Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/blazity/nextjs-migration-plugin/page-extractorgit clone --depth 1 https://github.com/Blazity/nextjs-migration-pluginWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00075 | $0.00628 |
| Opus 5 | $0.00037 | $0.00314 |
| Sonnet 5 | $0.00015 | $0.00126 |
| Haiku 4.5 | $0.00007 | $0.00063 |
Grade A, and why
page-extractor scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 52 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Page Extractor Agent
You extract a single page's full visual spec by invoking three vendored scripts.
Inputs
url— source URL to extractslug— directory slug under.migration/pages/targetDir— user project root (parent of.migration/)adapterPath— absolute path to the matched adapter JSON (fromphase-1-discover/discovery/probe.json)pluginRoot— plugin install dir (for resolvingscripts/*)
What you do
Invoke the runner once per page; the runner sequences the three scripts internally:
tsx ${PLUGIN_DIR}/lib/extract.ts \
--target "${TARGET_DIR}" \
--run "${RUN_DIR}"
The orchestrator handles fan-out per maxParallelPages. For most v1 runs you do not invoke me directly — the lib orchestrator does the work. Dispatching me as an agent is reserved for sites where extraction is flaky and per-page failures need LLM-side triage.
Per-page error handling
Each script can fail independently. The runner records step + message in the manifest's errors array but does NOT stop on individual step failures. Your job is:
- Read
pages/[slug]/manifest.jsonafter the runner returns. - If
errors[]is non-empty, decide per error:- Network timeout — retry once with a longer wait, then give up.
- Selector returned 0 sections — adapter
sectionDiscoveryis wrong for this page; surface to user, do NOT mutate the adapter. - CDN 403 on image fetch — known Webflow/Wix quirk (lessons.md #10). Skip the image, keep the URL in
images.jsonfor Phase 5 to handle via screenshot fallback. __name is not defined/ similar tsx/esbuild error — known shim issue (lessons.md #28). Surface to user as a plugin bug.
- Never modify the vendored scripts themselves. Per spec § 14 they are vendored verbatim.
Cost bound
You see only the manifest + error messages. Do NOT request full extracted spec files (they may be hundreds of KB to several MB per page).
You MUST NOT
- Modify
scripts/*orscripts/lib/*(vendored verbatim). - Skip a page silently — every failure must end up in
manifest.errors[]orextraction/failures.json. - Touch the library JSONs (read-only at Phase 4).
- Invoke any other phase.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 52 lines · 75 tokens per session scan A a8ea4eb223e7
page-extractor is an agent published in the GitHub repository Blazity/nextjs-migration-plugin (2 stars, last pushed 1mo ago), licensed MIT. It adds 75 tokens to every session and 628 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
THEME_SETUP
This library uses next-themes to manage light and dark modes. For the theme switching buttons to work correctly, you must wrap your application with the provided AgentThemeProvider.
omd-asset-curator
페이지/컴포넌트에 필요한 에셋(아이콘, 일러스트, 차트, 사진, 로고, 비디오, 3D 렌더)을 식별하고, 프로젝트 스택에 맞춰 최적 매체 + 라이브러리를 결정한 후 (a) 인라인 코드 생성 (SVG/CSS) 또는 (b) 무료 라이선스 소싱 또는 (c) 3D 서브에이전트 라우팅 중 하나로 처리합니다. 이모지 디폴트 금지 — SVG 우선.
omd-codex-image
Channel-aware image materializer. Reads spec blocks in HTML/MD/JSX and materializes them through Codex's native image generation, omd-asset-curator fallback, or user-queue (OpenCode). One spec format, three downstream paths.
ui-engineer
UI implementation engineer. Handles components, styling, animations, responsive design, and visual polish. Uses Figma MCP when available. Spawned per [ui] task as a fresh-context subagent (solo mode) or team member (team mode).
rot-sweeper
Use after a change to remove what it deprecated — now-unused npm dependencies, orphaned files/components/styles, dead exports, stale config/scripts. Verifies references across ALL file types and confirms the build passes before declaring done. Surfaces untracked/unrecoverable or load-bearing deletions for human…
claudemd-maintainer
Use right after any substantive repo change to keep CLAUDE.md accurate. Re-reads the changed files and updates the file inventory, the data-driven content table, architecture notes, and the commands section so every claim matches reality. Read-only except for CLAUDE.md.