Following the consolidated catalogue and the additive PR #57, one more idea for the skill layer, offered as options rather than recommendations: three further merges that would take the set from 27 to 24. I flag each as a genuine trade-off, because each one trades a precise per-tool activation match for removing one live overlap, and only pays off if the merged body keeps both modes obvious. You deliberately kept these separate in the consolidation, so treat this as "here is the case for each," not a push.
1. impact into preflight
impact is the highest-frequency "what breaks if I change this" entry. Folding it into preflight (the bundled single-symbol gate) removes the soft overlap between them, but risks weakening the raw-blast-radius activation unless a "quick blast-radius only" mode stays obvious in the merged body. Trade: one fewer overlap, at the cost of a useful precision distinction (raw impact vs bundled gate).
2. session-continuity plus daily into a catch-up umbrella
Two distinct windows: resume-from-your-own-anchor (session-continuity) versus a fixed last-24h briefing with hotspots and a self-audit (daily). An umbrella that routes by "which window" removes the overlap, but the merged body has to carry both modes or one intent degrades. Trade: fewer entries, at the cost of two genuinely different session moments sharing one activation surface.
3. codebase-exploration plus graph into an architecture umbrella
codebase-exploration is the narrative structured overview; graph is the graph-algorithm layer (PageRank, bridge symbols, communities, dependency paths). Merging them removes the "explore versus run-an-algorithm" overlap, but loses the narrative-versus-algorithm split unless the merged body keeps both depths. Trade: one fewer entry, at the cost of a real capability distinction (briefing versus algorithms).
Each is realizable as a thin-router umbrella with a references/ file per mode (the same progressive-disclosure shape the PR #57 directive docs describe), if you decide any is worth it. Happy to turn any of the three into a PR, or to leave them as-is. No strong opinion from me here: you have the design context I do not.
Following the consolidated catalogue and the additive PR #57, one more idea for the skill layer, offered as options rather than recommendations: three further merges that would take the set from 27 to 24. I flag each as a genuine trade-off, because each one trades a precise per-tool activation match for removing one live overlap, and only pays off if the merged body keeps both modes obvious. You deliberately kept these separate in the consolidation, so treat this as "here is the case for each," not a push.
1.
impactintopreflightimpactis the highest-frequency "what breaks if I change this" entry. Folding it intopreflight(the bundled single-symbol gate) removes the soft overlap between them, but risks weakening the raw-blast-radius activation unless a "quick blast-radius only" mode stays obvious in the merged body. Trade: one fewer overlap, at the cost of a useful precision distinction (raw impact vs bundled gate).2.
session-continuityplusdailyinto acatch-upumbrellaTwo distinct windows: resume-from-your-own-anchor (
session-continuity) versus a fixed last-24h briefing with hotspots and a self-audit (daily). An umbrella that routes by "which window" removes the overlap, but the merged body has to carry both modes or one intent degrades. Trade: fewer entries, at the cost of two genuinely different session moments sharing one activation surface.3.
codebase-explorationplusgraphinto anarchitectureumbrellacodebase-explorationis the narrative structured overview;graphis the graph-algorithm layer (PageRank, bridge symbols, communities, dependency paths). Merging them removes the "explore versus run-an-algorithm" overlap, but loses the narrative-versus-algorithm split unless the merged body keeps both depths. Trade: one fewer entry, at the cost of a real capability distinction (briefing versus algorithms).Each is realizable as a thin-router umbrella with a
references/file per mode (the same progressive-disclosure shape the PR #57 directive docs describe), if you decide any is worth it. Happy to turn any of the three into a PR, or to leave them as-is. No strong opinion from me here: you have the design context I do not.