
Citations
Forces claims to be anchored to source documents, URLs, or quoted evidence.
When to use: Research, fact-checking, professional reporting, claim verification.
Include citations for each factual claim.
Instruction Verbs
Classify
Assigns each item to a defined label based on criteria.
When to use: Use when you have known labels and need consistent assignment.
Classify each comment as question, objection, praise, or confusion.
Prompt Operations
Cluster
Asks the AI to find natural groups of related items based on similarity, theme, behavior, or shared characteristics.
When to use: Use when you have messy notes, keywords, comments, transcripts, or terms and want the AI to discover meaningful patterns before naming categories.
Cluster these customer comments by underlying pain point, then name each cluster.
Data Analytics
Cohort Analysis
Groups users, customers, or records by a shared starting condition so behavior can be compared over time.
When to use: Retention, lifecycle analysis, onboarding performance, repeat behavior.
Analyze this by cohort, not as one blended average.
Grounding
Comparative
Triggers the model to examine opposing or different data points for balance.
When to use: Decision-making, buying guides, neutral reporting, tradeoff analysis.
Give me a comparative analysis of the options.
Instruction Verbs
Compare
Shows similarities and differences between items.
When to use: Use when the reader needs to understand tradeoffs or options.
Compare these two prompt versions.
Skill Design
Comprehensive
Requires the AI to cover the relevant scope thoroughly and avoid leaving out important details.
When to use: Use when completeness matters more than brevity, such as research synthesis, revised guides, or protocols.
Create a comprehensive revised guide that covers all major conflicts and resolutions.
Formatting
Concise
Penalizes wordiness and removes low-signal chatter.
When to use: Short summaries, social copy, mobile-friendly output.
Be concise. No preamble.
Prompt Operations
Condense
Shortens content while preserving the most important meaning and structure.
When to use: Use when a draft is too long, repetitive, or needs to fit a constrained format.
Condense this into three bullets without losing the core distinction.
Data Analytics
Confidence Interval
A range that estimates where the true value is likely to fall, given the sample data.
When to use: Forecasting, experiment interpretation, uncertainty communication.
Include a confidence interval or explain why one cannot be calculated.
Reasoning
Consider
Expands or frames the AI’s reasoning by telling it what context, nuance, constraints, or implications should be weighed.
When to use: Use when the answer should account for conditions beyond the immediate source text.
Consider the full context, audience maturity, and cultural nuances before recommending changes.
Context Engineering
Constraint
A rule or boundary that limits how the AI can complete the task.
When to use: Use when output must follow style, length, format, legal, data, or brand requirements.
Constraint: do not use em dashes and do not invent statistics.
Context Engineering
Context Block
A clearly marked section that tells the AI what background information should inform the answer.
When to use: Use when the AI needs source facts, audience details, project constraints, or brand rules before completing the task.
Use the context block below before drafting the answer.
Context Engineering
Context Hierarchy
The priority order that tells the AI which context wins when instructions or sources conflict.
When to use: Use when a prompt includes multiple sources, brand rules, stakeholder notes, or older drafts.
Context hierarchy: brand guide first, source transcript second, draft notes third.
Context Engineering
Context Rot
The gradual loss or dilution of important instructions as a conversation gets longer and newer messages dominate attention.
When to use: Use when explaining why long AI chats drift, contradict earlier instructions, or start producing generic output.
Check for context rot and re-anchor the original constraints before continuing.
Context Engineering
Context Window
The amount of information the AI can consider at one time during a conversation or task.
When to use: Use when planning long prompts, large documents, multi-turn workflows, or when the model starts forgetting earlier instructions.
Keep the answer within the current context window and summarize anything that may be lost.
Grounding
Context-Aware
Prioritizes provided input over generic model knowledge or training patterns.
When to use: RAG, source-grounded work, brand voice tasks, internal docs.
Be context-aware and use only the provided material.
Instruction Verbs
Contrast
Emphasizes differences between items, approaches, or outcomes.
When to use: Use when the distinction matters more than the overlap.
