forge-openclaw-plugin 0.2.25 → 0.2.26

package/skills/forge-openclaw/entity_conversation_playbooks.md

@@ -7,6 +7,8 @@ Forge correctly, and gather only the structure that still matters.
 
 ## Core stance
 
+- Lead with what the user is trying to preserve, change, resolve, or make true, not
+  with the entity label.
 - Ask only for what is missing or still unclear.
 - Before every question, decide the one missing thing you are trying to clarify.
 - Know where the conversation is headed before you ask the next question.
@@ -20,6 +22,10 @@ Forge correctly, and gather only the structure that still matters.
   brief reflection -> one orienting question.
 - Another good shorthand for the same rhythm is:
   short reflection -> one orienting question.
+- When the user is vague, ask for the smallest real example, desired outcome, or stake
+  before you ask for wording.
+- When the user is clear, say what the record seems to be becoming and move straight to
+  the last missing structural detail.
 - For straightforward logistical entities such as tasks, calendar events, work blocks,
   timeboxes, and task runs, use a fast path:
   one brief confirming sentence -> one operational question.
@@ -33,6 +39,8 @@ Forge correctly, and gather only the structure that still matters.
   yourself and ask whether it fits.
 - If the user already answered the usual opening question, do not repeat the stock
   opener. Move straight to the next missing clarification.
+- After a substantive answer, briefly say what is becoming clear so the user can
+  correct the direction early.
 - Once the record is clear enough to name, stop exploring broadly and ask only for the
   last missing structural detail.
 
@@ -71,6 +79,40 @@ Update record:
 
 - "Before I change it, what feels newly true now, and what should stay intact?"
 
+## Steering moves
+
+Use these small moves to keep the intake natural and intentional.
+
+When the user is still searching:
+
+- reflect the stake in one sentence, then ask for the smallest concrete example or the
+  desired outcome
+
+When the user already knows what they mean:
+
+- offer the working formulation, then ask only for the last missing placement, timing,
+  or ownership detail
+
+When the record carries emotion but is not Psyche:
+
+- reflect what the user is trying to hold onto, repair, or not lose, then ask one
+  orienting question
+
+When the user is updating an existing record:
+
+- ask what is changing, what should remain true, and what prompted the change now only
+  if those answers would alter the record shape
+
+When you are about to save:
+
+- give one short working summary in the user's language and ask whether it feels true
+  enough or needs one correction
+
+When an adjacent record becomes visible:
+
+- name it gently and ask whether it should be linked now, saved separately later, or
+  left alone for now
+
 ## Question design rules
 
 - Let each question have one job:
@@ -83,15 +125,29 @@ Update record:
 - Do not over-warm or over-therapize logistical records. For those flows, one brief
   confirming sentence plus one question is usually enough.
 - Ask the more meaning-bearing question before the more administrative one.
+- Prefer "what", "when", and "how" before "why" when the user's meaning is still
+  forming. "Why" is often better after the experience or outcome is already clear.
 - If the user is uncertain, ask for a recent example before asking for an abstraction.
 - If the user is clear and decisive, confirm the working formulation and move directly
   to the one missing structural detail.
 - Avoid dead-form prompts such as "What should this be called?" when the user is still
   figuring out what the thing is.
+- For labels such as `tag`, `event_type`, and `emotion_definition`, do not open with a
+  naming question unless the meaning is already clear and only the wording is missing.
 - Before the final save question, it is often better to offer a candidate formulation
   than to ask for raw wording from scratch.
 - When useful, say what you think the record is becoming before asking the next
   question. That helps the user correct the shape early.
+- For reusable vocabulary or taxonomy records such as `tag`, `event_type`,
+  `emotion_definition`, `preference_catalog`, and `preference_context`, ask what
+  distinction the label should help the user notice, sort, or retrieve later before
+  you ask about naming or aliases.
+- For emotionally meaningful vocabulary records such as `event_type`,
+  `emotion_definition`, and many `self_observation` entries, start from the lived
+  moment or felt meaning before you move to reuse or retrieval language.
+- For collection-like records such as `preference_catalog` and
+  `questionnaire_instrument`, ask what they are for before you ask what should go
+  inside them.
 
 ## Ready-to-save check
 
@@ -104,6 +160,11 @@ Before saving, make sure you can answer all of these in plain language:
 
 If the answer to the last question is "none", save it instead of prolonging intake.
 
+Before the final write, it is usually worth asking one light confirmation such as:
+
+- "That sounds like the right shape to save. Do you want to keep it that way, or is
+  there one thing you want adjusted first?"
+
 ## Update loop
 
 Use this when the user is updating an existing record rather than creating a new one.
@@ -113,6 +174,9 @@ Use this when the user is updating an existing record rather than creating a new
 3. Ask what prompted the update now if that changes the shape of the record.
 4. Then ask only for the missing structural detail required by the change.
 
