The network, and what this site does not own
Seven ideas in this corpus belong to other sites. This site states each in three lines and links out, rather than re-arguing them.
| Site | Owns | This site |
|---|---|---|
| skills.sgit.ai | The skills thesis, the catalogue, authoring, the skill economy, skills + identity | — |
| wardley-maps.sgit.ai | The skill lifecycle (Explorer → Town Planner, English-to-code, the token-spend line) | Three lines and a link |
| open-source.sgit.ai | The OWASP monetisation thread, fork economics, the villagers argument | Owns the skills mechanism; links for the arguments |
| graphs.sgit.ai | Meaning through connectivity, G³ | Owns the application of graphs to skills |
| coding.sgit.ai | Code conventions, /for-agents/ | The skills-describe-intent claim is the same argument from the other side; cross-linked |
| standards.sgit.ai | Vault-as-substrate, the subset method | Base vaults are that architecture applied to skills |
| llms.sgit.ai | sg.llm.*, the chat pane | Source of the when-not-to-make-a-skill decision; that site owns the capability |
| sgit.ai | Vaults, the catalogue | Base vaults live there physically |
The lifecycle, in three lines
Skills “start as massive LLM, held together with string and gum… all the way to code, and it almost becomes a commodity” — the Explorer→Product→Town Planner arc, with the consequence “anybody who spends a lot of money on tokens has an engineering problem.” Read the full argument on wardley-maps.sgit.ai →
The house pattern
This site was built by copying pki.sgit.ai’s structure via standards.sgit.ai, its most recent carrier: the same nav/footer generator (admin/build/chrome.py), the same validate-then-tag-then-deploy pipeline, the same footer boilerplate. What this site adds on top: the catalogue and divergence generator, the skill-specific CSS components, and llms-full.txt from day one rather than added later — because this site’s subject is the artefact type agents load, so agent self-sufficiency is an acceptance criterion, not an extra.