Contrast a vague prompt with an engineered prompt.
Instruction Verbs
Convert
Changes content from one format or unit into another.
When to use: Use when the underlying information stays the same but the format changes.
Convert this outline into a checklist.
Instruction Verbs
Correlate
Looks for variables or themes that move together without assuming one caused the other.
When to use: Use for hypothesis generation and pattern detection.
Correlate the objections with customer segment, but do not imply causation.
Data Analytics
Correlation
A relationship where two variables move together, without proving that one caused the other.
When to use: Pattern discovery, hypothesis generation, early-stage analysis.
Identify correlations, but do not imply causation.
Skill Design
Criteria
The standards used to judge whether the output is good, complete, or acceptable.
When to use: Use when a skill needs an internal quality check or decision rule.
Criteria: each term must explain its effect inside a prompt and include an example phrase.
Reasoning
Deconstruct
Forces the model to identify the parts of a problem before drawing conclusions.
When to use: Audits, complex diagnosis, technical reviews, strategy breakdowns.
Deconstruct the workflow before recommending changes.
Reasoning
Deep-dive
Shifts the model from summary-level response to exhaustive exploration.
When to use: Research, technical documentation, comprehensive reports.
Deep-dive into the operational risks.
Skill Design
Deliver
Signals that the AI should produce the finished artifact or final packaged response.
When to use: Use when the prompt is moving from analysis into a completed template, document, plan, or output.
Deliver a complete prompt template with role, context, process, and output format.
Instruction Verbs
Diagnose
Identifies the likely cause of a problem from available evidence.
When to use: Use when something is broken, unclear, or underperforming.
Diagnose why this prompt is producing generic output.
Instruction Verbs
Distill
Reduces complex material to the essential signal or core takeaway.
When to use: Use when a draft or source is noisy and the key point needs to stand out.
Distill this into the one decision the reader needs to make.
Negative Constraint
Do Not Apologize
Prevents the model from entering a politeness loop when it hits a constraint.
When to use: Error handling, refusal protocols, operational assistants.
Do not apologize. State the issue and the next step.
Skill Design
Do not use when
Defines the conditions where the skill should not activate, even if some words seem related.
When to use: Use to prevent false triggers and keep skills from overreaching.
Do not use when the user only asks for a quick definition.
Skill Design
Ensure
Creates a quality-control requirement the final output must satisfy.
When to use: Use when the AI should verify that the deliverable meets a specific standard before finishing.
Ensure each protocol is practical, specific, and directly tied to source evidence.
Instruction Verbs
Evaluate
Judges quality, fit, strength, or value against criteria.
When to use: Use when you need an assessment, not just a description.
Evaluate this prompt against clarity, constraints, and output quality.
Prompt Operations
Expand
Adds more detail, examples, reasoning, or supporting explanation to existing content.
When to use: Use when the answer is too thin, assumes too much knowledge, or needs stronger educational value.
Expand this section with one concrete agency example.
Persona / Semantic
Expert
Sets the baseline knowledge level, judgment, vocabulary, and standards the AI should use.
When to use: Use in role sections when the task requires domain-specific authority or professional judgment.
Act as an expert presentation designer and information architect.
Prompt Operations
Extract
Pulls specified information out of source material into a cleaner list, table, or structure.
When to use: Use when mining transcripts, research notes, drafts, reports, or examples for reusable material.
Extract all trigger words, definitions, and example uses into a table.
Skill Design
Fallback
The alternate action the AI should take when ideal inputs, tools, or source material are missing.
When to use: Use to keep a skill useful without letting it hallucinate or stall.
Fallback: if source examples are missing, produce a blank template and mark examples as needed.
Context Engineering
Few-shot Examples
A small set of examples that show the AI the desired pattern, format, voice, or reasoning style.
When to use: Use when instructions alone are not enough and you need the model to imitate a pattern.
Use these few-shot examples to match the structure and level of detail.
Data Analytics
Granularity
The level of detail in the data, such as user-level, session-level, campaign-level, or monthly totals.
When to use: When deciding whether the data is detailed enough to answer the question.
