mirror of
https://github.com/Alkatrazz24/Funk-lab.git
synced 2026-07-08 10:04:42 +02:00
fix: ask-agent /no_think + SKILL.md renforcé + tests 10/10
- ask-agent : ajout /no_think pour agents system et monitor (Qwen3-1.7B en mode thinking consommait tous les tokens sur le raisonnement interne, laissant content: "" vide) - SKILL.md v1.1.0 : règles OBLIGATOIRES de délégation, distinction ask-agent vs delegate_task, exemples concrets, syntaxe exacte - test.sh : suite 10 tests end-to-end — CLI disponible, LiteLLM, system/monitor/brain, données réelles, skill présence/activation, Hermes funk-ai appelle Terminal: ask-agent Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
parent
bd7181b8db
commit
5e93c1d3d7
3 changed files with 240 additions and 49 deletions
|
|
@ -33,6 +33,13 @@ AGENT="$1"
|
|||
QUESTION="$2"
|
||||
shift 2
|
||||
|
||||
# Qwen3 thinking mode : les agents rapides (system/monitor) doivent répondre
|
||||
# directement sans passer du temps en raisonnement interne.
|
||||
# /no_think désactive le mode thinking pour system et monitor.
|
||||
case "$AGENT" in
|
||||
system|monitor) QUESTION="/no_think ${QUESTION}" ;;
|
||||
esac
|
||||
|
||||
while [ $# -gt 0 ]; do
|
||||
case "$1" in
|
||||
--max-tokens) MAX_TOKENS="$2"; shift 2 ;;
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue