D&D 5e Ability Checks & Skills Questions for Dungeon Masters

Player decisions should establish the approach to a social encounter, while character abilities should resolve uncertain execution. A persuasive player should not automatically bypass a character's poor Charisma, and a reserved player should not be penalized for being unable to perform the eloquent speech their character could deliver.

Begin by asking what the character wants and how they intend to obtain it. Offering a merchant profitable information, threatening an official with exposure, and appealing to a priest's beliefs are different strategies even when all involve conversation. The NPC's motives and circumstances determine whether each approach is plausible.

If the outcome remains uncertain, choose the appropriate ability check and relevant proficiency. A convincing argument may change what the NPC is willing to consider, whether a roll is necessary, or how difficult the attempt becomes. It should not automatically replace the character's mechanical abilities.

Consider a player who proposes an excellent bargain but whose character has low Charisma. The proposal might establish favorable negotiating conditions without guaranteeing acceptance. Conversely, a highly charismatic character cannot persuade someone to provide information they do not possess.

This approach rewards player creativity without requiring players to personally reproduce their characters' fictional talents.

Rules reference: D&D 5e SRD 5.2.1 — Ability Checks and Skills.

Call for an ability check when the character's approach could plausibly succeed, could meaningfully fail, and the outcome is uncertain enough that resolving it matters. Do not make players roll for routine actions with no pressure, and do not use a roll to disguise an outcome that the fiction has already made impossible.

Player skill should determine decisions: what to attempt, what information to use, what questions to ask, and what risks to accept. Character skill should determine capabilities the person at the table is not expected to possess. A shy player should not need real-world eloquence to benefit from a persuasive character, just as a player should not need to lift a door physically to use Strength. Good description can change the circumstances of a check—revealing leverage, avoiding a hazard, or eliminating uncertainty—but it should not replace the character sheet whenever the action depends on trained capability.

If the player's method fully resolves the problem, skip the roll. If it only improves the odds, reflect that in the difficulty, Advantage, or available consequences.

Rules reference: D&D 5e SRD 5.2.1 — D20 Tests.

No. Appearance does not determine Charisma, and a frightening character can have high Charisma. SRD 5.2.1 describes Charisma as measuring confidence, poise, and charm; those qualities can produce persuasion, command, intimidation, or presence without requiring the character to be conventionally attractive or pleasant.

Likewise, a low Charisma score does not require a stutter, illiteracy, or any specific disability. The rules describe scores from 2–9 as a weak capability and 10–11 as the human average, but they do not prescribe a personality or mannerism for each number.

For roleplay, decide what happens when the character tries to make an impression. A scarred undead with high Charisma might be composed, magnetic, terrifyingly articulate, or impossible to ignore. One with low Charisma might be unsettling but poor at directing that reaction, easily misunderstood, or socially withdrawn.

NPCs can still react to an unusual appearance according to the situation. That fictional reaction is separate from the character’s underlying Charisma score.

Rules reference: D&D 5e SRD 5.2.1 — Ability Scores and Charisma.

Multiple sources of Advantage do not normally turn into extra d20s, and multiple sources of Disadvantage do not normally deepen the penalty. When both Advantage and Disadvantage apply to the same roll, they cancel rather than being counted source by source. That keeps the probability system intentionally simple.

The useful design question is what additional teamwork does after one source has already produced the roll state. Let later contributions change a different part of the situation. One character can impose Disadvantage while another blocks an escape route, removes cover, protects an ally, forces movement, spends a resource to create a follow-up opportunity, or changes what failure will cost.

This preserves the rule while ensuring that the second and third helpful ideas are not meaningless. It also gives the DM a way to reward strong fictional positioning without inventing escalating three-die or four-die penalties.

When adjudicating an unusual combination, identify which effect actually grants Advantage or Disadvantage and then resolve every other contribution through its own stated effect or through the fictional situation rather than stacking the same mechanic repeatedly.

At the table, a useful distinction is to use Perception for noticing sensory information and Investigation for reasoning from evidence that has already been found. Perception answers questions such as “What do I notice?” while Investigation is better for “What does this arrangement imply?” or “How could this mechanism have been constructed?”

Avoid turning the two skills into interchangeable buttons. A character might use Perception to spot fresh scratches around a bookcase, then Investigation to infer that the case swings outward and locate the most likely trigger. Conversely, no Investigation check should be required to notice something that is plainly visible, and a failed Perception check should not automatically erase clues that the adventure requires to continue.

The practical rule is to identify the uncertainty before calling for a roll. If success depends on sensory acuity, lean toward Perception; if it depends on connecting facts, reconstructing events, or examining how pieces fit together, lean toward Investigation. When both matter, let one reveal the clue and the other deepen it rather than demanding two rolls for the same information.

Yes. D&D 5e allows ability checks without an associated skill proficiency. A Dungeon Master determines which ability is relevant to an uncertain task, then considers whether any skill proficiency or other applicable proficiency should modify the roll. If no relevant proficiency applies, the character makes a plain ability check using the appropriate ability modifier.

For example, a straight Strength check might measure the effort needed to force a stubborn mechanism. An Intelligence check could represent reasoning through an unfamiliar abstract relationship, while a Wisdom check might assess a judgment that does not naturally fall under a listed skill. The exact distinction depends on the task and the DM's adjudication.

Skills describe particular forms of proficiency; they do not represent an exhaustive menu of actions characters are permitted to attempt. Consequently, players should describe what their characters do rather than choose a skill name and expect every action to fit it.

The DM should also avoid requiring rolls for actions whose outcomes are already certain. Determine whether failure is possible and meaningful before selecting an ability check.

This approach keeps the underlying ability scores relevant and prevents a handful of frequently used skills from replacing the broader ability-check system.

Rules reference: D&D 5e SRD 5.2.1 — Ability Checks and Skills.

Skill checks work best when the table first decides whether a roll is needed at all. Ask what the character is trying to accomplish, whether failure is possible, and whether failure would create a meaningful consequence. If success is automatic or failure would merely force repeated rolls, resolve the action without a check.

When a roll is needed, describe the situation before naming the skill. The player states an approach; the DM chooses the relevant ability and decides whether a proficiency applies. This keeps skills from becoming buttons that players press regardless of method. Different approaches can justify different abilities, difficulties, risks, or information.

Use one roll to resolve one uncertain question. Avoid chains of Perception, Investigation, and Insight checks that all gate the same clue. For essential information, give the core fact and use checks to determine extra detail, speed, safety, or advantage.

Also establish how the table handles assistance, repeated attempts, passive awareness, and group tasks. Consistency matters more than perfect numerical precision. The goal is for checks to resolve uncertain actions while preserving the fictional logic that made the roll necessary in the first place.

Multiple sources of Advantage do not stack in D&D 5e because Advantage is a single mechanical state rather than an accumulating bonus. Whether a character has one source or several, they roll two d20s and use the higher result. Multiple sources of Disadvantage work similarly, and the presence of both cancels the mechanic regardless of their relative number.

This rule places a ceiling on how much a character can improve a d20 roll through favorable circumstances alone. Without it, combining numerous sources could make difficult checks and attacks increasingly predictable, rewarding the accumulation of similar effects over more varied tactical decisions.

Once Advantage is already available, players should generally look for benefits that operate through different mechanics. Additional damage, numerical bonuses, improved positioning, protection, movement, or denying an opponent an action may provide more value than another redundant source of Advantage.

However, redundant sources are not always useless. One source might disappear, become unavailable, or apply only to certain rolls. Maintaining an alternative can provide reliability when battlefield circumstances change.

The practical distinction is between improving a roll that already benefits from Advantage and protecting access to that benefit under changing conditions.

Rules reference: D&D 5e SRD 5.2.1 — Advantage and Disadvantage.

The Help action should represent meaningful cooperation, not an automatic benefit awarded whenever another player says they are helping. In D&D 5e, the Dungeon Master determines whether the proposed assistance is possible and relevant to the task being attempted.

A second character holding a damaged mechanism steady might help someone repair it. Another adventurer searching a different section of a library may contribute useful information. By contrast, an untrained observer repeating encouragement does not necessarily make a delicate operation easier.

Ask the assisting player what their character actually does. The description need not be elaborate, but it should establish a plausible connection between the assistance and the task. Some checks cannot benefit from cooperation, particularly when only one person can perform the action or additional participants would interfere.

For extended tasks, consider whether cooperation changes the method rather than simply supplying an extra die roll. Several characters working together might finish earlier, cover more ground, or combine separate discoveries.

Avoid making assistance artificially difficult merely to suppress Advantage. If the fictional circumstances support teamwork, the mechanics should recognize it. The objective is to keep cooperation meaningful without allowing Help to become a universal, unexplained statistical bonus.

Rules reference: D&D 5e SRD 5.2.1 — Help Action and Working Together.

Use a straight Wisdom ability check when a character faces an uncertain task involving awareness or judgment and no skill proficiency fits. Do not invent a universal 'common sense' skill, and do not require a check merely because the player asks whether their character recognizes something obvious.

First decide whether there is meaningful uncertainty. If a character would notice a fraying rope, tell the player directly. If time pressure or subtle clues make noticing a danger uncertain, decide which ability best represents the task. Wisdom may fit awareness; Intelligence may fit recalling or reasoning through information.

Then ask whether a specific proficiency applies. Perception, Insight, or Survival might be relevant depending on what the character is actually doing. A straight ability check simply means no applicable skill or tool proficiency is being added; the ability modifier still matters. Avoid treating the absence of a named skill as proof that the character cannot attempt the task.

Explain what success and failure mean before the roll. Failure might cost time or leave an ambiguous impression rather than making a sensible character behave irrationally. Consistent checks help players understand the stakes instead of trying to guess which ability the DM prefers.

Rules reference: D&D 5e SRD 5.2.1 — Ability Checks.

Yes. Under the 2024 D&D 5e rules, a natural 1 is an automatic miss for an attack roll, but ability checks use the normal D20 Test procedure: roll, add the relevant modifiers, and compare the total with the Difficulty Class. If the modified total reaches the DC, the ability check succeeds even when the d20 itself shows 1.

A failed check can also reveal an item if finding the item was not the uncertainty the roll was resolving. Decide the stake before the roll. The check might determine whether the character searches quietly, finds the compartment before guards arrive, notices additional information, or avoids damaging something. Failure can then expose the item while creating a different consequence.

That distinction matters especially when an object is necessary for the adventure to continue. Do not place essential progress behind a single binary check if failure would only stop play. Put speed, cost, danger, extra information, or control of the discovery behind the roll instead.

With new players, state the stake in one sentence before asking for the check. That makes it clear that a low die result resolves a particular uncertainty; it does not automatically mean that every possible outcome must be bad.

Rules reference: D&D 5e SRD 5.2.1 — D20 Tests; Rolling 20 or 1.

A maximum Dexterity score does not grant automatic permission to perform any feat that sounds “superhuman.” In D&D 5e, Dexterity contributes a modifier to relevant D20 Tests, but a character’s result can also depend on proficiency, Expertise, Advantage, class features, magic, and—before any roll—the DM’s judgment about whether the attempted action is possible and whether uncertainty requires a check.

That distinction keeps an ability score from functioning as a universal power description. A character with exceptional Dexterity is unusually capable at tasks involving agility, reflexes, balance, and precision, but the rules do not convert a high score into a fixed catalogue of cinematic stunts.

If the campaign wants more overtly superhuman martial action, define where that enters play. High-level class features, magic items, species traits, or explicit rulings about heroic-scale tasks can expand what is possible. The useful approach is consistency: if running across collapsing rooftops or catching a falling ally is possible for exceptional characters, establish the difficulty and consequences in a way players can learn.

Rules reference: D&D 5e SRD 5.2.1 — D20 Tests; Ability Checks; Dexterity.

A DM should roll a player character's check only when hiding the result creates information the character genuinely should not have. If knowing the die result would itself reveal whether the character succeeded, failed, noticed something, or was deceived, a hidden roll can preserve uncertainty. For ordinary actions, let the player roll; the roll is part of their interaction with the game.

Before using a hidden check, ask whether a roll is even necessary. If the character would obviously know the outcome, secrecy adds nothing. If uncertainty matters but the character should not know the result immediately, the DM can roll behind the screen or use an existing passive value when appropriate to the situation.

Do not use hidden rolls merely to prevent players from “metagaming” every uncertain result. That can make the DM responsible for too many character-facing mechanics and reduce player involvement.

The important distinction is between uncertainty in the fiction and uncertainty about the rules. Hide a result when the character lacks information; otherwise keep the procedure visible. Consistency matters more than secrecy, so explain the table's approach before hidden checks become important.

Under the 2024 D&D 5e rules, camouflage does not by itself let a creature use the Hide action while plainly visible. To Hide, the creature must succeed on a DC 15 Dexterity (Stealth) check while Heavily Obscured or behind Three-Quarters or Total Cover, and it must be outside every enemy’s line of sight.

That means visual camouflage is most naturally represented by helping a creature meet those conditions rather than overriding them. Dense patterned clothing in matching terrain might justify an environmental circumstance, Advantage when the DM considers it appropriate, or easier access to concealment, but it does not automatically replace cover, heavy obscurement, or the line-of-sight requirement.

Once the Hide check succeeds, the creature gains the Invisible condition and records the check total as the DC for creatures trying to find it with Wisdom (Perception). The condition ends under the specified triggers, including being found, making an attack roll, making a sound louder than a whisper, or casting a spell with a Verbal component.

So “hiding in plain sight” needs a specific feature or DM ruling; ordinary camouflage supports hiding but does not create a general exception.

Rules reference: D&D 5e SRD 5.2.1 — Hide action and Invisible condition.

A failed ability check should usually change position, pressure, information, or opportunity enough that simply rolling again is no longer the same decision. Failure can remain a genuine “no”; it does not have to become success at a cost. The important part is that the situation moves.

Use three questions when choosing a consequence: what becomes harder, what becomes urgent, or what new option appears? A failed lockpick attempt might leave the door locked but break the tool, make noise, consume time while a patrol approaches, or reveal that another route exists. None of those outcomes grants the original goal, but each changes what the players decide next.

Match the consequence to stakes that were already visible. If time matters, losing time is natural. If secrecy matters, noise is natural. Avoid inventing an unrelated punishment only after seeing a bad roll, because players should be able to understand what kind of risk their action carries.

If failure would leave everything exactly unchanged and repeated attempts are possible without cost, reconsider whether a roll is needed at all. Ability checks are most useful when uncertainty and meaningful failure can actually change play.

Rules reference: D&D 5e SRD 5.2.1 — Ability Checks.

A Dungeon Master should call for a straight Strength check when a task depends on physical force and Athletics proficiency does not meaningfully apply. Strength (Athletics) is appropriate when the task involves athletic training or exertion of the kind covered by that skill. The distinction depends on the fictional action, not simply how difficult it is.

For example, forcing a jammed mechanism, holding a heavy gate against pressure, or pulling apart an object might call for a straight Strength check. Scaling a difficult surface or swimming against a strong current more naturally involves Athletics. These examples are judgments about the relevance of training, not an exhaustive division between technique and muscle.

The DM should describe the obstacle and decide what capability it tests before choosing the check. A player should not automatically add Athletics proficiency to every physical action merely because their character possesses it. Conversely, proficiency should not be excluded arbitrarily when athletic training would genuinely help.

This distinction preserves the importance of Strength itself. A physically powerful character can excel at appropriate tasks without requiring every display of muscle to be treated as a trained athletic performance. The 2024 rules also use specific procedures for some combat maneuvers, which should not be replaced automatically with Athletics checks.