Check whether the data granularity supports this conclusion.
Prompt Operations
Group by
Asks the AI to organize items under a specific field, label, dimension, or category that you define.
When to use: Use when you already know the organizing rule, such as channel, audience, stage, category, status, or theme.
Group these prompt terms by function: reasoning, formatting, grounding, and editing operations.
Instruction Verbs
Identify
Finds and names relevant items, patterns, risks, entities, or gaps.
When to use: Use when the AI should locate important signals inside source material.
Identify the main risks in this launch plan.
Drift Prevention
Ignore Previous
Resets the working focus for a sub-task and re-applies the current instruction set.
When to use: When a long chat has drifted or accumulated irrelevant context.
Ignore previous conversational noise and re-apply these constraints.
Skill Design
Inputs
The information the skill needs before it can produce a useful output.
When to use: Use when designing a skill that should ask for or expect specific source material, audience, format, or constraints.
Required inputs: source text, target audience, output format, and tone rules.
Prompt Operations
Insert
Places new material at a specific point inside existing content.
When to use: Use when the addition belongs in the middle of a draft, outline, table, prompt, or workflow rather than at the beginning or end.
Insert a short example after the second bullet without changing the other bullets.
Persona / Semantic
Insightful
Pushes the model beyond surface facts toward synthesis and connection-making.
When to use: Strategy documents, trend analysis, thought leadership.
Give me an insightful synthesis, not a summary.
Instruction Verbs
Interrogate
Challenges the logic, assumptions, evidence, and implications behind an idea.
When to use: Use when you want deeper scrutiny instead of supportive agreement.
Interrogate the assumptions behind this recommendation.
Formatting
JSON / Schema
Forces strict programmatic syntax and prioritizes structure over conversational filler.
When to use: API workflows, data extraction, automations, integrations.
Output only valid JSON matching this schema.
Grounding
Latest / Recent
Signals the model or search tool to prioritize timestamped, current information.
When to use: Market research, news, technical updates, current platform behavior.
Use the latest available sources.
Instruction Verbs
List
Returns items plainly, usually without extended explanation.
When to use: Use when you need quick inventory, extraction, or scannable output.
List the assumptions in this strategy.
Anti-sycophancy
Logical Friction
Forces the model to find flaws, risks, or weak assumptions before agreeing.
When to use: Reviewing plans, offers, messaging, or strategic assumptions.
Apply logical friction to this plan.
Formatting
Markdown
Asks the model to use headers, lists, tables, and code blocks for readable structure.
When to use: Documentation, README files, structured reports.
Format the answer in Markdown.
Anti-sycophancy
Neutrality Buffer
Presents multiple perspectives to reduce one-sided or overly agreeable answers.
When to use: Subjective topics, decision support, policy questions, tradeoff analysis.
Use a neutrality buffer and show three perspectives.
Negative Constraint
No Preamble
Skips filler openings like “Sure, I can help” and goes directly to the result.
When to use: Production workflows, concise outputs, token-sensitive tasks.
No preamble. Directly provide the result.
Data Analytics
Normalization
Adjusts data into a consistent format or scale so comparisons are fair.
When to use: Comparing sources, cleaning inconsistent fields, preparing reports.
Normalize the naming conventions before comparing performance.
Instruction Verbs
Normalize
Makes inconsistent inputs comparable by standardizing names, formats, units, or structure.
When to use: Use before grouping, comparing, scoring, or analyzing messy source material.
Normalize these labels before grouping the records.
Grounding
Note
Highlights constraints, limitations, red flags, or special details the AI should explicitly account for.
When to use: Use when a prompt needs the AI to call out caveats or avoid missing important boundaries.
Note any limitations in your knowledge and any phrases to avoid.
Persona / Semantic
Nuanced
Penalizes generic completions and pushes the model toward specificity and sophistication.
When to use: Professional analysis, thought leadership, sensitive tradeoffs.
Give me a nuanced analysis of the tradeoffs.
Persona / Semantic
Objective
Flattens sentiment and reduces agreeable or biased framing.
