integrate-gestalt
Über
Diese Fähigkeit synthetisiert multidomänale Eingaben von `expand-awareness` zu einer einzigen, emergenten Erkenntnis, die größer ist als die Summe ihrer Teile. Sie kartiert Spannungen und Resonanzen zwischen Perspektiven, um ein kohärentes Ganzes zu formen, ohne vorschnell abzuschließen. Nutzen Sie sie nach dem Sammeln roher Wahrnehmungen und vor der endgültigen Kommunikation, um einen vereinenden Satz zu formulieren, den keine einzelne Domäne hervorbringen könnte.
Schnellinstallation
Claude Code
Empfohlennpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/integrate-gestaltKopieren Sie diesen Befehl und fügen Sie ihn in Claude Code ein, um diese Fähigkeit zu installieren
Dokumentation
Integrate Gestalt
Form coherent whole from expand-awareness panoramic perception → not avg / compromise / pick-best, but emergent pattern no single perspective alone could produce.
Use When
expand-awarenesssurfaced multi-domain perception → need unified insight- Multi-domain perspectives available → none accounts for all evidence
- Problem analyzed from angles → separate analyses must become > list
- "What does it all mean together?" → no obvious answer
- Synthesis keeps collapsing → "pick best domain" rather than forming new
- Before
express-insight(needs formed gestalt as input)
In
- Req: Multi-domain obs from
expand-awareness(or equivalent) - Opt: Original q / problem → prompted multi-domain scan
- Opt: Known constraints gestalt must satisfy
- Opt: Prior failed integration attempts (collapsed → single-domain)
Do
Step 1: Map Tensions
For each domain pair from panoramic perception → characterize rel. 3 possible: tension (disagree), resonance (reinforce from diff angles), orthogonal (unrelated).
Tension-resonance map:
Tension-Resonance Map
+-------------------+-------------------+-------------------------------+
| Domain Pair | Relationship | Detail |
+-------------------+-------------------+-------------------------------+
| A vs B | tension / | |
| | resonance / | |
| | orthogonal | |
| Evidence: | | What specifically disagrees, |
| | | reinforces, or is unrelated? |
| Implication: | | What does this relationship |
| | | suggest for the whole? |
+-------------------+-------------------+-------------------------------+
| A vs C | ... | ... |
+-------------------+-------------------+-------------------------------+
| B vs C | ... | ... |
+-------------------+-------------------+-------------------------------+
Fill row per pair. N domains → N(N-1)/2 pairs. >10 rows → group first, map between groups.
Prioritize tensions → most integrative info. Resonances confirm; orthogonals set aside; tensions demand resolution → gestalt found in how they resolve.
→ Map done, all pairs characterized w/ evidence. ≥1 genuine tension — no tensions → domains not different enough for emergence.
If err: All pairs resonance → agreeing at surface. Dig: agree for diff reasons? = hidden tension. No rels → expand-awareness too shallow → return + deepen obs.
Step 2: Find Figure
Gestalt psych: figure emerges from ground. Ground = Step 1 map. Figure = dominant pattern unifies most domains w/ fewest contradictions.
- Scan map for clusters → which domains resonate? Clusters → candidate figures
- Per candidate → "What single perspective makes sense of most obs?"
- Figure ≠ compromise (weakening domains till agree) ≠ selection (picking strongest). Is new frame recontextualizing obs
- Test: state candidate in 1 sentence. Feels like 1 input domain? If yes → not gestalt yet → domain answer in disguise
- Look at tensions: true figure often lives in space between disagreeing domains, not in either
Signs figure emerging:
- Multi tensions resolve together under same reframe
- Contradictory obs → complementary aspects of same phenomenon
- Figure explains why each domain saw what it saw, incl. why disagreed
→ 1-2 candidate figures as single sentences. Each recontextualizes obs (not selects). Accounts for major tensions.
If err: No figure → integration premature. Two paths: (a) return to expand-awareness + add missing domain → key perspective absent; (b) sit w/ tensions w/o forcing → some gestalts need incubation. Note state, return later.
Step 3: Test Whole
Candidate must survive 3 tests.
Test A — Tension account: Walk thru every Step 1 tension. Gestalt resolves / reframes / acknowledges as irreducible trade-off? Unaddressed → premature.
Test B — Single-domain origin: Could insight come from single domain? Specialist nods "we knew that" → collapsed back. True gestalt surprises every domain — each recognizes contribution not whole.
Test C — Coherence under rotation: Approach from each domain's perspective. Holds shape? Or looks diff per domain? Robust → same insight any angle; fragile → changes meaning.
Scoring:
- All 3 pass → Step 4
- A fails → incomplete → back to Step 2 w/ unresolved tensions as constraints
- B fails → not emergent → Step 2 + explicit exclude single-domain framings
- C fails → not coherent → may be 2 insights masquerading as 1. Split + test each
→ Candidate passes all 3, or failure mode clearly identified → guides Step 2 return.
