Extract newsletter scripts to a neutral scripts/newsletter/ engine and add side-by-side support for all three AI coding tools off one source of truth. - Move 9 scripts + substack config from .claude/skills/**/scripts to scripts/newsletter/ (history preserved); fix PROJECT_ROOT depth and cross-require/config paths; repoint all SKILL.md invocations - Add canonical AGENTS.md; reduce CLAUDE.md to an @AGENTS.md import - Add opencode.json (permissions, no MCP); skills auto-discovered in place - Add codex/prompts/*.md (6 prompts) + copy installers (install.sh/.ps1) - Add docs/multi-tool-usage.md (setup, per-tool invocation, teardown) + README pointer
3.7 KiB
AGENTS.md
Canonical instructions for AI coding tools (Claude Code, OpenCode, Codex) working with this Hugo blog. This is the single source of truth; CLAUDE.md imports it.
Project Info
- Type: Hugo static site with Vietnamese tech content
- Theme: hugo-theme-stack
- Language: Vietnamese (with common English tech words)
- Timezone: Asia/Ho_Chi_Minh (UTC+7)
Directory Structure
content/post/
└── YYYY/
└── MM/
└── DD/
└── index.md
Starting Local Development Server
hugo server -D
The site will be available at http://localhost:1313
Options:
-Dor--buildDrafts: Include draft content-For--buildFuture: Include future-dated content--disableFastRender: Full re-render on all changes
Shared Engine
All portable newsletter scripts live in scripts/newsletter/ and are invoked from the repo root with plain Node (stdlib only, no deps):
node scripts/newsletter/add-url.js "<url>" # classify + dedup a URL → JSON route
node scripts/newsletter/find-newsletter-number.js # next newsletter number
node scripts/newsletter/list-existing-tags.js # existing tag frequencies
node scripts/newsletter/detect-image-source.js "<url>" # detect Substack image + uuid
node scripts/newsletter/find-substack-post.js --uuid <uuid>
node scripts/newsletter/fetch-via-defuddle.js "<url>" # fallback fetch (defuddle proxy)
These are shared by all three tools — no tool-specific copies.
Newsletter Workflow Routing
The newsletter workflow adds URLs (articles, YouTube videos, images) to today's newsletter post and manages tags. How it surfaces depends on the tool:
- Claude Code / OpenCode — invoke skills (both read
.claude/skills/<name>/SKILL.mdnatively):mt-add-url— meta dispatcher: classifies each URL, auto-invokes the right handler. Default entry for adding URLs.mt-add-post— article/blog URL → newsletter main contentmt-add-video— YouTube link → Bonus → Videosmt-add-image— image → Bonus → Images (labels Substack images via source-post lookup)mt-add-tags— add/update tags in post frontmattermt-webfetch— fallback web fetcher (defuddle proxy); use only when built-in WebFetch is blocked
- Codex — no skill auto-dispatch. Run
/prompts:mt-add-url <url>(installed prompts; see "Using Codex" below), or read this file and call thescripts/newsletter/*.jsscripts directly.
mt-add-url dispatches article / youtube / image; other types (direct video files, documents, unknown) prompt the user to add or extend a handler.
Skill implementations live in .claude/skills/.
Using Codex
Codex has no skill auto-dispatch and loads custom prompts only from ~/.codex/prompts/. Install the repo's prompt sources once, then invoke them as slash commands:
- Linux/macOS:
bash codex/install.sh - Windows:
pwsh codex/install.ps1(orpowershell codex/install.ps1)
Then in a Codex session at the repo root: /prompts:mt-add-url <url> (also /prompts:mt-add-post, mt-add-video, mt-add-image, mt-add-tags, mt-webfetch).
The installer copies prompts (not symlinks — Windows symlinks need admin/Developer Mode). Re-run it after editing any codex/prompts/*.md so ~/.codex/prompts/ re-syncs. Even without installing, Codex can read this file and call the scripts/newsletter/*.js scripts directly.
Git Workflow Rules
Before any git commit, check all staged content/post/*/index.md files for minimal tags. If any have only ["AI-Assisted"] or empty tags, run the mt-add-tags workflow on them first before committing.