When to use: Risk reviews, evaluations, neutral analysis, decision support.
Evaluate this objectively.
Context Engineering
Operating Context
The current working frame for the AI, including role, objective, source material, constraints, and expected output.
When to use: Use when you want the AI to stay aligned across a multi-step task.
Operating context: you are editing a prompt glossary for agency operators.
Data Analytics
Outlier
A data point that sits far outside the expected range and may distort interpretation.
When to use: Trend analysis, forecasting, reporting QA, anomaly review.
Identify outliers before summarizing the trend.
Skill Design
Output
Specifies the final deliverable format the AI should produce.
When to use: Use in execution sections when you need a structured document, table, template, or artifact.
Output a structured document with key findings and actionable protocols.
Skill Design
Outputs
The deliverables or response format the skill should produce.
When to use: Use when a skill needs consistent output, such as a table, checklist, draft, audit, or file.
Output: a Markdown table with Term, Effect, When to Use, and Example Prompt.
Persona / Semantic
Persona: [role]
Weights the output toward vocabulary and norms associated with a specific role or audience.
When to use: Tailoring communication to a professional standard or audience lens.
Persona: senior agency COO.
Prompt Operations
Prepend
Adds new material to the beginning of an existing output before the current content.
When to use: Use when the AI should add an intro, summary, warning, context note, or framing section without changing the rest.
Prepend a two-sentence executive summary before the current analysis.
Prompt Operations
Preserve
Tells the AI what must stay unchanged while other edits are made.
When to use: Use when exact language, formatting, citations, names, voice, or approved sections should not be altered.
Preserve the quoted client language exactly, but revise the surrounding explanation.
Instruction Verbs
Prioritize
Determines what should be handled first based on explicit criteria.
When to use: Use when you need a practical next-step order, not just a list.
Prioritize these fixes by customer impact and implementation effort.
Skill Design
Procedure
The ordered steps the AI should follow when using the skill.
When to use: Use when consistency matters and the model should not improvise the workflow each time.
Procedure: inspect the source, extract terms, classify by function, then write examples.
Skill Design
Provide
Directs the AI to give a specified deliverable, explanation, overview, or set of materials.
When to use: Use in execution sections when the final response should include a concrete answer or artifact.
Provide a structured, comprehensive overview of the findings.
Persona / Semantic
Provocative
Encourages the model to take a stand or explore unconventional perspectives.
When to use: Brainstorming, editorial writing, debate preparation.
Give me a provocative angle on this issue.
Instruction Verbs
Rank
Orders items by relative strength, importance, relevance, or fit.
When to use: Use when you want the AI to make a judgment call using criteria.
Rank these ideas by revenue potential and ease of implementation.
Reasoning
Rationale
Requires the model to explain the why behind an answer or recommendation.
When to use: When you need transparency, verification, or decision logic.
Give your recommendation with rationale.
Instruction Verbs
Read
Directs the AI to process the provided source material as the first step.
When to use: Use when you want to prevent the AI from jumping straight to generation before absorbing the input.
Read the full source material before identifying key concepts.
Prompt Operations
Refactor
Restructures content or logic without changing the intended outcome.
When to use: Use for prompts, code, SOPs, workflows, outlines, or messy drafts that need cleaner organization.
Refactor this prompt into Role, Objective, Process, Constraints, and Output Format.
Context Engineering
Reference Material
The source content the AI should use to complete the task, such as notes, transcripts, data, brand guides, or examples.
When to use: Use when the AI must work from specific material instead of general knowledge.
Use the reference material below. Do not add outside concepts unless clearly labeled.
Instruction Verbs
Reframe
Changes the lens, angle, or interpretation of the same information.
When to use: Use when the facts are right but the positioning or mental model needs to shift.
Reframe this as an Expert Bottleneck problem.
Prompt Operations
Remove
Deletes specified content and signals that it should not be replaced unless requested.
When to use: Use when cutting filler, banned language, unsupported claims, duplicated sections, or irrelevant details.