If err: Repeated fails → domains may not form natural gestalt. Not every multi-domain obs → emergence. Honest answer = structured list of perspectives + tensions. Deliver map as output rather than force false unity.
Step 4: Name Insight
Articulate gestalt in single sentence domain specialist would not write from their domain alone. This sentence = deliverable.
- Write sentence. Should be:
- Specific → actionable / falsifiable
- General → encompass all contributing domains
- Surprising → ≥2 input domains
- Free of single-domain jargon (or deliberate recontextualized)
- Test against 3 criteria from Step 3 one more
- Opt: 1-para expansion → how gestalt emerged from domain contribs (= provenance, not insight)
- Record which domains contributed, key tensions, figure-ground rel → metadata → supports future integration
Named insight + provenance → input to express-insight.
→ Single sentence capturing gestalt + brief provenance para. Passes "no single domain" test. Any practitioner recognizes contribution but couldn't arrive alone.
If err: Sentence collapses → domain language → negation test: state what it's NOT. "Not security rec, not perf opt, not arch pattern — is [gestalt]." Negations clear frames → space for emergent formulation.
Check
- Tension-resonance map complete all pairs w/ evidence
- ≥1 genuine tension (not just diff of emphasis)
- Candidate articulated as reframe (not compromise / selection)
- Test A: major tensions resolved / reframed / acknowledged
- Test B: no single domain could produce alone
- Test C: holds shape from each domain's view
- Final insight → single sentence + provenance
Traps
- Averaging: Weaken each until superficial agree. Mush, not gestalt. Bland = averaging.
- King-making: Pick strongest domain's answer + dress in multi-domain lang. Test B catches → unsurprised specialist nod = king-making.
- Premature closure: Accept first candidate w/o testing. First = often obvious, not integrative.
- Forced unity: Insist gestalt exists when domains orthogonal. Orthogonal → structured lists, not gestalts — valid outcome.
- Jargon blending: Mix terms from domains → sounds integrative, means nothing. Every term must be independently meaningful.
→
expand-awareness— produces raw panoramic perception this skill integrates; always precedesexpress-insight— communicates formed gestalt; always followsbuild-coherence— selects between options w/ structured eval; integrate-gestalt forms new wholebrahma-bhaga— creates from void; integrate-gestalt creates from abundancemeditate— clears prior ctx → clean perception; useful before expand-awarenesscoordinate-reasoning— manages info flow in multi-path eval; complementary when coordinating reasoning threads
GitHub Repository
Verwandte Skills
evaluating-llms-harness
TestenDiese Claude Skill führt den lm-evaluation-harness aus, um LLMs über 60+ standardisierte akademische Aufgaben wie MMLU und GSM8K zu benchmarken. Sie wurde für Entwickler entwickelt, um Modellqualität zu vergleichen, Trainingsfortschritt zu verfolgen oder akademische Ergebnisse zu berichten. Das Tool unterstützt verschiedene Backends, einschließlich HuggingFace- und vLLM-Modelle.
cloudflare-cron-triggers
TestenDiese Fähigkeit bietet umfassendes Wissen zur Implementierung von Cloudflare Cron Triggers, um Workers mithilfe von Cron-Ausdrücken zu planen. Sie behandelt das Einrichten periodischer Aufgaben, Wartungsjobs und automatisierter Workflows, während häufige Probleme wie ungültige Cron-Ausdrücke und Zeitzonenprobleme behandelt werden. Entwickler können sie zum Konfigurieren geplanter Handler, zum Testen von Cron-Triggers und zur Integration mit Workflows und Green Compute verwenden.
webapp-testing
TestenDiese Claude Skill bietet ein Playwright-basiertes Toolkit zum Testen lokaler Webanwendungen durch Python-Skripte. Es ermöglicht Frontend-Verifizierung, UI-Debugging, Screenshot-Aufnahme und Log-Einblick bei gleichzeitiger Verwaltung von Server-Lebenszyklen. Nutzen Sie es für Browser-Automatisierungsaufgaben, führen Sie Skripte jedoch direkt aus, anstatt deren Quellcode zu lesen, um Kontextverschmutzung zu vermeiden.
finishing-a-development-branch
TestenDiese Fähigkeit unterstützt Entwickler dabei, abgeschlossene Arbeiten zu finalisieren, indem sie testet, ob Tests bestehen, und dann strukturierte Integrationsoptionen präsentiert. Sie leitet den Workflow für das Zusammenführen von Code, das Erstellen von PRs oder das Bereinigen von Branches nach Abschluss der Implementierung. Nutzen Sie sie, wenn Ihr Code bereit und getestet ist, um den Entwicklungsprozess systematisch abzuschließen.
