Alibi
SkillDev tools"Routes Alibi users to the right explanation, confidence,
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the Alibi skill
What this skill tells your AI
The instructions your AI receives, as published by vectorspacelab/arex-skill in skills/repositories/repo-skills/alibi/SKILL.md and read by ahel’s review.
Alibi is a Python library for model explanation, confidence scoring, prototype selection, and explanation persistence.
When to use this skill
Use this skill when a user asks how to:
- install Alibi or choose extras
- inspect public Alibi APIs
- run a small reproducible Alibi workflow
- decide which explanation family fits a task
- diagnose
MissingDependency, backend, or predictor-shape issues
First reads
references/workflows.mdfor the route mapreferences/optional-dependencies.mdfor extras andMissingDependencyplaceholdersreferences/troubleshooting.mdfor import, backend, and workflow failuresreferences/repo-provenance.mdwhen checking staleness or before refresh workreferences/api-reference.mdfor verified constructor signatures and return shapes
Minimal install and import
python -m pip install alibi- From a checkout:
python -m pip install -e . - Minimal import check:
python -c "import alibi; print(alibi.__version__)" - If the import trips over a spaCy /
clickissue, repair the base environment before assuming the package is broken.
Route map
sub-skills/global-tabular-explanations/
Use for ALE, partial dependence, PD variance, permutation importance, and other global tabular feature-effect workflows.
sub-skills/anchors-local-explanations/
Use for AnchorTabular, AnchorText, and AnchorImage workflows, including text sampling and image segmentation guidance.
sub-skills/attribution-and-shap/
Use for KernelShap, TreeShap, and IntegratedGradients, plus optional SHAP / TensorFlow backend selection.
sub-skills/counterfactuals-and-similarity/
Use for Counterfactual, CEM, CounterfactualProto, CounterfactualRL, and GradientSimilarity workflows.
sub-skills/confidence-prototypes-utilities/
Use for TrustScore, LinearityMeasure, ProtoSelect, and save/load round-trips.
Helper scripts
scripts/core_smoke.pyruns a tiny CPU smoke across the base tabular workflows.scripts/check_optional_backends.pyreports which optional exports are placeholders and which extra enables them.
What not to do
- Do not treat a
MissingDependencyplaceholder as a verified runtime path. - Do not send users to the source checkout for docs, examples, or notebooks.
- Do not assume optional backends are installed until the helper says so.
Signals
- GitHub stars
- 266
- Forks
- 21
- Last commit
- Sep 2026
ahel review
K1binfo
installs-packages
Automated review, not a security audit. Ruleset v1+k2.
Advanced
- Catalog kind
- skill
- Gateway key
alibi- Source
- github.com/vectorspacelab/arex-skill