factory-bootstrap

Plugin factory-bootstrap from marketplace ge-agent-factory in vamsiramakrishnan/ge-agent-factory. Commands read from the repo's manifest — 1 month ago.

Skills (37)

• impeccable

  Use when the user wants to design, redesign, shape, critique, audit, polish, clarify, distill, harden, optimize, adapt, animate, colorize, extract, or otherwise improve a frontend interface. Covers websites, landing pages, dashboards, product UI, app shells, components, forms, settings, onboarding,

• ge-agent-pipeline

  Use for GE Demo Generator pipeline work: capability-driven local mock data, fixture-backed source adapters, validation, preview, deploy plans, publish plans, and explicit live deploy/publish handoffs. Prefer the top-level ge CLI and workspace manifest over direct file edits. - "mock data" - "generat

• ge-agent-tracker

  Use to answer workspace and agent status questions in the capability-driven GE Demo Generator harness. Reads workspace.json, run events, artifacts, versions, selected skills, readiness, and nextActions. Explains what is ready, what is missing, and the next safe command. - "where is my agent" - "agen

• ge-api-service-adapter

  Use when a GE use case includes REST APIs, GraphQL, gRPC, RFC/BAPI, SOAP, SFTP/API, MCP, or other service interfaces. Generates fixture-backed source adapter contracts: OpenAPI, MCP SSE tool manifests, optional Cloud Run/API Gateway deployment notes, and ADK tool wiring guidance. - "API" - "REST" -

• ge-cloud-mock-data

  Use to orchestrate cloud-ready deterministic data generation as a collection of datastores and service adapters. Composes source mapping, Snowfakery YAML structured generation, OLTP packagers for AlloyDB/Firestore/Bigtable, OLAP BigQuery packaging, unstructured blob packaging for Cloud Storage, and

• ge-contextual-agent-interviewer

  Use for turning fuzzy GE demo requests into generation-ready briefs. Selects department, catalog use case or freeform path, systems, evidence, data volume, anomalies, and local/deploy/publish boundary with minimal questions. Produces explicit source-system, schema, document, API, datastore, row-coun

• ge-foolproof-adk-agent-builder

  Use when building or repairing a complete local-first Python ADK agent workspace for GE Demo Generator. Produces root_agent, source-system tools, deterministic fixtures, uv project files, tests, README, and capability manifest without requiring Google auth or live deployment. Refuses to scaffold an

• ge-iterative-agent-refinement

  Use after a workspace exists to audit, repair, and validate it against the GE local-first quality bar. Focuses on business specificity, fixture credibility, ADK correctness, evidence quality, UI/demo readiness, and capability manifest accuracy. - "iterate" - "refine" - "improve" - "make it better" -

• ge-olap-bigquery-packager

  Use to package Snowfakery-generated analytical rows for BigQuery: fact tables, dimension tables, metric snapshots, scoring outputs, external feed snapshots, schemas, NDJSON load files, and dataset manifests. - "BigQuery" - "OLAP" - "analytics table" - "fact table" - "metric snapshot" primary: mock_d

• ge-oltp-cloud-packager

  Use to package Snowfakery-generated OLTP data for Google Cloud operational datastores: AlloyDB for SQL, Firestore for document/JSON operational records, and Bigtable for high-volume keyed event/time-series data. - "AlloyDB" - "Firestore" - "Firebase" - "Bigtable" - "OLTP packager" - "NoSQL" primary:

• ge-snowfakery-structured-data

  Use for YAML-controlled structured data generation with Snowfakery. Owns OLTP and OLAP row recipes, relationships, distributions, edge cases, and canonical generated records before datastore-specific packaging. - "Snowfakery" - "snowfakery recipe" - "YAML data generation" - "structured rows" - "OLTP

• ge-unstructured-blob-packager

  Use to generate and package unstructured evidence blobs for mock systems: PDF, DOCX, Markdown, HTML, email transcripts, chat transcripts, contracts, policies, plan documents, and Cloud Storage manifests. - "PDF" - "DOCX" - "document blob" - "unstructured" - "Cloud Storage documents" - "plan document

• ge-usecase-source-mapper

  Use to inspect GE slide use cases and classify each source system into a datastore collection. Reads src/components/slides/use-cases architecture metadata and emits source maps for OLTP SQL, OLTP NoSQL, OLAP, unstructured blobs, API/service adapters, and runtime/model dependencies. - "source map" -

• admitting-agents

  Operates the release admission layer of the GE Agent Factory — minting and verifying signed Agent Passports and running the admission gate that fronts handoff. Use when a proven agent needs a passport (ge passport emit), when verifying evidence integrity before a release, when evaluating or enforcin

• authoring-okf-specs

  Converts a generated GE agent spec into a conformant OKF (Open Knowledge Format) v0.1 Knowledge Bundle and back, making OKF a portable, authorable BRD spec exchange format. Use when exporting a generated spec as a portable OKF knowledge bundle, authoring or exchanging a BRD in Open Knowledge Format,

• bringing-your-own

  Operates the BYO (bring the enterprise's own systems, evals, models, and policies) insertion points of the GE Agent Factory — profiling and redacting real systems, recording and comparing bounded read traffic, modeling twin mutations, compiling reviewed bindings into dispatch policy, validating and

• browsing-the-library

  Browses and instantiates the Agent Library — the catalog of pre-built OKF agent blueprints (300+ across verticals) — by showing inventory counts, searching and filtering blueprints, inspecting one blueprint's full metadata and lifecycle readiness, and creating a receipt-backed workspace from a chose

• building-simulators

  Builds, plans, materializes, calibrates, and validates GE mock data and enterprise source-system simulators. Use when a harness must generate scenario data, run Snowfakery, seed simulator overlays, profile and compare a real system read-only, model safe twin mutations, validate simulator conformance

• checking-workspaces

  Checks and repairs generated GE agent workspaces. Use when validating workspace readiness, running doctor or repair, inspecting spec-code trace, or unblocking preview, promotion, deploy-plan, or publish-plan gates.