+If the current title or shape may no longer fit, offer one revised formulation yourself
+before asking the user to rewrite it from scratch.
+
 ## Goal
 
 Aim: clarify the direction and why it matters, not just produce a title.
@@ -253,6 +317,34 @@ Preferred opening question:
 
 - "What recurring move are you trying to strengthen or loosen?"
 
+## Tag
+
+Aim: create a label that helps future retrieval or grouping, not just another vague
+bucket.
+
+Arc:
+
+1. Ask what the tag should help the user notice, group, or find later.
+2. Ask what kinds of records should belong under it.
+3. Offer a concise label if the meaning is clearer than the wording.
+4. Ask about color, kind, or parent grouping only if that changes how it will be used.
+
+Helpful follow-up lanes:
+
+- what the tag is for later
+- what should count as inside versus outside the label
+- whether the user already has nearby tags that this should stay distinct from
+
+Ready to save when:
+
+- the tag has a stable label
+- the grouping meaning is clear enough to reuse later
+- any important distinction from nearby tags is clear
+
+Preferred opening question:
+
+- "What do you want this tag to help you notice or find again later?"
+
 ## Note
 
 Aim: preserve the useful context and link it to the right places without turning the
@@ -279,7 +371,7 @@ Ready to save when:
 
 Preferred opening question:
 
-- "What feels important to preserve from this?"
+- "What feels important enough here that you would want to be able to find it again later?"
 
 ## Insight
 
@@ -426,6 +518,9 @@ Arc:
 3. Ask what it may connect to: pattern, belief, value, mode, task, project, or note.
 4. Ask for tags or extra context only if that will help later review.
 
+If the user already gave the moment or timing, move straight to what they noticed most
+clearly instead of re-asking when.
+
 Ready to save when:
 
 - the observation itself is clear
@@ -455,7 +550,7 @@ Ready to update when:
 
 Preferred opening question:
 
-- "What about this night feels worth remembering or linking?"
+- "What about this night feels important enough to remember or connect?"
 
 ## Workout Session
 
@@ -475,33 +570,121 @@ Ready to update when:
 
 Preferred opening question:
 
-- "What about this workout feels most worth capturing?"
+- "What about this workout feels most worth remembering or connecting?"
 
-## Preference Item Or Context
+## Preference Catalog
 
-Aim: understand whether the user is trying to save a preference signal, define a
-context, or inspect the learned model.
+Aim: define a useful comparison pool, not just a list with no decision purpose.
 
 Arc:
 
-1. Ask what preference question the user is trying to answer.
-2. Distinguish whether this is:
-   a direct item signal,
-   a context change,
-   or a request to inspect what Forge already knows.
-3. If the user wants live comparisons, hand off to the Preferences game UI.
-4. If the user wants a direct signal or context change, ask only for the missing domain,
-   item, or context detail.
+1. Ask what preference question this catalog is meant to support.
+2. Ask what domain or concept area it belongs to.
+3. Ask what kinds of items should be included or excluded.
+4. Offer a working catalog name once the purpose is clear.
 
-Ready to act when:
+Helpful follow-up lanes:
+
+- what decision or taste question this catalog should help answer
+- what belongs in scope
+- what would make the catalog immediately useful instead of bloated
 
