skills/multimodal-hypothesis-lab/SKILL.md
--- name: multimodal-hypothesis-lab description: Scientific reasoning loop that uses multiple MMX modalities to test a hypothesis. Activates when: (1) Arif proposes a geological model or interpretation; (2) a complex question needs visual + textual + data evidence; (3) Arif asks to "prove" or "test" an interpretation. Runs: hypothesis → request image/video evidence → mmx vision analysis → GEOX consistency check → CLAIM/PLAUSIBLE/HYPOTHESIS verdict. metadata: {"openclaw": {"emoji": "🔬"}} --- #
npx skillsauth add ariffazil/openclaw-workspace skills/multimodal-hypothesis-labInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Geoscience is not a single-modality discipline. A depositional model must be consistent with core, wireline, seismic, and analogs. This skill runs a multi-step hypothesis test using MMX's modalities as a evidence engine.
HYPOTHESIS
↓ propose in text
PLAN EVIDENCE NEEDS
↓ what modalities would confirm or refute?
REQUEST EVIDENCE
↓ mmx image (schematic), mmx video (animation), mmx vision (real photos), mmx text (analog data)
ANALYZE EVIDENCE
↓ GEOX labels on all evidence
CONSISTENCY CHECK
↓ does evidence support / refute / is inconclusive?
VERDICT
↓ CLAIM (supported) / PLAUSIBLE (weak support) / HYPOTHESIS (unsupported)
State the hypothesis clearly with GEOX labels:
INT: The Upper Sandstone in Field X is a turbidite channel complex.
SPEC: This interpretation is based on gamma ray log motif alone.
NEEDS: seismic amplitude character + analogous core photo to confirm.
What would confirm or refute this? List:
mmx image (schematic) or ask Arif to providemmx visionmmx text chat --model MiniMax-Text-01mmx search# Generate schematic of expected seismic character
mmx image "seismic section showing turbidite channel complex, amplitude map, two-way-time depth"
# Request analogous core photo
mmx image "core photo turbidite sandstone amalgamated channels"
# Search published analogs
mmx search "turbidite channel seismic amplitude character field analog"
For each image received, run through geo-vision-translator to get OBS/DER/INT labels.
EVIDENCE ITEM 1: [description]
- Supports hypothesis? YES/PARTIAL/NO
- GEOX label: OBS/DER/INT
- Uncertainty: [what it can't confirm]
EVIDENCE ITEM 2: ...
| Verdict | Definition | When to Use | |---|---|---| | CLAIM | Multiple independent evidence types support it | Consistent seismic + well + analog | | PLAUSIBLE | Some support, but key evidence missing | Supported by one modality, unconfirmed by others | | HYPOTHESIS | Plausible but unsupported | Single line of evidence or theoretical only | | REFUTED | Evidence contradicts the hypothesis | Explicit counter-evidence from data |
Hypothesis: "The N5 sandstone in Well A is shoreface, not turbidite."
STEP 1 — PROPOSE
INT: N5 sandstone = shoreface depositional environment
SUPPORTING: gamma ray motif blocky,proximal
REFUTING: absence of turbidite indicators
STEP 2 — PLAN
Needed: (1) seismic amplitude at N5 level, (2) analogous shoreface core photo, (3) wireline shape comparison
STEP 3 — REQUEST
mmx image "shoreface sandstone core photo well sorted"
mmx search "shoreface vs turbidite gamma ray motif distinction"
STEP 4 — ANALYSIS (geo-vision-translator on core photo)
OBS: well-sorted, fine-to-medium sandstone, horizontal lamination
INT: consistent with shoreface, not diagnostic of turbidite
SPEC: turbidite would show amalgamation surfaces and clast variety
STEP 5 — CONSISTENCY
Evidence 1 (core analog): SUPPORTS shoreface (partial)
Evidence 2 (regional search): Gamma blocky motif ambiguous — both possible
Evidence 3 (mmx text): Shoreface model plausible but not confirmed
STEP 6 — VERDICT: PLAUSIBLE
"Not refuted, but insufficient evidence to distinguish from turbidite.
Need: biostratigraphy, seismic amplitude at N5, pressure data."
MULTIMODODAL HYPOTHESIS TEST
Hypothesis: [stated clearly with GEOX label]
Confidence before: [HIGH/MEDIUM/LOW]
Evidence gathered:
- [modality] | [finding] | supports: YES/PARTIAL/NO | label: OBS/DER/INT
Consistency: [strong / mixed / weak]
VERDICT: [CLAIM / PLAUSIBLE / HYPOTHESIS / REFUTED]
Confidence after: [HIGH / MEDIUM / LOW]
Remaining gap: [what would upgrade the verdict]
development
Check every skill’s “use when” and “do not use when” clauses for collisions, missing negatives, and vague verbs like “help,” “assist,” or “improve.” Load when linting, reviewing, or validating trigger boundaries.
development
Bootstrap, design, and package new skills. Load when capturing user intent for a new skill or drafting its initial instruction framework.
content-media
Diagnose which federation services are up, down, or drifting. Produce a prioritized remediation plan.
business
Scan a repo or workspace for exposed secrets, tokens, keys, and credentials. Produce a findings report with remediation steps.