Krill
Community@Krilliac
Krill maintains a 16-skill operations registry for the Sonder Runtime, covering diagnostics, self-modification, QLoRA training, fleet orchestration, and change control.
Agent Skills by Krill
Showing 31 vetted skills indexed across 2 GitHub repositories.
sparkengine-change-control-and-release-readiness
Manages SparkEngine release gates, readiness work items, and same-SHA claim promotion rules.
sparkengine-rendering-rhi-rendergraph-and-shaders
Documents SparkEngine's RHI backends, RenderGraph contracts, and shader compilation workflows.
sparkengine-failure-archaeology
Documents SparkEngine's solved bugs with symptoms, root causes, fix commits, and current status.
sparkengine-debugging-playbook
Diagnoses SparkEngine subsystem failures using symptom-to-cause lookup tables and discriminating experiments.
sparkengine-ecs-lifecycle-threading-and-memory
Documents SparkEngine ECS phase wiring, threading rules, physics stepping, and memory ownership contracts.
sparkengine-validation-and-qa
Guides test registration, execution, and CI gate validation for the SparkEngine C++ test suite.
sparkengine-networking-security-and-multiplayer
Guides SparkEngine UDP multiplayer networking, authentication security, and replication work with verified code anchors.
sparkengine-editor-scenes-and-reflection
Diagnose and fix SparkEditor scene loading, reflection-driven Inspector editing, and undo/redo issues.
sparkengine-research-frontier-and-methodology
Guides engine research, benchmarking, and maturity-ladder promotion for SparkEngine development decisions.
sparkengine-architecture-contract
Enforces SparkEngine's subsystem ownership, service locator, and ECS phase-ordering design invariants.
sparkengine-assets-import-and-package-integrity
Diagnose SparkEngine asset loading, SparkPak archive integrity, and VFS path-containment issues.
sparkengine-build-ci-and-dependencies
Diagnose and resolve SparkEngine CMake builds, CI failures, and third-party dependency issues.
sparkengine-persistence-save-and-migrations
Diagnose and modify SparkEngine save files, async database transactions, and JSON store durability.
sparkengine-modules-sdk-abi-and-hot-reload
Diagnose and manage SparkEngine game-module DLL loading, ABI gating, and hot reload.
sparkengine-run-package-and-release
Launch, smoke-test, package, and publish SparkEngine binaries via CPack and GitHub release workflows.
sparkengine-docs-wiki-and-generated-truth
Regenerates and validates SparkEngine wiki, README, and generated documentation against source code.
sonder-diagnostics-and-tooling
Measure Sonder Runtime health using doctor, preflight, status, logs, and read-only benchmark scripts.
sonder-security-and-privacy
Reviews diffs against Sonder Runtime security boundaries and privacy consent gates.
sonder-failure-archaeology
Documents historical incidents, root causes, and fix disciplines for the Sonder Runtime repository.
sonder-agents-and-fleets
Operate Sonder autopilot runs, orchestrator fleets, model fanout, and multi-PC Ollama worker pools.
sonder-docs-and-writing
Writes and maintains Sonder Runtime documentation across ADRs, runbooks, wiki pages, and evidence ledgers.
sonder-config-and-flags
Catalogs every Sonder Runtime environment variable, config precedence rule, and consent gate.
sonder-change-control
Classifies repository changes by blast radius and enforces CI gates before commits.
sonder-server-decomposition-campaign
Executes incremental strangler-migration slices decomposing server.py into the sonder_runtime package.
Frequently Asked Questions About Krill
FAQPage SchemaWhat tasks can I perform using Krill's Sonder Runtime skills?▼
You can operate the runtime lifecycle (start, stop, drain, backup), run health diagnostics, triage failures, execute server.py decomposition slices, manage QLoRA adapter training and rollback, orchestrate autopilot fleets and Ollama worker pools, enforce commit gates, and write ADRs, runbooks, and release notes.
Who are these Sonder Runtime skills designed for?▼
They target maintainers and engineers working inside the Sonder Runtime repository: operators running the server day to day, developers preparing commits and PRs under blast-radius gating, and engineers running self-modification deploys, decomposition slices, or model training and promotion cycles.
How do I set up and run the Sonder Runtime environment?▼
Use sonder-build-and-env to recreate the working environment from a bare checkout: create the venv, install the correct requirements file, and run compile checks and pytest collection. Then sonder-run-and-operate covers launching via python -m sonder_runtime subcommands, serving, and backups.
What prerequisites and configuration do the Sonder skills require?▼
A Sonder Runtime repository checkout is required. sonder-config-and-flags documents every SONDER_* environment variable, config precedence, runtime policy, and consent gate; sonder-build-and-env handles venv and dependency installation before any operational, testing, or training skill is exercised.
How do the skills handle security and safe changes?▼
sonder-security-and-privacy defines the security model, privacy boundaries, and redaction guarantees a maintainer must never weaken. sonder-change-control classifies changes by blast radius, mandates gates before commit or merge, and specifies what must never be committed, including protected control-plane files.