Skip to main content
Alphabetical one-line definitions. For full explanations with diagrams, follow the Learn More links to the relevant concept page.
TermDefinitionLearn More
AgentThe system being versioned. Its configuration is captured as a manifest; traces, manifests, and datasets are all scoped to an agent.Execution Model
agentversionThe open spec (and PyPI package) for the manifest, diff, and compatibility-decision format DecimalAI builds on.Versioning & Compatibility
BaselineThe manifest a regression check diffs against — typically the last manifest seen in production.Regression Check
Compat StatusConvenience field on a trace: keep, repair, replay, or drop. (compatible/incompatible are accepted as legacy aliases that fold to keep/drop.) Authoritative source is the TraceCompat table.Versioning & Compatibility
Compatibility ReportAnalysis generated when a new manifest is registered, classifying every existing trace as keep/repair/replay/drop.Versioning & Compatibility
Compatibility Score0.0–1.0 metric measuring how well a trace matches a manifest version. Category = compatibility on the EvalScore record.Evaluation
Compatibility VerdictPer-trace outcome: keep, repair, replay, or drop.Versioning & Compatibility
ComponentA single versioned piece of a manifest: tool, model, prompt, skill, subagent, or output_schema.Versioning & Compatibility
Component VerdictPer-component outcome when diffing old vs new manifests: COMPATIBLE, REPAIRABLE, INCOMPATIBLE, or MISSING.Versioning & Compatibility
Content HashPer-component SHA-256 fingerprint. Changes when the component’s definition changes.Versioning & Compatibility
DatasetCurated training data built from filtered, scored production traces.Skills & Data Pipeline
Decision EngineSystem that combines quality + compatibility scores into a single keep/repair/replay/drop verdict per trace.Evaluation
DegradedParent trace status when some (not all) child traces errored.Multi-Agent Systems
DelegationOrchestrator → sub-agent task assignment. Control returns to orchestrator after sub-agent completes.Multi-Agent Systems
Detection SourceHow a manifest was created: auto (from traces) or manual (via SDK/API).Versioning & Compatibility
DPODirect Preference Optimization. Dataset format with chosen/rejected pairs for preference training.Skills & Data Pipeline
DriftWhen a sub-agent’s actual config diverges from what the orchestrator’s manifest expects.Multi-Agent Systems
EpisodeCompatibility engine synonym for “trace” — the same RunTrace record, just referred to in a compat context.Execution Model
Eval ScoreSingle evaluation result: name, score (0.0–1.0), passed (bool), source, category.Evaluation
Eval VerdictAggregate trace-level outcome: pass, fail, or review. Computed from all eval scores.Evaluation
EvaluatorA configured quality check — deterministic, LLM-as-judge, or custom.Evaluation
HandoffLateral transfer of control between peer agents. Unlike delegation, the original agent may not regain control.Multi-Agent Systems
Impact ReportThe structural, per-PR output of a regression check: each production trace marked HIGH / MEDIUM / LOW IMPACT by the manifest diff. The structural axis — distinct from the Compatibility Report (the data-lifecycle side).Regression Check
Impact SeverityHIGH / MEDIUM / LOW — how structurally a manifest change touches a trace (“was this trace affected?”). Orthogonal to the keep/repair/replay/drop Compatibility Verdict. Not the same scale as component Severity (none/minor/moderate/major).Compatibility Policies
LLM CallSingle model invocation — rendered prompt, completion, tokens, latency, cost, tool calls.Execution Model
ManifestSnapshot of an agent’s full configuration (tools, models, prompts, skills, sub-agents) at a point in time.Versioning & Compatibility
Manifest HashSHA-256 fingerprint of manifest structure. Same hash + same agent = idempotent.Versioning & Compatibility
Manifest StatusLifecycle state: active, superseded, or draft.Versioning & Compatibility
OrchestratorAn agent that delegates to sub-agents. Inferred from manifest components or trace linkage — not explicitly declared.Multi-Agent Systems
Parent TraceAn orchestrator’s trace record that child traces link back to via parent_trace_id.Multi-Agent Systems
Quality ScoreEval score measuring output quality (relevance, helpfulness, safety). Category = quality on EvalScore.Evaluation
Regression CheckThe pre-deploy GitHub Action that diffs a candidate manifest against the baseline and posts an Impact Report comment on the PR.Regression Check
RepairDeterministic fix of a trace to match a new manifest. Zero LLM cost.Skills & Data Pipeline
ReplayRe-running a historical trace against the current agent to compare outputs.Skills & Data Pipeline
RevertWhen the same manifest hash reappears — the old manifest is reactivated, the current one superseded.Versioning & Compatibility
SessionGroup of traces forming a multi-turn conversation. Linked by session_id.Execution Model
SeverityHow impactful a component change is: none, minor, moderate, or major.Versioning & Compatibility
SFTSupervised Fine-Tuning. Dataset format with input→output pairs for imitation learning.Skills & Data Pipeline
SkillReusable instruction file (SKILL.md) that modifies agent behavior. Not a tool.Skills & Data Pipeline
Skill ActivationRecord of which skills were loaded during a trace.Skills & Data Pipeline
skillevaluationThe open spec (and PyPI package) for A/B benchmarking a skill — runs each test case with and without the skill, then reports the measured lift.skillevaluation
SkillScoreThe 0–100 quality composite that ranks skills in the registry — from benchmark lift, live eval pass rates, and AI-judge quality, not install counts.SkillScore
Source TypeWhere a trace came from: production, playground, test, or replay.Execution Model
SpanTimed segment within a trace (llm, tool, retriever, other). Nests via parent_span_id.Execution Model
Sub-agentAgent receiving delegated work. Identified by parent_trace_id on its trace.Multi-Agent Systems
SurfacePolicy grouping for compatibility rules: tool_registry, model_runtime, prompt_stack, skill_registry, subagents, output_contract.Versioning & Compatibility
TraceA single, complete agent execution from input to output. The atomic unit of the platform.Execution Model
TrajectoryML/RL term for a full (state, action, reward) sequence. DecimalAI traces are similar but use eval scores instead of explicit rewards.Execution Model
TurnA single interaction within a session. Each turn produces one trace.Execution Model
Version LabelHuman-readable manifest identifier (v1, v2, v3). Auto-incremented — not semantic versioning.Versioning & Compatibility