Tuckman team stages · IN SHORT
Use this skill to examine how a team is working together after a change or during a persistent disagreement. It offers a working interpretation of the team stage, an alternative explanation and actions suited to your role.
Your brief
In a fictional project team, people disagree about who has authority and meetings end without a decision. Storming may be a useful interpretation; forming remains an alternative if roles were never established.
Possible draft
One practical intervention is to agree who decides each type of question at the next meeting. Review whether that agreement changes the behaviour before drawing a stronger conclusion about the team.
01 / THE FOUNDATION
The output links those observations to a possible stage, explains the second most likely interpretation and suggests actions for both the leader and team members, with moments to carry them out.
02 / SEE HOW IT WORKS
In a fictional project team, people disagree about who has authority and meetings end without a decision. Storming may be a useful interpretation; forming remains an alternative if roles were never established.
One practical intervention is to agree who decides each type of question at the next meeting. Review whether that agreement changes the behaviour before drawing a stronger conclusion about the team.
Teams do not necessarily pass through every stage in a fixed sequence. A stage is not a judgement of individual ability, and the skill cannot promise a date for the next transition.
03 / GET STARTED
Describe what people do without labelling an individual as the problem.
Consider forming, storming, norming, performing or adjourning, including a plausible alternative.
Give each intervention a concrete moment and note signals that would require reassessment.
04 / MAKE IT YOURS
The English ZIP contains the SKILL.md instructions and is free to download without an account. The file declares the MIT licence; retain its source and licence notices when adapting or sharing it. Need help applying this skill? Book a call.
Full informationDownload. Apply. Refine.
File language: English
Explore the details at your own pace.
You describe what is happening in your team, and Claude determines which of the five Tuckman phases that team is in: forming, storming, norming, performing or adjourning. With the reasoning attached, four interventions for the leader, two for the team itself, the leadership style that fits this phase and the relapse risk. The conflict does not disappear because of it. It gets a place, and with that a next action.
the lines above come from the zip on this page · SKILL.md is 11,991 bytes
This is the example from the SKILL.md, written out here to show what the skill does with it. Note the third line of the answer: the skill always names the second most likely phase and why it did not choose that one. A diagnosis that only confirms itself is not a diagnosis.
The AI Tuckman Team Phases Coach is a free skill from our own skill library for Claude. A skill is an instruction file, SKILL.md, that gives an AI assistant a fixed method for a task. No software, no subscription, no link to your HR system: a text file of 1,863 words that tells Claude which signals belong to which team phase, which questions it should ask first, how it resolves doubt between two phases, and in which seven blocks the answer arrives.
The problem the skill solves is that team trouble rarely has a name. People grumble in the corridor, meetings produce no decision, two people stop speaking up, and nobody can say exactly what is wrong. What happens next is predictable: everyone looks for someone to blame, because pointing at a person is easier than recognising a pattern. The Tuckman model gives that pattern a name, and that changes the conversation. Clashing over the approach is not proof that the team is no good, but a pattern the model invites you to examine, not a stage every team is known to undergo.
It is written for managers, project leads, scrum masters and for team members who are not leaders themselves but want to understand what is happening. That last group is named explicitly in the file, because question four of the intake asks what your role is: manager, team member or outsider. The answer determines which interventions you get back, because a team member simply has different levers than a manager.
What the skill is mainly meant for: the moment a team effectively starts over. A new team member, a departing team member, a new manager, a merger, a reorganisation, a new project with the same people, or the move from office to hybrid working. The file lists that as a separate trigger list, and that is not a coincidence: that is exactly the moment people think something is wrong, while the model says something normal is happening.
New to this kind of file? Then start by reading what Claude skills are NL and how to install them NL. Want to write one yourself for your own way of working? The structure is explained in writing a SKILL.md NL. This skill belongs to the skill library that we make available for free from the AI and automation service.
The signals below summarise those in the SKILL.md, because that is what the skill diagnoses on. Not on what you think is going on, but on the behaviour you describe. The line above shows what the model revolves around: performance may dip during conflict and improve as collaboration develops; this drawing is illustrative, not a measured or necessary trajectory. The blue card is the phase from the FIG.01 example. The yellow line below runs the other way: that is possible relapse, not evidence that relapse occurs more often than progress.
progression according to Tuckman's model, not a measurement of an actual team
The SKILL.md uses the model to put conflict in context. Conflict may be part of negotiating roles and agreements, but a lack of visible clashes does not prove that a team cannot reach norming. The stage is a working interpretation based on described behaviour, not a universal law. The skill names that interpretation and its evidence before suggesting interventions.
A phase is not a judgement of people. The file says so explicitly: storming says something about the age of the collaboration, not about the standard of the team. That looks like a detail, but it sets the whole tone of the output. A team that is told it is in storming hears that this is normal. A team that is told there is too little collaboration hears a reproach.
The order is a pattern, not a law. Teams skip phases when the members already know each other. A group that has already done three projects together does not land back in forming on a fourth. And the reverse is true too: any major change in composition, assignment or leadership can set a team back. That is why the intake asks what has changed in the last three months, and why the relapse gets its own block in the output.
The right style belongs to the phase, not to the person. This is the skill's practical application of the model, not a universal intervention prescription from Tuckman's research. Directive in forming, coaching in storming, facilitating in norming, delegating in performing. Many managers have one fixed style and apply it everywhere. A leader who already delegates in forming leaves a group to flounder, and a leader who still directs in performing removes exactly the autonomy the team had just built up.
Interventions are actions, not attitudes. For every intervention, it forces itself to name a concrete moment. Not be clearer, but set out the decision rules at the next meeting and write down who decides on what. That difference is the difference between advice you nod along to and advice you carry out on Monday.
In diff form, with the rule from the instruction file: how attitude advice turns into an action.
Every diagnosis the skill produces has the same seven blocks in the same order, and that order is not formatting but reasoning: first what, then why, then evidence, only then advice. The mono line under each block explains why it sits there.
That fixed order has another practical advantage: two diagnoses can be compared with each other. Run the skill in January and again in April, and blocks one and three immediately show whether anything has shifted, and block seven shows whether the risk you named back then actually happened.
The file opens with frontmatter: the name ai-tuckman-team-stages-coach, version 1.1.0, an MIT licence, and a description that determines when Claude picks the skill up. Alongside technical terms like tuckman, team phases, group dynamics and forming storming norming performing, it also lists everyday phrases as trigger words: new team member, new manager, team is clashing, team is not working, we cannot reach agreements. That detail matters more than it looks, because almost nobody types the word Tuckman when they have a problem.
Next comes a chapter on when the skill should switch itself on, and, notably, also when not to. The file states: do not trigger when the question is about an individual rather than the group, because one employee's performance problem is not a team phase. And then the instruction that makes the difference: say so, and do not treat it as if it were one. A model that fits everywhere fits nowhere well.
The theory chapter summarises Tuckman briefly and adds two warnings that must stay explicit: the order is a pattern and not a law, and a phase is not a judgement of the quality of the people. It then lists, per phase, the signals, what the team needs in that phase, and the leadership style that fits. All of that is summarised in FIG.02 above.
The method has three phases. Phase one is the five-question intake, asked all at once and not one by one: describe the team, what is concretely happening in behaviour rather than conclusions, what has changed in the last three months, what your role is, and what you have already tried. If question two or question three is missing, the skill asks again. It may fill in the rest itself, provided it says that it is doing so.
Phase two is the diagnosis, with two hard rules. The first: always name the second most likely phase and why you did not choose it. The second: when in doubt between two phases, choose the earlier of the two, because a team placed too far ahead gets interventions it cannot handle. There is also an explicit relapse check built in: if someone has joined or left in the last three months, or there is a new manager, check whether a team previously in norming or performing now shows signals of forming or storming. A change alone does not establish relapse.
Phase three is the interventions: four for the leader, two for the team, each an action with a moment. After that comes the output format with the seven blocks, the chapter on what the skill does not do, a tone chapter that bans team-coach language and metaphors about journeys and mountains, the two sources, and the worked example you see in FIG.01.
NOTE · fix 2026-08-10: the published zip file contained an incomplete SKILL.md without the method, output format and example, while this page already described those parts. The SKILL.md has been completed based on Tuckman's model and the structure this page already named, and republished under the same path. Everything after this comes straight from that file.
The SKILL.md contains a chapter with six things the skill does not do, and that chapter matters more for a team diagnosis than for most other skills. After all, you are talking about colleagues who are not in the room. Below are the six rules as request and response.
