Lambda AgentOS v0.1: skeleton (standards, constitution, prompts, scripts, CI)

[codegen-sh]

Requested by: Jake Ruesink

Summary
This PR bootstraps a Spec‑Driven development skeleton for our repos, blending GitHub Spec‑Kit’s workflow with Agent‑OS’ layered standards. It’s agent‑agnostic and safe to land without impacting runtime code.

What’s included

• .agentos/standards: tech‑stack, code‑style, best‑practices, medusa‑guidelines
• .agentos/product: mission, architecture, roadmap, decisions/
• .agentos/memory/constitution.md: non‑negotiable gates (simplicity, contract‑first, typed data flow, drift control)
• .agentos/templates: docs (spec/plan/tasks templates) and prompts (specify/plan/tasks/implement/review/medusa-plugin)
• .agentos/scripts: preflight/postflight + spec helpers (create-spec, generate-tasks, implement-tasks, medusa-checks)
• .agentos/agents/AGENTS.md: roles → prompt templates
• .github/workflows/lambda-agentos.yml: CI job to run pre/post‑flight on PRs
• .github/pull_request_template.md: prompts for spec link + gates

How to use

1. Create a new spec folder
   • ./agentos/scripts/create-spec.sh "Feature Title" → .agentos/specs/NNN-feature/
2. Write the spec (WHAT/WHY) → spec.md (use template)
3. Plan the implementation → plan.md (contracts first; RR7 loaders/actions, Tailwind, Prisma/Medusa only if applicable)
4. Generate tasks → tasks.md (parallel groups [P], explicit deps)
5. Implement
   • .agentos/scripts/preflight.sh
   • follow tasks.md (small commits referencing spec path)
   • .agentos/scripts/postflight.sh
6. Open PR and link the spec path (PR template asks for it)

Notes

• This repo is frontend‑only; medusa‑checks.sh is a noop here but is included for org consistency when we reuse the skeleton in Medusa repos.
• No app code changed; CI still runs existing lint/build/type/tests.

Follow‑ups (optional)

• Seed an example spec (e.g., “001 Contract‑first Fetch + Route example”) and drive it through plan → tasks → implement.
• Add a tiny‑spec path (template) for small fixes.

References

• Agent‑OS: https://github.com/buildermethods/agent-os
• Spec‑Kit: https://github.com/github/spec-kit
• Spec‑Driven methodology: https://github.com/github/spec-kit/blob/main/spec-driven.md

---

💻 View my work • About Codegen
⛔ Remove Codegen from PR • 🚫 Ban action checks

[codegen-sh]

Requested by: Jake Ruesink

Added the Research Agent prompt + wiring for Lambda AgentOS (Context7-first):

• New prompt: .agentos/templates/prompts/research.md
• Registered agent: .agentos/agents/AGENTS.md (added “Researcher (Research Agent)”)
• Usage notes (optional): .agentos/templates/docs/research-usage.md
• Dry‑run script stub (optional): .agentos/scripts/research.sh

Notes

• Follows existing prompt tone/structure (second-person, Goal + concise sections) while preserving the open-ended, evidence-driven flow.
• Context7-first path is explicit (resolve-library-id → get-library-docs, tokens≈4000); safe fallbacks to browser/search MCP tools; secrets redaction reminders included.
• Output structure matches acceptance criteria: classification, plan, 2–3 sources, synthesized findings, recommendation, repo next steps, risks.
• Local lint/typecheck passed.

If you want me to expand the dry-run into an automated check, I can add a tiny CI step that calls the stub and scans for required section headers (non-destructive).

[codegen-sh]

Requested by: Jake Ruesink

Follow-up tweak: made the Research Agent tool selection explicitly tool-agnostic (no Context7-first bias).

Changes

• .agentos/templates/prompts/research.md: neutralized ordering; Context7, Exa/Perplexity, and browser/search MCP are suggested options if available; otherwise rely on local repo/framework knowledge.
• .agentos/templates/docs/research-usage.md: updated “Tooling Options” section; removed “preferred/fallback” language; added guidance for no-tools scenario.
• .agentos/scripts/research.sh: wording updated accordingly (still a safe dry-run stub).

No code paths or CI changed; lint/typecheck still pass.