+Ready to save when:
+
+- the catalog has a stable purpose
 - the domain is clear
-- the user intent is clear: inspect, compare, signal, or context-edit
-- the missing item or context detail is clear enough
+- the boundary of what belongs inside is clear enough
+
+Preferred opening question:
+
+- "What preference question do you want this catalog to help answer?"
+
+## Preference Catalog Item
+
+Aim: add one candidate in a way that will make later comparisons feel clear and fair.
+
+Arc:
+
+1. Ask what makes this item worth including in the catalog.
+2. Ask what catalog or domain it belongs to if that is still unclear.
+3. Ask for a short clarifying description only if the label would be ambiguous later.
+4. Ask about aliases or tags only if they help retrieval.
+
+Helpful follow-up lanes:
+
+- why this item belongs in the comparison pool
+- what would distinguish it from nearby items
+- whether the label alone will be clear later
+
+Ready to save when:
+
+- the item label is clear
+- the parent catalog is clear
+- there is enough context to recognize it later if the label is ambiguous
+
+Preferred opening question:
+
+- "What makes this item worth comparing in that catalog?"
+
+## Preference Context
+
+Aim: define a real operating mode for preferences, not a decorative label.
+
+Arc:
+
+1. Ask what situation or mode this context is meant to represent.
+2. Ask what should count inside that context and what should stay outside it.
+3. Ask whether it should be active, default, or kept separate from other evidence.
+4. Offer a concise name if the mode is clearer than the wording.
+
+Helpful follow-up lanes:
+
+- what decisions this context should shape
+- what belongs inside versus outside the mode
+- whether it should be default or explicitly separate
+
+Ready to save when:
+
+- the context has a stable purpose
+- its boundary is clear enough to use consistently
+- any default or sharing choice that matters is clear
+
+Preferred opening question:
+
+- "What situation or mode should this context help Forge treat differently?"
+
+## Preference Item
+
+Aim: save one concrete preference candidate or signal without losing the context that
+makes it meaningful.
+
+Arc:
+
+1. Ask what preference or taste question this item belongs to.
+2. Ask what domain or context it should live in.
+3. Ask whether the user is trying to save it as a comparison candidate or as a direct
+   signal such as favorite, veto, or compare-later.
+4. Ask what makes this item distinct enough to compare usefully only if it is still a
+   comparison candidate.
+5. If the user is really trying to send a direct signal, ask only for the missing
+   detail and then act.
+
+Helpful follow-up lanes:
+
+- what domain this belongs to
+- what context makes the preference meaningful
+- whether this is a signal or a comparison candidate
+- what distinguishes the item from nearby options
+
+Ready to act when:
+
+- the item is clear
+- the domain or profile context is clear enough
+- any needed distinguishing detail is captured
 
 Preferred opening question:
 
-- "What preference are you trying to make clearer right now?"
+- "What preference are you trying to make clearer by saving this item?"
 
 ## Questionnaire Instrument
 
@@ -512,7 +695,8 @@ Arc:
 
 1. Ask what the questionnaire is meant to measure or surface.
 2. Ask who it is for and when it should be used.
-3. Ask what style of questions it needs before getting into item wording.
+3. Ask what kind of questions would make it honest and useful before getting into
+   item wording.
 4. Move to draft creation once the purpose is clear.
 
 Ready to act when:
@@ -551,9 +735,14 @@ consistent.
 
 Arc:
 
-1. Ask what category the label should capture.
+1. Ask what kind of moment or incident this label should capture in lived terms.
 2. Ask how narrow or broad it should be.
-3. Ask for a short description only if the label could be ambiguous later.
+3. Ask what would count as inside versus outside the category if that boundary is
+   still fuzzy.
+4. Ask for a short description only if the label could be ambiguous later.
+
+If the user already offered a candidate label, keep the wording provisional and ask
+what kinds of moments belong inside it before you ask whether the label is right.
 
 Ready to save when:
 
@@ -562,7 +751,7 @@ Ready to save when:
 
 Preferred opening question:
 
-- "What kind of incident should this event type capture?"
+- "What kind of moment do you want this event type to help you name more consistently?"
 
 ## Emotion Definition
 
@@ -570,7 +759,7 @@ Aim: define one reusable emotion entry clearly enough that future reports stay p
 
 Arc:
 
-1. Ask what emotion this should represent.
+1. Ask what this feeling is like in lived terms when the user says it.
 2. Ask what distinguishes it from nearby emotions if that matters.
 3. Ask for a short description only if later reports would benefit from it.
 
@@ -581,4 +770,4 @@ Ready to save when:
 
 Preferred opening question:
 
-- "Which emotion do you want this entry to help name more clearly?"
+- "When you say this feeling, what does it actually feel like for you?"