Two sources, both named in the file, and nothing more. That is deliberate: the model is old and widely cited, and precisely because of that, countless extensions and figures circulate that Tuckman himself never wrote down.
Tuckman, B.W. (1965). Developmental sequence in small groups. Psychological Bulletin. Tuckman worked as a psychologist at the Naval Medical Research Institute and reviewed 50 articles about small-group development. From that he distilled the four phases forming, storming, norming and performing. The article can be found via PsycNET and is listed with its source in the SKILL.md.
Tuckman, B.W. and Jensen, M.A.C. (1977). Stages of small group development revisited. Twelve years later, Tuckman and Mary Ann Jensen added the fifth phase, adjourning, about the group breaking up. In practice this phase is the one most often skipped, even though it is exactly where the knowledge the team built up gets lost.
What you will not find on this page are percentages of teams that get stuck, average durations per phase, or scores per team type. Those are not in the sources, so they are not in the skill either. The model is a language for ordering behaviour, not a measuring instrument.
Want to approach team dynamics from a different angle? There are two skills that sit close to this one. The SCARF Model Coach NL looks at why people feel threatened by something that, on paper, takes nothing away, which is often the underlying cause in the storming phase. And the Radical Candor Coach NL is about the conversation itself, once the diagnosis tells you it needs to happen.
The zip contains a folder, ai-tuckman-team-stages-coach, with the SKILL.md inside it. Installing is a matter of putting that file in the right place, and that place differs per environment. SKILL.md has been an open standard since December 2025, so the same skill also works in Codex, Cursor and Gemini CLI. So you are not downloading a Claude file, but a working instruction that any modern AI assistant can read.
~/.claude/skills/ or into .claude/skills/ inside your project./ai-tuckman-team-stages-coach inside the session.AGENTS.md in your repo.AGENTS.md.After that, you simply describe what is going on. The more concrete the behaviour, the sharper the diagnosis: write down what you saw happen in the last meeting, not what you thought of it. A fuller explanation of installing in different environments is in installing Claude skills NL, and the broader background on working with AI is in the knowledge base.
One more practical note about privacy, because this question comes up with this skill more than with most others: you are describing colleagues. Use first names or roles instead of full names, and keep personal circumstances out of it. The diagnosis is none the worse for it, because the skill diagnoses on behaviour in the group, not on individuals.
It is at its strongest right after a change in the team. A new team member, a departing colleague, a new manager, a merger or a reorganisation: then there is a clear moment when the behaviour changed, and that is exactly what the diagnosis needs. It is also strong with persistent conflict. Storming that does not resolve itself has a cause, and the distinction between a role question and a process question often makes that cause visible in a conversation.
A third good moment is the start of a new project. Then you should expect forming, not top performance, and that helps you calibrate your expectations and your planning. A team that is still asking for a framework in week two is not slow, it is forming.
There are also situations where you are better off leaving it be. The first is in the file itself: when the question is about an individual rather than the group. Someone who is not performing, someone who is overloaded, someone you need to have a difficult conversation with: that is not a team phase, and a five-stage team model will not help there. The Nonviolent Communication Coach and the GROW Coach fit better there.
The second limit is more serious. At signs of bullying, exhaustion or unsafety, the skill stops diagnosing and refers to a person. That is one of the six rules from FIG.04, and it is not a formality: a model that treats such signals as a phase trait does harm. If you are not sure whether that is what is happening, talk to a confidential adviser or an occupational physician, not an AI assistant.
And then the most honest limit: the skill does not know your team. It works with what you describe, and your description is, by definition, your perspective. Two team members running the same skill on the same team can get back two different phases, and that is not an error but information. If the cause of the trouble does not lie in the group but in a change coming from above, the Kotter Change Coach is the better tool.
This skill is the free do-it-yourself version of work we also deliver as a service. The skill stays complete and with no catches, but be clear about what a skill is: it teaches your AI how to do something, while every new session starts empty. The skill is not the engine and not the memory. With a team diagnosis you feel that right away: explaining all over again who is on the team, what happened in March and which intervention you already tried, takes more time than the diagnosis itself.
Rung 3 deserves more than one paragraph, because this is the difference between a clever chat and a system you can build on. A team diagnosis is, above all, something that gains value over time: the previous diagnosis, the intervention you chose and what happened afterwards. Jarvis is the organisation brain that remembers what your AI tools need to know, divides up the work and keeps track of what happened.