Rules reference: D&D 5e SRD 5.2.1 — Ability Checks and Skills.

Use passive Perception for a character's general awareness when the DM is checking whether they notice something without actively searching; use a Wisdom (Perception) check when the character is actively trying to notice something and the outcome is uncertain. Those are related tools, not interchangeable versions of the same roll.

In the 5e rules, passive Perception is 10 plus the character's modifier for Wisdom (Perception), including applicable modifiers. It represents what the character may notice without the DM asking for a roll. The Search action, by contrast, can call for a Wisdom check using Perception when a character deliberately looks, listens, or otherwise searches.

Passive Perception is therefore not automatically a minimum result for every active Perception check. If a character actively rolls and the total is lower than their passive score, that does not by itself mean the roll should be replaced.

At the table, describe anything obvious without a check. Use passive Perception when hidden information may be noticed incidentally, and call for an active check when a player's declared method creates genuine uncertainty. This keeps Perception from becoming a repetitive “roll whenever you enter a room” procedure.

Advantage and Expertise improve a check in different ways: Advantage improves the die roll, while Expertise improves the modifier. Advantage lets you roll two d20s and use the higher result; Expertise doubles the Proficiency Bonus you add to an eligible ability check. They can also apply together.

The practical difference is consistency. Expertise raises every possible total by a fixed amount, so even an ordinary die result becomes better. Its extra bonus also grows as Proficiency Bonus increases. Advantage does not raise your modifier or your maximum possible die result; instead, it makes low rolls less likely to be the result you keep. Its value therefore depends on the target number: it is especially helpful when either die has a reasonable chance to succeed, but it cannot make an impossible total possible.

A useful mental model is “training versus circumstances.” Expertise represents a character who is unusually reliable at a skill. Advantage represents this particular attempt having better odds because of position, help, information, or another favorable effect.

Rules reference: D&D 5e SRD 5.2.1 — Advantage and Disadvantage; Expertise.

A natural 20 on a Perception check is not automatically a success in D&D 5e. Automatic success on a natural 20 is a rule for attack rolls; an ability check succeeds when the total result meets or exceeds the DC.

That also means a 20 does not let a character perceive something that cannot be perceived. If a secret is behind a solid wall with no sensory clue, or the character is looking in the wrong place, a high Perception result does not create information from nowhere. The DM should call for a check only when success and failure are both genuinely possible.

When the total does succeed, give the player the strongest information the character could reasonably obtain from that method. A very high result can justify clearer detail, noticing secondary clues, or reducing ambiguity, but that is adjudication rather than a special critical-success rule.

If the player rolled without being asked, first decide whether a check was appropriate at all. You do not need to invent a hidden door or bonus discovery simply because the die showed 20. Reward good observation with useful, actionable description, not with information that breaks the fiction.

No, not by the Charmed condition alone. Being Charmed restricts how you can act against the creature that charmed you and gives that creature an advantage in social interaction; it does not say that you automatically believe the charmer, lose awareness of contradictions, or stop receiving information from other effects.

For a lie-detecting magic item, read the item’s own text next. If it passively vibrates whenever its wearer is told a lie and it has no exception for Charmed creatures, the condition does not inherently switch the item off. The wearer could therefore notice the vibration while still being Charmed.

What that information means is a separate question. A Charmed character can know that a statement appears false and still be constrained by the condition’s actual effects. The DM can describe the resulting tension: the item warns the character, but the magical influence still affects what actions are available toward the charmer.

If the earrings are homebrew, their wording controls the unresolved details. Decide whether they detect deliberate lies, factual falsehoods, half-truths, or mistaken statements. Those definitions matter more to the item than the Charmed condition itself and should be established consistently before the next ambiguous case.

Yes. In the 2024 rules represented by SRD 5.2.1, Influence is still a defined action. It covers attempts to change a creature’s attitude through a Charisma check using Deception, Performance, Intimidation, or Persuasion, or a Wisdom check using Animal Handling where appropriate.

