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
Template/example gaps surfaced by comparing the shipped .dd-config.yaml.j2 against the live dd-sk config (2026-08-15 usage audit). The template's examples teach the config surface; these are places real usage diverged because the examples under-teach it.
Exclusion-pattern roles are undemonstrated. The live config leans on negative role matches (!*manager*, !*director*, !*vp*, !*recruiter*, !*intern*) — a real, shipped feature that appears nowhere in the template's role examples. Add one !-prefixed example with a comment explaining the syntax. (Also relevant to the "VP role-exclusion gap" noted in prior planning, and to job-search: keyword-based exclusion (negate keywords / match description, not just title) #150's broader exclusion design.)
max_enrich_fit: 50 reads as a reasonable default but is far too low for real use. The live workspace runs it at 5000 (effectively uncapped). Either raise the example, or comment it as "first knob to raise for a full-scale search" so users don't discover the cap by silently under-enriched runs.
domain_keywords replace-vs-merge behavior is documented only in the user's own workspace comment, not the shipped template (tracked as Setting domain_keywords silently replaces the package defaults #211 — whatever the semantics decision there, the template comment should state it).
Also worth considering
The daily-learning command outgrew its container: the workspace rewrote daily-learning.md (the other manifest-confirmed managed edit) from a flat dated log into a multi-track system (learning/<track>/{syllabus,progress}.md, cross-track flashcards.csv spaced-recall deck, lenses.md, a session operating protocol loaded before material). Whether to upstream the richer structure or leave daily-learning intentionally minimal is a product decision — filing the observation here rather than prescribing.
Type: docs / design
Template/example gaps surfaced by comparing the shipped
.dd-config.yaml.j2against the live dd-sk config (2026-08-15 usage audit). The template's examples teach the config surface; these are places real usage diverged because the examples under-teach it.!*manager*,!*director*,!*vp*,!*recruiter*,!*intern*) — a real, shipped feature that appears nowhere in the template's role examples. Add one!-prefixed example with a comment explaining the syntax. (Also relevant to the "VP role-exclusion gap" noted in prior planning, and to job-search: keyword-based exclusion (negate keywords / match description, not just title) #150's broader exclusion design.)max_enrich_fit: 50reads as a reasonable default but is far too low for real use. The live workspace runs it at 5000 (effectively uncapped). Either raise the example, or comment it as "first knob to raise for a full-scale search" so users don't discover the cap by silently under-enriched runs.domain_keywordsreplace-vs-merge behavior is documented only in the user's own workspace comment, not the shipped template (tracked as Settingdomain_keywordssilently replaces the package defaults #211 — whatever the semantics decision there, the template comment should state it).Also worth considering
The daily-learning command outgrew its container: the workspace rewrote
daily-learning.md(the other manifest-confirmed managed edit) from a flat dated log into a multi-track system (learning/<track>/{syllabus,progress}.md, cross-trackflashcards.csvspaced-recall deck,lenses.md, a session operating protocol loaded before material). Whether to upstream the richer structure or leave daily-learning intentionally minimal is a product decision — filing the observation here rather than prescribing.