integrate-gestalt
À propos
Cette compétence synthétise des entrées multidomaines provenant de `expand-awareness` en une seule et émergente perspicacité, supérieure à la somme de ses parties. Elle cartographie les tensions et les résonances entre les perspectives pour former un tout cohérent, en évitant une clôture prématurée. Utilisez-la après la collecte de perceptions brutes et avant la communication finale pour formuler une phrase unificatrice qu'aucun domaine isolé ne pourrait produire.
Installation rapide
Claude Code
Recommandénpx 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-gestaltCopiez et collez cette commande dans Claude Code pour installer cette compétence
Documentation
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
Dépôt GitHub
Compétences associées
evaluating-llms-harness
TestsCette compétence Claude exécute le lm-evaluation-harness pour évaluer les modèles de langage sur plus de 60 tâches académiques standardisées telles que MMLU et GSM8K. Elle est conçue pour permettre aux développeurs de comparer la qualité des modèles, de suivre les progrès de l'entraînement ou de rapporter des résultats académiques. L'outil prend en charge différents backends, incluant les modèles HuggingFace et vLLM.
cloudflare-cron-triggers
TestsCette compétence fournit une connaissance complète pour la mise en œuvre de Déclencheurs Cron Cloudflare afin de planifier des Workers à l'aide d'expressions cron. Elle couvre la configuration de tâches périodiques, de travaux de maintenance et de flux de travail automatisés, tout en traitant des problèmes courants tels que les expressions cron non valides et les problèmes de fuseau horaire. Les développeurs peuvent l'utiliser pour configurer des gestionnaires planifiés, tester des déclencheurs cron et intégrer avec Workflows et Green Compute.
webapp-testing
TestsCette Compétence Claude fournit une boîte à outils basée sur Playwright pour tester des applications web locales via des scripts Python. Elle permet la vérification frontend, le débogage d'interface utilisateur, la capture d'écrans et la consultation des journaux, tout en gérant les cycles de vie du serveur. Utilisez-la pour les tâches d'automatisation de navigateur, mais exécutez les scripts directement plutôt que de lire leur code source pour éviter la pollution du contexte.
finishing-a-development-branch
TestsCette compétence aide les développeurs à finaliser leur travail en vérifiant que les tests passent, puis en présentant des options d'intégration structurées. Elle guide le processus de fusion, de création de PRs ou de nettoyage des branches une fois l'implémentation terminée. Utilisez-la lorsque votre code est prêt et testé pour finaliser systématiquement le cycle de développement.