• deploying-the-control-plane

  Builds and deploys the factory's OWN apps — factory gateway, console, presentation UI, worker, MCP services — and runs the self-service installer into a target project. Use when deploying or updating the control-plane Cloud Run services, building app images, running ge infra apply / installer/build-

• driving-live-proof

  Operates the live-behavior layer of the GE Agent Factory — driving the deployed agent, compiling and recording evalsets, live proof, and latency benching. Use when verifying a shipped agent through its streaming assist surface, recording or replaying cassettes, running ge drive / ge prove --live / g

• enriching-okf-blueprints

  Improve GE Agent Factory OKF blueprints after quality classification by adding serious evals, fixtures, multi-turn coverage, trajectory assertions, state assertions, adversarial tests, and domain invariant coverage. Use when working on OKF catalog specs, eval coverage, blueprint maturity, or enrichm

• grounding-interviews-with-documents

  Grounds the spec interview in user-supplied documents — upload a BRD/policy/schema, parse it server-side, preview/confirm, then inject it so the interview's questions and the agent-spec are evidence-backed; then edit the spec in the canvas and register. Use when the operator has source documents to

• guarding-the-factory

  The factory's safety rail for an autonomous operator — what is reversible, what is outward-facing, and what needs explicit human sign-off. Use when about to deploy, load data, publish, change IAM or cross-project, or take any destructive action, and whenever deciding whether to act or escalate.

• installing-the-factory

  Installs the GE Agent Factory from a bare machine to a proven working install — repo clone, mise bootstrap, toolchain provisioning (Bun, Python, uv, Terraform), dependency install, the ge command on PATH, and a verified first proof — checking each phase before starting the next. Use when setting up

• interviewing-specs

  Turns rough enterprise agent ideas into normalized GE use-case specs through a harness-driven interview. Use when Antigravity or another harness should interview the user, infer systems/data/tools/evals, register an interview-authored spec, or check spec completeness before factory planning.

• managing-access

  Manages who can reach the factory's surfaces — Cloud Run IAM run.invoker, IAP, and the opt-in Firebase (Google) sign-in on the console and deck — and keeps deploys single-tenant. Use when a service returns 403/is locked, when granting access, when choosing IAP vs Firebase vs run.invoker, or when enf

• navigating-factory-line

  Use when operating the full GE Agent Factory line end to end, choosing the right skill for the current step, moving from interview to spec to build to gates to release, and explaining what happens next in simple terms.

• observing-runs

  Observes factory and cloud runs without babysitting a terminal — polling submitted run status, fetching a stage's result/errors, tracking an agent's lifecycle state, reading eval coverage, and reading platform readiness. Use when a long-running build or release was submitted (especially a detached,

• operating-console

  Operates and extends the GE console. Use when working on apps/console command registry, persisted jobs, preflight checks, activity streams, Autopilot, Mission Contract display, workspace gates, or API route sentinels.

• operating-the-factory

  Drives the GE Agent Factory end to end like a console operator — turns a goal into the right next station, watches runs, resolves blockers, and decides build/repair/ship/escalate. Use when an autonomous harness (Antigravity) must operate the factory or console without a human clicking, or when decid

• planning-missions

  Plans Factory and Autopilot ownership for a GE agent run. Use when deciding local vs remote behavior, target gates, build ownership, repair/observe ownership, handoff, and resumability.

• recording-evidence

  Records and validates evidence for GE Factory and Autopilot runs. Use when handling ledger events, provenance, artifacts, blockers, repairs, resumability, repeated failures, or upstream fix recommendations.

• running-factory

  Runs GE factory build stages through the harness and CLI surfaces. Use when handling factory plans, workspace generation, data generation, package data, harness refinement, structured output parsing, or next-stage selection.

• running-release

  Runs the release end of the GE Agent Factory line. Use when handling deploy plans, cloud data load, runtime deployment, runtime polling, tool registration, Gemini Enterprise publishing, live verification, or local-to-cloud handoff.

• standing-up-the-platform

  Brings the factory's cloud planes up and green — factory/data/tool planes via ge up / data up / mcp deploy — and reads ge doctor to remediate readiness failures before any build or release. Use when doctor reports blockers, a plane is down, remote mode needs the platform stood up first, or an MCP/da

• triaging-runs

  Diagnoses a stuck, blocked, or failed factory run from its event timeline and blockers, then decides resume vs re-run vs upstream fix vs escalate. Use when a run is blocked/failed, a stage errored, a build/deploy didn't produce its artifact, or the operator must recover a run.

MCP servers (1)

• ge-agent-factory

  runs: bun tools/mcp-server.mjs

Similar plugins