Skip to content

Glossary

omk docs (blog posts, SKILL.md, CLI output, report pages) freely mix industry-standard ML / statistics / measurement terms. These words are de facto standard in the English community, so this table is a quick-reference index: each entry gives a one-line definition + where it shows up in omk.

Scope: a reader's cheat sheet, not a design spec. omk maintainers follow this vocabulary when writing new docs.

Sibling docs: terminology spec (maintainer-internal decision record) / statistical rigor / composite-score construct validity


1. Statistics / measurement

TermOne-line definitionWhere it shows up in omk
bootstrap CIPercentile confidence interval computed by resampling the registered sampling unit (1000 draws by default)Core Bootstrap-family Analysis
Δ (delta)Mean difference in composite score between treatment and controlBootstrap comparison estimate and Studio projection
95% CIAn interval from a procedure with 95% long-run coverage under its assumptions. A treatment-minus-control interval excluding 0 is directionally significantCore Bootstrap-family Analysis
significantThe registered comparison interval excludes 0 at its effective family-corrected alphaCore Decision evidence
Pearson rPearson correlation coefficient. 1 = perfectly aligned / 0 = unrelated / -1 = perfectly opposedJudge-ensemble or Gold agreement diagnostics
MADMean absolute difference among observed judge-member means for the same sampleJudge Ensemble Analysis
Krippendorff αAgreement statistic using the registered interval-distance definitionExplicit Gold comparison or preregistered Agreement Analysis
effect sizeThe treatment-minus-control estimate on the registered score scaleBootstrap comparison and practical-effect gate
varianceDispersion across independent run-mean composite valuesEvaluation Series Analysis
holdout (set)Independent validation samples the skill never explicitly covered, used to guard against sample-set overfittingpost-evaluation follow-up recommendations
construct validityWhether the measurement actually measures the intended thing (vs measurement error)scoring.md: composite-score construct-validity argument
ad hocAn implementation choice made without a principled justification — typically "ship it first, justify later"scoring.md: equal-weight composite aggregation is ad hoc
sample-set overfittingThe evaluation set happens to be "already answered," inflating scoresscoring.md / evaluation blog caveat section
length debiasCorrection for the known LLM-judge bias of scoring longer answers higheron by default; disable with omk eval --no-debias-length

2. omk evaluation concepts

TermOne-line definitionWhere it shows up in omk
artifactThe unified abstraction for omk's "thing under evaluation": skill / prompt / agent / workflow / baselinedetermined by experiment role (--control / --treatment / baseline), not a standalone flag
executorHow the model is run: claude / codex / openai-api--executor parameter; execution-environment fingerprint
ensemble (judge)Multiple LLMs act as judges and score independently, then combine--judge-models claude:opus,claude:sonnet
judgeAn LLM scoring against a rubricjudge model parameter; evidence table
rubricThe detailed criteria a judge follows when scoring (must recognize X / must include Y / at least N items / ...)rubric field in sample config
anchorA method for calibrating the LLM judge against human standards--gold-dir human anchors
gate (layer gate)A registered treatment-layer threshold evaluated from authenticated Composite-layer evidenceCore release Decision
verdictOne of six conclusions: PROGRESS / REGRESSION / CAUTIOUS / NOISE / UNDERPOWERED / SOLOCore Report, CLI route, and Studio projection
sample (evaluation sample)A single evaluation caseeval-samples.json
eval-samplesThe sample config file (each entry has prompt / rubric / assertion / capability)omk eval --samples
baseline (reserved variant)The empty-artifact variant; omk reserves this variant name, but it becomes the control only when selected for that experiment role--control baseline
treatmentThe experiment role compared with control; it may contain any artifact kind--treatment <name>
controlThe experiment role used as the reference side; it may contain a baseline or any other artifact kind--control <name>
composite (score)Equal-weight mean of observed present fact / behavior / judge layers on a 1-5 scale; zero observed layers is missingCore Composite table and Studio projection
fact (layer)Explicitly classified fact-criterion pass weight mapped to 1-5Assertion-layer Analysis
behavior (layer)Explicitly classified behavior-criterion pass weight mapped to 1-5Assertion-layer Analysis
judge (layer)Ensemble consensus or dimension aggregate bound as the judge sourceComposite Analysis
dimensionAn Analysis aggregate bound one-to-one to a Metric and upstream judge-ensemble resultDimension Analysis
evidence coveragePlanned, observed, missing, invalid, failed, unavailable, and not-started evidence retained through lineageCore Analysis and Decision gates
managed recordA .omk/governance/managed/<id>.json fact record from omk install (source / contentHash / distribution / evidence / decisions)omk install; evidence-gated management
lifecycle (installed / measurable / stale)Read-time state of a managed skill: installed (no valid evidence) → measurable (eval evidence bound) → stale (content drifted off its evidence)deriveManagedState; omk eval "→ measurable"
evidence (managed)A ManagedEvidenceRef an eval run appends to a managed record, bound to the content fingerprint it measured (report id / sample coverage / verdict / comparability)omk eval auto-write

3. Machine learning / AI general

TermOne-line definition
promptThe input text given to an LLM
system promptBackground instructions injected before user input; for evaluation omk injects the entire SKILL.md as the system prompt
agentAn AI that can call tools and run over multiple turns
workflowA multi-step AI process orchestration
skillOne of omk's core evaluation targets, usually in the form of a SKILL.md
tool callAn external function the LLM invokes during execution
turnOne interaction unit of "LLM output + user/tool response"
contextAll the history the LLM sees while generating
fingerprintA stable runtime or content identity used to verify consistency and comparability across runs
session traceThe event stream of one complete AI conversation (prompt / tool calls / output / scoring), the object observe parses

4. omk's three stages

omk's loop runs in three stages — doctor (preflight health) → eval (offline A/B + verdict) → observe (production traces) — together covering knowledge evaluation + management + insight. See the three stages for the full mental model.


Writing conventions

The terms above are the shared vocabulary for omk docs. Detailed naming decisions for omk-internal terms (artifact / executor / variant / verdict, etc.) live in the terminology spec (a maintainer archive). The Chinese docs additionally follow GB/T 15834 punctuation and a set of translation rules, documented in the Chinese glossary; they don't apply to English prose.