package/skills/forge-openclaw/psyche_entity_playbooks.md

@@ -27,17 +27,64 @@ Forge without turning the conversation into a worksheet.
 - Do not rush to naming, reframing, or repair before the experience itself is clearer.
 - Once the formulation is coherent, help with wording instead of forcing the user to
   generate all the language alone.
+- When the user is close to the experience but far from the wording, do some of the
+  formulation work for them. Offer one careful candidate name or summary and invite
+  correction.
 - Ask only one lane at a time. Do not jump from the example straight into belief,
   mode, value, and repair all in one turn.
 - If the user gives a rich answer with several possible lanes, choose the one that most
   improves understanding and leave the others for later.
 - Before the next question, reflect back what you just heard in one or two sentences so
   the user can feel understood and correct you if needed.
+- Before you ask for change, naming, or repair, usually ask what the experience is
+  trying to protect, prevent, or hold onto.
 - The warmth should come from accuracy and steadiness, not from extra softness,
   diagnosis language, or a polished performance of care.
 - If the user becomes flooded, vague, or circular, narrow the focus to the smallest
   concrete slice you can ask about: one moment, one sentence, one body signal, or one
   thing they did next.
+- When the user gives an abstraction first, gently bring it back to one recent moment
+  before helping them name the pattern.
+- Prefer "what", "when", and "how" early. Save "why" for later unless the user is
+  already grounded enough that it will deepen rather than blur the picture.
+- Before you ask for the user's preferred fix, understand what the response is trying
+  to protect, achieve, or prevent.
+
+## Therapist micro-skills
+
+Use these in small doses rather than as a script.
+
+Reflect:
+
+- briefly mirror the user's felt sense, stake, or sequence in plain language
+
+Locate:
+
+- ask where this shows up most clearly: the last moment, the first cue, the body
+  signal, the feared consequence, or the move that happened next
+
+Deepen:
+
+- choose one lane that improves understanding most:
+  situation,
+  sequence,
+  meaning,
+  protection,
+  cost,
+  longing/value,
+  or naming
+
+Formulate:
+
+- once the experience is coherent, offer one careful working sentence or title and ask
+  whether it feels true, too sharp, or still misses something important
+
+Connect:
+
+- near the end, notice whether this belongs with a value, belief, pattern, mode, note,
+  goal, project, or task and ask whether the user wants to map that too
+- when another record is clearly present, ask whether the user wants to link it now,
+  save it separately later, or just keep it in the note/context for now
 
 ## First-turn rule
 
@@ -59,6 +106,13 @@ Good first-turn shapes:
   that moment?"
 - "I can help slow it down. When do you notice this most clearly?"
 
+If the user is already direct and just wants to save a clear Psyche record, shorten the
+opening:
+
+- one brief reflection
+- one missing-detail question
+- then move toward the write instead of forcing exploration
+
 ## Lane chooser
 
 After each real answer, choose the next best lane. Do not mix several lanes at once.
@@ -118,6 +172,11 @@ Example shape:
   and pulling back gives you a little protection fast. What does withdrawing prevent in
   that moment?"
 
+Another strong shape when the user is getting abstract:
+
+- "So the painful part seems to be less the event itself and more what it starts to
+  mean about you. What does it seem to prove in that moment?"
+
 ## Formulation rules
 
 - Do not front-load a finished case formulation.
@@ -130,6 +189,8 @@ Example shape:
   detail into normalized fields.
 - Before saving, give a short working summary in the user's own language and ask
   whether it feels true enough.
+- A good final check is often:
+  "Does that feel true enough to save, or is there one part that needs to be said more accurately?"
 
 ## Naming turn
 
@@ -164,6 +225,21 @@ If the entity is not yet clear:
   meaning
 - then move toward naming and save
 
+## Psyche update loop
+
+Use this when the user is revising an existing Psyche record rather than creating one
+from scratch.
+
+1. Ask what feels newly true, newly visible, or newly inaccurate.
+2. Ask what still feels essentially right and should not be lost.
+3. Re-check the name only if the old wording no longer fits the lived experience.
+4. Ask for one concrete recent example if the update is abstract or sweeping.
+5. Then change only the parts the new understanding actually affects.
+
+Good update opener:
+
+- "Before we change it, what feels newly true about this now, and what still feels basically right?"
+
 ## Safety rule
 
 If the user shows imminent risk of self-harm, suicide, violence, inability to stay
@@ -224,6 +300,7 @@ Helpful follow-up lanes:
 
 - the earliest cue or vulnerability factor
 - the moment the loop starts to make sense internally
+- where the pattern stops feeling like a choice
 - what the loop protects or prevents
 - what it costs later
 - what a preferred response would still need to honor
@@ -299,6 +376,7 @@ Arc:
 Helpful follow-up lanes:
 
 - the feared consequence inside the belief
+- what the moment seems to prove
 - how old or familiar the rule feels
 - evidence for
 - evidence against
@@ -318,7 +396,7 @@ Ready to save when:
 
 Preferred opening question:
 
-- "When that reaction hits, what does it start telling you?"
+- "When that reaction hits, what does it start telling you about you, them, or what happens next?"
 
 ## Mode Profile
 
@@ -356,7 +434,7 @@ Ready to save when:
 
 Preferred opening question:
 
-- "When this part takes over, what is it trying to protect?"
+- "When this part takes over, what feels most at risk or most in need of protection?"
 
 ## Mode Guide Session
 
@@ -391,7 +469,7 @@ Ready to save when:
 
 Preferred opening question:
 
-- "What just happened that brought this part online right now?"
+- "What just happened, and what is this part trying to get you to do or stop doing right now?"
 
 ## Trigger Report
 
@@ -412,6 +490,7 @@ Helpful follow-up lanes:
 
 - what happened first
 - what emotions were present and how intense they felt
+- what felt most threatened or unbearable
 - what meaning or interpretation arrived fast
 - what action followed
 - what happened next short term and long term