# Collections

> Hand-picked sets of resources, each built around one job, with a note on why every item is there. Structured data is at `/api/v1/collections`.

14 collections.

- [Spec, TDD and review for coding agents](https://hebbmind.com/collections/spec-tdd-review-coding-agents.md): For developers and tech leads who want coding agents to work like careful engineers: guardrails first, then interview, spec, test-first build, debug, verify, review. Most run in Claude Code or Codex; the three Anthropic plugins are Claude Code only. Agent review assists; a person approves merges. (12 resources)
- [UI design taste for agent-built apps](https://hebbmind.com/collections/ui-design-taste.md): For designers and frontend devs making agent-built web UI look designed. Most items need a coding agent like Claude Code or Codex, Node.js and web code. Set the system, explore, commit, review, polish, animate, test accessibility. Figma MCP: paid Dev/Full seat. Checks don't replace design review. (12 resources)
- [Code-made video: launches and explainers](https://hebbmind.com/collections/code-to-video-launch-explainers.md): For founders, marketers, developer advocates and teachers: Claude Code or Codex on your computer scripts, voices, builds, captions and publishes the video. ElevenLabs, HeyGen and Postiz cloud have paid plans; for-profit firms over 3 people need a Remotion license. Label AI voices and avatars. (15 resources)
- [Office documents: Word, Excel, slides](https://hebbmind.com/collections/office-documents-word-excel-slides.md): For office staff, analysts, consultants: read files and scans, build the sheet, write the report, make the deck, handle PDFs, share. Anthropic's docx, xlsx, pptx and pdf skills power Claude's file features; others need Python or Node. Check every number; keep confidential files in approved tools. (15 resources)
- [Product discovery, specs and delivery](https://hebbmind.com/collections/product-discovery-specs-delivery.md): For PMs and founders, in order: research, market scans, competitors, priorities, spec, prototype, tickets, analytics, re-planning, updates. Most install from a terminal into Claude Code or Codex. Lovable, Linear, Jira, PostHog need accounts. Keep customer data in approved tools; you set priorities. (15 resources)
- [SEO and AI search visibility](https://hebbmind.com/collections/seo-ai-search-visibility.md): For in-house marketers, content leads and solo founders: connect Search Console, audit, check AI crawlers and citations, research keywords, refresh posts, edit. Most run in Claude Code. Search-volume data costs money; AI-citation tracking is immature. Improve your pages; don't mass-produce. (13 resources)
- [Contract review and NDA triage](https://hebbmind.com/collections/contract-review-nda-triage.md): In-house counsel and legal ops: triage NDAs, review clauses and DPAs, score risk, redline in Word, sign. US-law defaults. Plugins need Claude Cowork or Code; others also run in Codex. Use an employer-approved plan for client data; counterparty files can hide prompt injections. A lawyer reviews all. (12 resources)
- [Month-end close and reconciliation](https://hebbmind.com/collections/month-end-close-reconciliation.md): Controllers, accountants, FP&A: plan the close, pull GL and bank data, draft accruals and JEs, reconcile, build statements, explain variances, QA the workbook. Skills run in Claude Code or Codex. Let agents draft, not post; a qualified accountant reviews. Assumes US GAAP/SOX. Use approved AI only. (14 resources)
- [K-12 lesson planning and classroom tools](https://hebbmind.com/collections/k12-lesson-planning.md): For US K-12 teachers and instructional coaches: unpack standards, plan, differentiate, write math checks, make slides, prep, post to Google Classroom or Canvas. K-12 skills use US standards and come with Claude for Teachers. De-identify student data, follow district AI policy, review before class. (11 resources)
- [Literature review to journal submission](https://hebbmind.com/collections/literature-review-to-submission.md): For grad students and postdocs in any field: search, screen and read, draft, fit a journal, check claims, tidy references, review and revise. Some steps need keys or sign-ins. Tools assist; you verify every citation and claim, follow the journal's AI policy and keep unpublished data private. (15 resources)
- [Building agents: from model API to evals](https://hebbmind.com/collections/build-your-first-agent.md): For engineers new to agent work, in build order: model API, first agent, tools over MCP (the open protocol for agent tools), data, context and memory, a framework, evals, tracing, sandboxing. Skills run in Claude Code or Codex, plugins in Claude Code. Your agents need a model API key, usually paid. (13 resources)
- [Investment research: filings and models](https://hebbmind.com/collections/investment-research-filings-models.md): For analysts, finance students and investors: get market and macro data, read SEC filings and transcripts, build and check comps and DCF models, test ideas with quant tools. Some need paid plans or keys; keep non-public data out of hosted tools. Research aid, not investment advice; check figures. (13 resources)
- [Applied economics with coding agents](https://hebbmind.com/collections/applied-economics-with-agents.md): For economics PhD students, applied microeconomists and policy analysts: set up a project, get FRED and SEC data, run Stata, R or Python, draft and revise. Agents assist; check every merge, estimate and sentence. Local tools still send what the agent reads to the AI provider: check data agreements. (14 resources)
- [Bioinformatics for bench biologists](https://hebbmind.com/collections/bioinformatics-for-bench-biologists.md): For wet-lab biologists moving to computational work: look up genes, find public data, run sequencing pipelines, then single-cell, stats and figures. Works in Claude Code and Codex unless noted; most need local Python. Keep patient data off untrusted services; have a bioinformatician review results. (12 resources)