Remove the generic AI hype language and leave the practical explanation.
Prompt Operations
Replace
Swaps one piece of content for another and signals that the original should not remain.
When to use: Use when a phrase, section, format, or assumption is wrong and needs a direct substitution.
Replace “AI replaces judgment” with “AI accelerates judgment.”
Instruction Verbs
Repurpose
Turns one asset or source into another content type or use case.
When to use: Use when making derivative content from a core piece.
Repurpose this article into three LinkedIn posts.
Skill Design
Required fields
The minimum pieces of information needed before the skill can run reliably.
When to use: Use when the AI should ask a clarifying question instead of guessing.
Required fields: audience, source material, desired format, and success criteria.
Prompt Operations
Revise
Improves existing content while preserving its core intent and structure.
When to use: Use when the idea is right but the wording, clarity, tone, or flow needs improvement.
Revise this paragraph for clarity while preserving the original meaning.
Instruction Verbs
Rewrite
Creates a new wording of existing content while preserving the intended message.
When to use: Use when tone, clarity, rhythm, or audience fit needs improvement.
Rewrite this in a warmer voice, without changing the meaning.
Data Analytics
Sample Size
The number of observations used in an analysis, which affects reliability and confidence.
When to use: Experiment design, survey interpretation, performance reporting.
Check whether the sample size is large enough to support this conclusion.
Context Engineering
Scope
Defines what the AI should include, exclude, or stay within for the task.
When to use: Use when you need to prevent the AI from expanding into unrelated ideas or solving the wrong problem.
Scope this answer to prompt-writing mechanics, not AI ethics or model history.
Skill Design
Scope (Skill)
The boundary of what the skill is responsible for and what it should not attempt.
When to use: Use when a skill needs clear limits so it does not overreach or solve the wrong task.
Scope: this skill only handles prompt vocabulary, not general AI history.
Instruction Verbs
Segment
Breaks results or content into meaningful groups for clearer analysis.
When to use: Use when aggregate output hides important differences by audience, channel, stage, or behavior.
Segment these examples by audience maturity level.
Data Analytics
Segmentation
Divides data into meaningful groups so patterns are not hidden inside aggregate totals.
When to use: Audience analysis, campaign performance, channel comparison, ICP research.
Segment the results by channel and audience type.
Instruction Verbs
Sort
Orders items by a specified rule such as date, priority, impact, or sequence.
When to use: Use when order matters more than grouping.
Sort these tasks by urgency, highest first.
Context Engineering
Source of Truth
The document, dataset, or instruction set the AI should treat as authoritative when conflicts appear.
When to use: Use when multiple sources exist and you need to prevent the AI from blending outdated or lower-priority information.
Treat the brand guide as the source of truth if any sources conflict.
Skill Design
Specific
Forces precision and discourages vague, generic, or overly broad output.
When to use: Use when the output needs concrete steps, examples, constraints, or recommendations.
Give specific, actionable suggestions tied to the source material.
Drift Prevention
State Serialization
Captures the model's current operating mode as structured state so constraints stay active across turns.
When to use: Multi-turn workflows, production assistants, agent handoffs.
At the end, serialize the current operating mode as JSON.
Data Analytics
Statistical Significance
A measure of whether an observed difference is likely to be real rather than random noise.
When to use: A/B tests, experiment results, performance claims.
Flag any claim that lacks statistical significance.
Anti-sycophancy
Steel-man
Finds the strongest version of an opposing view before responding.
When to use: Strategy debate, positioning, critique, decision stress-testing.
Steel-man the opposing argument before you respond.
Reasoning
Step-by-step
Forces the model to generate intermediate reasoning steps before the final answer.
When to use: Complex problem solving, coding logic, multi-layered analysis.
Explain the decision step-by-step.
Instruction Verbs
Stress-test
Looks for weak points, edge cases, failure modes, and ways the output could break.
When to use: Use before relying on a prompt, workflow, strategy, or claim.
Stress-test this prompt and identify where it may fail.
Formatting
Structured
Forces the AI to organize output into clear sections, headings, tables, steps, or other readable architecture.