We have been running on this system ourselves for months. Every agent session, every task and every decision is logged in it and can be read back. A new session therefore does not start blank: it first retrieves the recorded decisions, the running projects and the latest changes, and carries on where the previous one stopped. So we are not describing a promise but the way we actually work every day.
See the four plans at jarvis/prijzen NL. Through the waiting list NL you only pass on your preferred plan, without obligation. That does not create an account, an order or a payment obligation. We discuss business custom work first.
Tuckman tells you where the team stands. What you do next depends on the cause: a change from above, behaviour that is not sticking, or a conversation that is not happening. All 100 skills in the library are free, and these seven sit closest to this one.
If the team is in storming because of something that came from outside, you address that process, not the team.
Kotter Change CoachEight steps to make a change land, from urgency to embedding it in the culture.SKILL Switch Framework Change CoachChange from head, heart and environment, useful when the resistance is emotional rather than rational.SKILLAgreements that should stick in norming but do not are a behaviour question.
Fogg Behaviour Model CoachWhy behaviour does or does not happen, broken down into motivation, ability and prompt.SKILL EAST Framework Behaviour CoachMakes the desired behaviour easy, attractive, social and timely.SKILLThe diagnosis points to the conversation. These three are about how you have that conversation.
Radical Candor CoachBeing direct without becoming unkind, exactly what storming asks of a leader.SKILL NL Nonviolent Communication CoachFor the conversation that has to be about behaviour, not about the person.SKILL SCARF Model CoachWhy a change feels threatening when, on paper, nothing is being taken away.SKILL NLWhere this skill comes from, and what else there is.
The whole skill library100 free skills in one place, sorted by topic.HUB AI and automationThe service behind it: from separate skills to working automation in your business.SRV GROW CoachFor the individual conversation, exactly where a team model no longer fits.SKILL AI trainingIf your team wants to learn to set this kind of work up itself.SRVNothing. The frontmatter says license: MIT, so you are free to use, share and adapt it. You download a 5.1 KB zip containing a folder and a single file, SKILL.md, and that is the complete skill. No account, no email address, no paid version and no sales email afterwards.
Yes. SKILL.md has been an open standard since December 2025, so the same file also works in Codex, Cursor, Gemini CLI and other tools that follow the standard. In Codex you unzip it into .agents/skills/ in your project, or into ~/.agents/skills/ for all your projects; Codex has supported SKILL.md directly since the open standard of December 2025. Putting the contents of SKILL.md into your AGENTS.md still works too. The instructions are plain readable text, so any assistant that accepts instruction files can handle it.
Seven fixed blocks: the phase the team is in, the reasoning behind that diagnosis, the traits you recognise in this team, four interventions for the leader, two interventions for the team itself, the leadership style that fits this phase, and the relapse risk with its early signal. That order is fixed in the file, so two diagnoses can be compared with each other.
No, and the skill says so explicitly. The order is a pattern and not a law. Teams skip phases when the members already know each other, and they may revisit an earlier stage after a major change in composition, assignment or leadership. A new team member or manager is a reason to check the behaviour, not proof of a return from performing to forming or storming.
No. The skill treats storming as a possible pattern during negotiation of roles and agreements, not a necessary stage for every team. A phase is an interpretation of the collaboration, not a measured age or a judgement of the people. The skill must never present a phase as a judgement of individuals.
On two publications named in the file. Bruce Tuckman reviewed 50 articles about small-group development and published Developmental sequence in small groups in Psychological Bulletin in 1965, with the four phases forming, storming, norming and performing. In 1977 he added the fifth phase, adjourning, together with Mary Ann Jensen, in Stages of small group development revisited.
When the question is about an individual rather than the group, because one employee's performance problem is not a team phase. And when there are signs of bullying, exhaustion or unsafety: then the skill names that and refers to a person, such as a manager, a confidential adviser, HR or the occupational physician. It does not make a psychological diagnosis.
Download the zip, put it in your AI environment and describe what you saw happen in the last two meetings. Within ten minutes you have a phase, a rationale and six interventions, at least one of which is doable this week. If your team already works with AI but without agreements on how, the AI training is a more logical first step than a new subscription. And if you want this kind of work prepared and ready, with a person approving it, we are happy to simply talk about that.
