aboutsummaryrefslogtreecommitdiffstats
path: root/docs
diff options
context:
space:
mode:
authorDanilo M. <danix@danix.xyz>2026-08-25 19:17:22 +0200
committerDanilo M. <danix@danix.xyz>2026-08-25 19:17:22 +0200
commit728d15912899d9c66838f0937931f61d75d217d4 (patch)
tree6e218ba3adc540be47654f25695ad5aca0b0bb1f /docs
parent6ae195baa09bc217d51f3fb0a38312dfdcc85fdb (diff)
downloadllamachat-728d15912899d9c66838f0937931f61d75d217d4.tar.gz
llamachat-728d15912899d9c66838f0937931f61d75d217d4.zip
fix: keep one-shot skills loaded until NewHEADmaster
Diffstat (limited to 'docs')
-rw-r--r--docs/superpowers/specs/2026-08-25-skills-design.md5
1 files changed, 3 insertions, 2 deletions
diff --git a/docs/superpowers/specs/2026-08-25-skills-design.md b/docs/superpowers/specs/2026-08-25-skills-design.md
index 20a3c0f..99baee6 100644
--- a/docs/superpowers/specs/2026-08-25-skills-design.md
+++ b/docs/superpowers/specs/2026-08-25-skills-design.md
@@ -13,8 +13,9 @@ with `name` and `description`, then the instructions.
This adds a skills subsystem: llamachat reads that directory, lets the model
load a skill on demand through a tool call, and lets the user load one by
typing `/name` or mentioning the name naturally. A loaded skill's text stays
-in context for the whole conversation in chat mode, and for the single reply
-in one-shot mode.
+in context for the whole conversation in chat mode. One-shot mode keeps it
+in memory (never in history) until New or the chip is clicked, so a skill
+loaded in a scratch window stays active for the rest of it.
## Skills directory and format