Why the labels look like this
The model
A directory can’t control how tools are used. What it can do is give every tool honest labels — where it comes from, what it opens, and what it takes to hold it — so judgment has something to work with.
The safe-use envelope
Four numbers, deliberately separate. Depth and quality are different things — a well-run depth-1 systems session can move a team further than a clumsy depth-4 ritual. The most common facilitation failure is a depth-3 exercise run with holding-1 competence; the labels make that mismatch visible before the workshop.
Depth — what it can open
| ●○○○ | Informational | Conceptual and cognitive engagement; participants talk ABOUT topics. Little or no self-disclosure beyond opinions. Examples: discussing a framework, mapping a system, sorting cards without personal stakes. |
| ●●○○ | Reflective | Personal reflection and meaning-making; participants share their own experience at a self-chosen comfort level. Examples: journaling prompts, structured dialogue rounds, values clarification. |
| ●●●○ | Emotional | Invites live feeling and self-disclosure with interpersonal risk: empathy practices, sustained eye contact, naming grief/anxiety/joy in the room, giving and receiving personal impressions. The exercise works because emotion becomes present. |
| ●●●● | Transformational edge | Deliberately works at participants' growing edge: identity, despair, shadow, deep somatic or transpersonal process. Can destabilize before it reintegrates; aftercare and voluntary, informed participation are part of the design, not optional extras. |
Holding — what it takes to hold
| 0 | Self-guided | Designed for solo use without a facilitator (guided audio, journaling prompt, self-assessment). |
| 1 | Host | Someone keeps time and reads instructions; no group-process skill required (energizers, simple structured rounds). |
| 2 | Facilitator | Group-facilitation craft required — framing, psychological safety, balancing participation, debriefing. |
| 3 | Space-holder | Experienced facilitator able to hold emotional process: welcoming and normalizing strong affect, working with triggering, closing well, arranging aftercare. |
| 4 | Specialist | Professional/clinical or lineage-specific training assumed (therapeutic methods, trauma work, deep somatic release, certified programs). |
Trust — what the group must already have
| ★ | any | Works in freshly formed or low-trust groups |
| ★★ | some | Requires some psychological safety — people speak without fear of ridicule |
| ★★★ | established | Requires established trust — people can risk disclosure and disagreement with each other |
Disclosure — what the design asks people to share
| none | No personal material in the design — task-focused |
| optional | Personal sharing is invited and meaningful |
| required | Personal sharing IS the mechanism — the tool does not run without it |
The pairing rule underneath: whoever holds a process should be able to work one level deeper than the exercise aims, because exercises open more than they intend. Where a design shares nothing — journaling, solo quests, silent practice — privacy is the container, and the labels say so.
What the labels cannot catch
Depth, holding, trust and disclosure describe what an exercise opens and what running it demands. They say nothing about what the work is for, and that is where this field goes wrong more often than it goes wrong on safety. The UNDP's Conscious Food Systems Alliance put it plainly in its evidence review, and it is worth carrying here rather than paraphrasing:
These approaches are “not about saying we need to change people's beliefs, values, and worldviews, as this would turn them into objects to be changed rather than seeing them as change agents”. Doing so “risks co-opting the concept of transformation to preserve business-as-usual”. And spaces for inner work that do not “include explorations of power and bias… may not lead to the deeper shifts needed, or worse may reinforce a dominant groups' values.”
A depth-2 exercise with holding 1 and trust any can still be an instrument for getting people to accept what they should be questioning. No label in this model detects that. The same review notes that evidence for whether these practices shift wider outcomes at all is “in their infancy” — which is why nothing here is described as proven.
Honesty fields
Mechanisms say how a tool works (attention-training, open-awareness, somatic-awareness, emotional-processing, cognitive-reframing, perspective-taking, …), which enables substitution and sequencing. Change scope keeps claims honest: most single exercises produce a state; practices build skills; only sustained, scaffolded engagement supports capacity — per the 2021 IDG report, qualities like Connectedness are not produced by training, only supported. Form names the genre: exercise, practice, method, or instrument.
Origin travels with every tool — creator, tradition, lineage notes, license. Exercises get renamed and re-attributed as they travel; this directory keeps receipts, and links instructions home instead of republishing them.
The full model with its rationale lives alongside the dataset — YAML records, controlled vocabularies and a validator — in the project repository.