You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Consider more opinionated sub-entities - problem statement, risk, assumption, insight, decision, principle #5
These wouldn't be full pages, but these things should be attached to OST nodes during product lifecycle.
Explore whether a structured entity type makes sense to define. Consider:
Pure AI assisted extraction of information from a page to populate a structure
Semi-structured conventions and templates for expressing these within a node/page (eg a risk table with a format matching the schema), potential validation helpers
Instead of a structured schema, define a set of descriptive goals, rules and activities that teams and agents can work from
Following research on #3 (findings in docs/torres-alignment-research.md):
A few refinements from the research:
Full page vs embedded is intentionally flexible. A schema defines an entity type (e.g. assumption, metric) in one place. Both a full-page file and an embedded heading within a primary entity page are valid expressions of that type — once #10 lands. The question of "should this be its own page?" is a content organisation decision, not a schema constraint.
Metric as a sub-entity attribute, not a standalone type. Rather than a new metric entity type, consider metric as a structured annotation on a goal or opportunity — a field or inline section with a defined shape. This keeps the entity hierarchy lean. A composable schema building block (#17) could define what a well-formed metric looks like for teams that want to enforce it.
Assumption and the fourth layer. Research confirms that Torres' bottom layer (assumption test) maps naturally to the sub-entities in this issue. For lightweight planning, assumptions can be inline sections on a solution page. For teams doing full discovery, they earn their own files. The same entity schema covers both. A fourth-layer entity type is being added to the general schema in #14.
This is becoming an overarching goal for being able to manage more structured information within a page.
That would suggest it involves some combination of #26#27#31#17 .
Imagining these things combined...
An opportunity page might have a heading with a problem statement. Because of defining natural language we can tell which section is the problem statement and parse it
The same page might have a list of assumptions. Because we can parse tables, a table of assumption entities with attributes can be parsed from the page too.
The template-sync knows how to make an opportunity page with the requisite sub-elements listed. This is because something in the schema metadata indicates the non-hierarchy entities (ie adjacent relationships) there are and how the should be rendered into a template. - eg assumptions are a table, with columns for the title and any required properties. problem statements are a heading. solutions are a bulleted list (and also happen to be part of the core hierarchy. Any of these can also be their own page if needed.
Activity
mindsocket commentedon Feb 26, 2026
Following research on #3 (findings in
docs/torres-alignment-research.md):A few refinements from the research:
Full page vs embedded is intentionally flexible. A schema defines an entity type (e.g.
assumption,metric) in one place. Both a full-page file and an embedded heading within a primary entity page are valid expressions of that type — once #10 lands. The question of "should this be its own page?" is a content organisation decision, not a schema constraint.Metric as a sub-entity attribute, not a standalone type. Rather than a new
metricentity type, considermetricas a structured annotation on a goal or opportunity — a field or inline section with a defined shape. This keeps the entity hierarchy lean. A composable schema building block (#17) could define what a well-formed metric looks like for teams that want to enforce it.Assumption and the fourth layer. Research confirms that Torres' bottom layer (assumption test) maps naturally to the sub-entities in this issue. For lightweight planning, assumptions can be inline sections on a solution page. For teams doing full discovery, they earn their own files. The same entity schema covers both. A fourth-layer entity type is being added to the general schema in #14.
mindsocket commentedon Mar 10, 2026
This is becoming an overarching goal for being able to manage more structured information within a page.
That would suggest it involves some combination of #26 #27 #31 #17 .
Imagining these things combined...
assumptionsare a table, with columns for the title and any required properties.problem statementsare a heading. solutions are a bulleted list (and also happen to be part of the core hierarchy. Any of these can also be their own page if needed.