Glossary

ForgeRun

Lifecycle container for a governed platform run.

EvidencePacket

Human-reviewable record linking intent, reasoning, execution, artifacts, and decision state.

Forge Agent

Bounded, governable agent session under a ForgeRun; recorded as AgentRun (arun_*) with a workcell id and WorkcellResult. See Forge Agent. Forge itself is not an agent.

Forge Campaign

Multi-repo, branch-aware automation effort under one ForgeRun, driven by a campaign manifest (forge/campaigns/<id>.yaml or .forge/runs/<frun_id>/campaign.yaml). See Forge Campaign. Not a marketing campaign or a charge/* Git branch.

Workcell

Bounded agent or tool runtime that accepts WorkcellRequest and emits WorkcellResult.

Versona

Discipline lens used to apply a specific review perspective or operating practice.

BlueprintContextPack

Portable, agent-readable process context generated from Blueprints and repo-local policy.

LCDLTraceSummary

Condensed trace of governed LLM reasoning, verification, and repair.

FleetTemplate

Approved execution recipe that Fleet can run under a known boundary.

ChatIntent

Safe transformation of chat/mobile ingress into a Forge-controlled request.

forge-workcells

Private implementation repo (autowww/forge-workcells) for bounded workcell runners. Platform owns schemas; workcells owns execution code. See Ecosystem reference.

micro-pack

Fixed prompt + manifest for a single micro-agent task (for example ux-ai-rule-v1 under Kitchen Sink tools/forge-micro-agent/packs/). Not a full Cursor rules tree.

local_llm_worker

Workcell id for local-LLM micro-agent runs (LCDL profile, KS-assembled context.json). Implemented in forge-workcells.