elizaos

star 3

End-to-end ElizaOS development, debugging, and operations for the LTCG monorepo. Use when working on ElizaOS agents, character files, plugin architecture (actions/providers/evaluators/services/routes), runtime settings, Convex-backed agent APIs, or agent docs in this repository.

LunchTable-TCG By LunchTable-TCG schedule Updated 2/8/2026

name: elizaos description: End-to-end ElizaOS development, debugging, and operations for the LTCG monorepo. Use when working on ElizaOS agents, character files, plugin architecture (actions/providers/evaluators/services/routes), runtime settings, Convex-backed agent APIs, or agent docs in this repository.

ElizaOS (LTCG Monorepo)

Overview

Use this skill for all ElizaOS work in this repository, including:

  • Plugin development in packages/plugin-ltcg
  • Character and runtime configuration in packages/plugin-ltcg/src/character.ts and packages/plugin-ltcg/src/characters/*.ts
  • Agent API integration in convex/http/*.ts
  • Agent docs in apps/docs-agents/src/content/docs
  • Runtime debugging, validation, and testing for ElizaOS agent behavior

Source Priority

Use this order to resolve uncertainty:

  1. Repository implementation (highest priority for this codebase)
  2. Official ElizaOS docs via Context7 (/elizaos/docs)
  3. DeepWiki ElizaOS pages for orientation and discovery, then verify against 1 and 2

Do not rely on DeepWiki alone for behavioral guarantees.

Load-Only-What-You-Need References

  • For official CLI, character format, and plugin interface patterns: read references/official-elizaos.md.
  • For repository file map, endpoint map, and environment variable map: read references/ltcg-elizaos-monorepo.md.
  • For concrete implementation procedures and checklists: read references/playbooks.md.

Workflow

1) Classify the request

Map the request into one primary lane:

  • Character/runtime lane: personality, model/provider, secrets/settings, startup behavior
  • Plugin lane: actions/providers/evaluators/services/routes/client wiring
  • API lane: agent registration, matchmaking, game actions, story/chat/deck endpoints
  • Docs lane: apps/docs-agents updates
  • Ops lane: debugging, testing, release hardening

2) Locate the implementation surface

Use rg and file maps from references/ltcg-elizaos-monorepo.md to find exact touch points before editing.

3) Apply the smallest coherent change

  • Keep changes scoped to the selected lane.
  • Reuse existing patterns already used in packages/plugin-ltcg/src.
  • Preserve established naming and component registration conventions.

4) Verify at the right level

Choose the smallest effective validation set:

  • Plugin-only checks in packages/plugin-ltcg
  • Repo-level type/tests only when the change crosses package boundaries
  • Endpoint checks when touching convex/http

5) Update docs when behavior changes

If runtime behavior, configuration, or API shape changes, update docs in apps/docs-agents/src/content/docs.

6) Update engineering memory

Before and after substantial work, use the repo memory loop:

bun run eng:start
bun run eng:log -- observation "<title>" "<details>"

If a wrong assumption or error occurred, log it with prevention:

bun run eng:log -- mistake "<title>" "<what went wrong>" "<prevention rule>"

Guardrails

  • Keep ElizaOS plugin registration coherent: if adding a component, wire it through the corresponding index and plugin export.
  • Never hardcode secrets. Read via runtime/environment settings.
  • Keep LTCG_API_KEY validation semantics intact (ltcg_ prefix) unless a deliberate migration is required.
  • For HTTP endpoints under convex/http, preserve auth wrapper choice (httpAction vs authHttpAction) intentionally.
  • When adding settings, update config schema and defaults together.
  • When adding API routes, ensure client constants and client methods stay aligned.

Quick Commands

Use these first:

# Monorepo
bun run dev:all
bun run type-check

# Plugin package
cd packages/plugin-ltcg
bun run dev
bun run start
bun run type-check
bun run test
bun run check-all

Completion Criteria

Consider work complete only when:

  1. Code changes are wired through all required registries and exports.
  2. Runtime/config/API compatibility is validated at least once.
  3. Agent docs are updated if user-facing behavior changed.
  4. No unresolved TODOs or partial wiring remain.
Install via CLI
npx skills add https://github.com/LunchTable-TCG/LTCG --skill elizaos
Repository Details
star Stars 3
call_split Forks 0
navigation Branch main
article Path SKILL.md
More from Creator
LunchTable-TCG
LunchTable-TCG Explore all skills →