When to use: Use when you want to prevent a wall of text and make the output easier to scan or reuse.
Output a structured presentation outline with sections and bullets.
Instruction Verbs
Summarize
Compresses source material into the main points.
When to use: Use when you need the gist without the full detail.
Summarize this transcript in five bullets.
Instruction Verbs
Surface
Brings hidden, implied, or easy-to-miss points into view.
When to use: Use when you suspect risks, assumptions, or patterns are buried in the material.
Surface the unstated assumptions in this recommendation.
Instruction Verbs
Synthesize
Combines multiple inputs into a coherent conclusion or framework.
When to use: Use when source material is scattered across notes, transcripts, or research.
Synthesize these notes into one clear operating principle.
Reasoning
Systematic
Weights the response toward method, order, classification, and repeatable process.
When to use: Workflow creation, audits, SOPs, technical writing.
Analyze this systematically.
Formatting
Tabular
Triggers comparative logic and organizes information into rows and columns.
When to use: Comparisons, pros and cons, feature matrices.
Summarize this in a tabular format.
Instruction Verbs
Tighten
Makes wording shorter, sharper, and less repetitive.
When to use: Use when a draft has the right idea but too much padding.
Tighten this paragraph without losing the main point.
Prompt Operations
Transform
Changes content from one format, style, or structure into another while keeping the underlying information.
When to use: Use when turning notes into a post, a table into a narrative, or a prompt into an SOP.
Transform this rough note into a structured prompt template.
Instruction Verbs
Translate
Changes language, technical level, tone, or audience framing.
When to use: Use when content needs to move from expert language to plain language, or from one audience to another.
Translate this technical explanation for a non-technical agency owner.
Instruction Verbs
Trend
Examines movement or pattern change over time.
When to use: Use when timing, sequence, growth, decline, or direction matters.
Trend these observations over time and call out the inflection points.
Skill Design
Trigger
The word, phrase, or situation that tells the AI when a skill should activate.
When to use: Use when defining exactly when a reusable skill should be applied and when it should stay quiet.
Trigger this skill when the user asks to turn source material into a publishing-ready asset.
Anti-sycophancy
Truth > Alignment
Gives the model permission to disagree with the user's framing when accuracy requires it.
When to use: High-stakes review, risk assessment, quality control.
Prioritize truth over alignment with my premise.
Skill Design
Use when
Defines the conditions where the skill should be applied.
When to use: Use in skill descriptions so the AI can decide whether the skill matches the task.
Use when the user asks to create, edit, or QA an HTML artifact.
Instruction Verbs
Validate
Checks whether a claim, output, or conclusion is supported by evidence or rules.
When to use: Use before trusting a factual claim, data interpretation, or generated output.
Validate this conclusion against the source data.
Skill Design
Validation
The check the AI performs to confirm the output follows the skill requirements.
When to use: Use when a skill produces files, structured data, prompts, or reusable templates.
Validation: confirm every row has a term, effect, use case, and example.
Grounding
Verify
Activates a self-checking loop against constraints, facts, or source material.
When to use: Data-heavy tasks, accuracy-critical work, QA passes.
Verify each claim against the source material.
Persona / Semantic
World-class
Raises the expected expertise, synthesis quality, and professional standard of the AI’s response.
When to use: Use in role sections when the task requires top-tier judgment, strategy, design, research, or synthesis.
Act as a world-class knowledge synthesis expert.
Formatting
YAML
A human-readable structured format that uses indentation and key-value pairs to organize instructions, variables, examples, and metadata.
When to use: Use when you want a prompt to be readable by humans but structured enough for an AI to follow consistently. Good for system prompts, reusable prompt templates, configuration blocks, and content briefs.
Use this YAML context block as the source of truth for the task.
Negative Constraint
Zero Hallucination
High-weight instruction to stay inside provided context or say when information is missing.
When to use: Sensitive facts, internal documentation, evidence-based writing.
Zero hallucination. If it is not in the source, say so.
No matching words. Try another search or group.