That does not mean every conversation should become a formal Influence roll. The social-interaction rules still depend on what the characters say and what the NPC wants, knows, fears, or refuses to do. The action provides a mechanical procedure when the outcome is uncertain and a check is warranted; it does not erase roleplay or force an NPC to accept impossible demands.

For a DM, the useful sequence is: establish the creature’s current attitude and interests, hear the player’s approach, decide whether the attempt can work, then call for the relevant check if uncertainty remains. A strong argument may change what is possible or which skill applies before the die is rolled.

Rules reference: D&D 5e SRD 5.2.1 — Actions: Influence; Social Interactions.

A character with low Charisma should still be able to attempt persuasion, deception, and other social approaches. A poor ability score changes the character's likelihood of succeeding at uncertain tasks; it does not remove the player's opportunity to propose clever plans or participate in conversations. The DM should evaluate the approach before deciding whether a roll is necessary.

Suppose a low-Charisma character tries to persuade a guard to permit entry. Offering a genuine benefit, presenting relevant evidence, or identifying the guard's personal interests may improve the situation even if the character struggles to communicate convincingly.

When an ability check is appropriate, apply the character's actual modifier and any relevant proficiency. A creative proposal should not automatically replace the roll, but the fictional circumstances may determine what outcomes are possible or how the attempt is adjudicated.

Failure can produce interesting developments without making the character incompetent at everything. The guard might misunderstand the request, offer a limited concession, demand proof, or become suspicious of the character's motives. The precise result should follow the NPC's personality and circumstances.

Low Charisma can therefore shape social encounters without restricting characters to silence. The goal is to preserve meaningful decisions while allowing mechanical weaknesses to influence how those decisions unfold.

Rules reference: D&D 5e SRD 5.2.1 — Ability Checks and Skills.

On a qualifying ability check, Halfling Fortune can reroll a natural 1, and Reliable Talents can still replace the resulting d20 value with 10 if that new result is 9 or lower. The reroll does not stop being part of the same ability check simply because Fortune changed which d20 result is used.

In SRD 5.2.1, Fortune says that when a halfling rolls a 1 on a D20 Test, they may reroll and must use the new result. Reliable Talents says that when a rogue makes an ability check using one of their skill or tool proficiencies, they may replace a d20 result of 9 or lower with 10.

That means the reroll remains potentially useful even on a check where Reliable Talents provides a floor. A reroll of 15 is better than 10; a reroll of 7 can become 10 if the check qualifies. Reliable Talents does not apply to every D20 Test, so attacks and saving throws do not gain that floor merely because Fortune can affect them.

Rules reference: D&D 5e SRD 5.2.1 — Halfling: Fortune; Rogue: Reliable Talents.

If the 2024 rule being used calls for a Dexterity (Sleight of Hand) check made with Thieves’ Tools, proficiency in the skill and proficiency with the tool are two different relevant proficiencies. Under the 2024 tool-check framework, having both can grant Advantage rather than adding the Proficiency Bonus twice.

The important point is how the check is constructed. “Sleight of Hand” identifies the skill proficiency that can apply to the ability check, while “using Thieves’ Tools” identifies the tool involved. A character proficient in only one of them can still receive the normal benefit that the rule assigns to that proficiency. A character with both does not double the Proficiency Bonus; the additional benefit is Advantage when the rule says both proficiencies apply.

Do not rewrite the check as “Dexterity (Thieves’ Tools)” merely because tools are being used. Use the wording of the specific lock rule your table is following, because 2014 and 2024 material describe tool checks differently.

This answer should be checked against the exact 2024 Dungeon Master’s Guide passage before publication, since that locked-door wording is outside the verified SRD material used for this batch.

