EquiStamp avatar

EquiStamp

Official

@equistamp · United States of America

0Followers
|
12Public Repos
|
23Published Skills

A tool for measuring AI algorithms

Skills Distribution
DomainAI Models & ...Cognitive Methodol.. (40%)Safety Evaluation .. (40%)Model Porting & Pr.. (20%)

Agent Skills by EquiStamp

Showing 23 vetted skills indexed across 2 GitHub repositories.

EquiStampEquiStamp

evaluation-design

Design and review AI safety evaluations against the 7-dimension rubric.

Official
Advanced
EquiStampEquiStamp

evaluation-running

Run and back-test AI safety evaluations across model generations.

Official
Advanced
EquiStampEquiStamp

pairwise-compare-evals

Compare AI safety evaluations pairwise across predefined criteria to generate rankings.

Official
Advanced
EquiStampEquiStamp

generate-porting-feasibility-report

Generate a Google Docs-friendly porting feasibility report from an eval YAML.

Official
Advanced
EquiStampEquiStamp

extract-eval-properties

Generate a porting feasibility assessment and YAML metadata for Inspect.

Official
Advanced
EquiStampEquiStamp

productionization

Convert validated EquiStamp evaluations into automated production monitoring systems.

Official
Advanced
EquiStampEquiStamp

compare-porting-feasibility

Orchestrate parallel porting feasibility assessments across evals into YAML, reports, and HTML dashboards.

Official
Advanced
EquiStampEquiStamp

scenario-ideation

Generate structured scenario ideation documents for EquiStamp's Phase 1 eval pipeline.

Official
Advanced
EquiStampEquiStamp

evaluation-building

Translate a validated Phase 3 evaluation design into a Phase 4 Build Plan.

Official
Advanced
EquiStampEquiStamp

risk-modeling

Transform Phase 1 scenarios into formal risk models with causal pathways and measurable targets.

Official
Advanced
EquiStampEquiStamp

tech-writing

Draft or edit technical documents using EquiStamp's causal-flow writing method.

Official
Advanced
EquiStampEquiStamp
5

focusing

Guides users through a six-step felt-sense protocol for intuitive problem-solving.

Official
Intermediate
EquiStampEquiStamp
5

bucket-errors

Separate entangled cognitive questions using a five-step bucket-error process.

Official
Advanced
EquiStampEquiStamp
5

trigger-action-planning

Design, practice, and execute trigger-action plans for automatic behaviors.

Official
Intermediate
EquiStampEquiStamp
5

resolve-cycles

Run structured 5-minute time-boxed problem-solving cycles to overcome procrastination.

Official
Intermediate
EquiStampEquiStamp
5

double-crux

Identify and resolve disagreements by uncovering shared underlying cruxes.

Official
Intermediate
EquiStampEquiStamp
5

goal-factoring

Factor actions into underlying goals and generate alternatives.

Official
Intermediate
EquiStampEquiStamp
5

hamming-questions

Guide structured introspection sessions to identify important problems and bottlenecks.

Official
Intermediate
EquiStampEquiStamp
5

murphyjitsu

Stress-test a concrete plan by imagining its failure modes and designing defenses.

Official
Advanced
EquiStampEquiStamp
5

pride-self-recognition

Guides a structured frustration-to-value loop with three modes for self-reflection.

Official
Intermediate
EquiStampEquiStamp
5

aversion-factoring

Decompose aversions into addressable sub-aversions with button tests.

Official
Intermediate
EquiStampEquiStamp
5

freedom-of-action

Map perceived constraints and expand action space through structured steps.

Official
Intermediate
EquiStampEquiStamp
5

internal-double-crux

Facilitate a 5-step dialogue to resolve internal conflicts between competing motivations.

Official
Intermediate

Frequently Asked Questions About EquiStamp

FAQPage Schema
What specific tasks can I perform using EquiStamp?

EquiStamp enables structured cognitive problem-solving, the design and execution of safety evaluations, and the generation of porting feasibility reports. You can perform pairwise comparisons of model evaluations, build formal risk models from scenarios, and convert validated evaluation designs into production monitoring systems.

Who is the target persona for these methodologies?

The target personas include safety researchers, model evaluators, and technical leads focused on rigorous performance assessment. These methodologies are designed for engineers and analysts who require structured frameworks to decompose complex cognitive tasks, stress-test plans, and standardize the evaluation of model safety and feasibility.

What are the prerequisites for running EquiStamp evaluations?

To run evaluations, you require a validated Phase 3 evaluation design and access to the Inspect framework for YAML metadata processing. Users should be prepared to map scenarios into causal pathways and utilize the provided 7-dimension rubric to ensure consistency across model generations.