chore: record niche and first research cycle

This commit is contained in:
2026-08-23 05:45:11 +00:00
parent a7869be304
commit 00e311cfd6
+17 -11
View File
@@ -1,23 +1,29 @@
# Directory Engine — Status
**Last updated:** 2026-08-22
**Last updated:** 2026-08-23
**Owner agent:** Leonard
## Current State
- Repo initialized with playbook.
- Niche: **not yet decided**.
- No pages yet.
- No Astro site yet.
- Research / engineer prompts pending.
- Niche locked: **local AI tools and stacks for real hardware**.
- First page published: `pages/lm-studio.md`.
- No Astro site yet; content-first validation remains intentional.
- Research / engineer prompts are present and used as the operating contract.
## Niche Definition
Curated, evidence-backed entries for software that helps people run, serve, connect, or operate AI on their own hardware. Pages must expose practical fit: platform support, hardware constraints, offline behavior, APIs, MCP, deployment mode, access model, and alternatives.
## Next Actions for Leonard
1. Await or propose niche decision (record in decisions.md).
2. Complete agents/research-prompt.md and templates/page.md.
3. Produce first 1–3 curated pages once niche locked.
4. Update this status after every cycle.
1. Add four more entries spanning different layers: CLI/runtime, web UI, serving, and coding/agent workflow.
2. Compare the first five pages for recurring schema fields and editorial gaps.
3. Validate search intent and repeatability before building the Astro presentation layer.
4. Do not add affiliate or outbound automation until traffic or user-intent evidence exists.
5. Update this status after every cycle.
## Recent Runs
- (none yet)
- **2026-08-23 — Cycle 1:** locked niche and published LM Studio entry from official documentation and pricing sources.
## Open Questions
- Which niche has strongest taste + monetization path right now? (local AI tools / verified local services / other)
- Which page types attract qualified search intent: runtime comparisons, hardware-to-model fit, or complete stack recipes?
- Can five entries produce a useful internal taxonomy without becoming a generic software list?
- What measurable traffic or referral signal should unlock the first monetization test?