The Help action should grant Advantage under the 2024 rules; adding the helper’s skill modifier would be a house rule, not RAW. For an ability check, the helper chooses one of their own skill or tool proficiencies and an ally close enough to assist. That ally gets Advantage on the next check using that skill or tool before the helper’s next turn, and the GM decides whether the assistance is actually possible. Help with an attack instead distracts an enemy within 5 feet and gives Advantage on the next allied attack against it.

Replacing Advantage with the helper’s full modifier creates a different scaling problem. A highly specialized character could become a portable bonus for someone else, and that bonus could stack with Advantage from another source. The result rewards modifier stacking more than cooperation.

If the goal is to make the quality of the helper matter, keep the existing math and make the fiction carry that distinction. Require a plausible method of assistance, enforce the proficiency requirement for ability-check Help, and refuse Help when a second person cannot meaningfully contribute. That keeps cooperation concrete without turning every task into bonus arithmetic.

Use NPC attitudes and faction loyalties to decide which requests are plausible, what evidence matters, and how difficult a social interaction should be. A successful D&D 5e social check should resolve uncertainty within the fiction; it should not automatically override a character's loyalties, knowledge, or interests.

Before asking for a roll, establish the NPC's immediate goal, existing opinion of the party, and obligations to other people. A guard might dislike an arrogant visitor but still accept valid authorization. A friendly guild representative might refuse to betray members even after a persuasive appeal. The circumstances determine whether a check is needed and what success can reasonably accomplish.

Let player choices change the situation. Offering reliable proof, helping an ally, exposing a faction dispute, or making a credible bargain may change which outcomes become possible. A threat could work in the moment while creating future hostility; a favorable conversation does not necessarily erase political consequences.

Record the NPC's response and any promises or debts afterward. For recurring factions, use a few clear relationship states instead of treating each conversation as an isolated die roll. If applying a specific published social-interaction procedure, check the chosen rules edition rather than assuming every D&D 5e book uses the same framework.

First decide whether the attempted theft is physically possible; if it is, Dexterity with Sleight of Hand proficiency is usually the most appropriate check for covert manipulation. D&D 5e does not require a roll simply because a player describes an unusual action: the DM first determines how the action can be resolved and calls for a check when the outcome is genuinely uncertain.

Sleight of Hand fits when the challenge is secretly manipulating or removing an accessible object without the other person noticing. Acrobatics would normally be relevant only if agility, balance, or unusual body movement were actually the obstacle. Performance or another Charisma-based approach might matter if the character first needs to convincingly create a distraction, but that is a different problem from physically taking the item.

Avoid automatically turning one strange action into a chain of three or four checks. Decide what the important uncertainty is. If the distraction is already established, one Dexterity check may be enough. If both gaining the person's cooperation and stealing the object are uncertain and independently consequential, resolving them separately can make sense.

The original golden-teeth example also demonstrates an important limit: a die roll does not make an impossible action possible. If the teeth are firmly attached, they cannot simply be pickpocketed during a kiss. The player would need a plausible method of removing them first, with whatever consequences that method creates.

Advantage and Disadvantage lose nuance when several meaningful circumstances collapse into the same mechanical result, but a DM can preserve those circumstances without making the dice mechanic more complicated. Rules-wise, multiple sources of Advantage do not stack, multiple sources of Disadvantage do not stack, and a roll affected by both has neither, regardless of how many sources exist on either side.

That simplicity is useful for speed, but it means a carefully prepared ambush and a small moment of assistance can both produce the same two-d20 roll. Rather than adding extra d20s, let additional favorable circumstances change what the roll can accomplish or what failure costs.

For example, Advantage might already represent attacking from concealment. A second favorable factor could mean the character also has an escape route, avoids exposing an ally, identifies the enemy leader before attacking, or retains good cover if the attack misses. Several disadvantages can work the same way: one may impose Disadvantage while another worsens the consequence of failure, restricts movement, or creates a new problem in the scene.

This keeps the core resolution quick while allowing the fiction to remember that three good ideas are better than one. If a table instead wants stacking numerical modifiers, make that an explicit and consistent house rule rather than improvising different bonuses whenever the standard mechanic feels insufficient.