chore(deps): update dependencies

- bump mcp-go to v0.47.1
- update cloud auth, otel, and various indirect deps

go.mod

@@ -21,7 +21,7 @@ require (
 	github.com/fsnotify/fsnotify v1.9.0
 	github.com/indaco/herald v0.13.0
 	github.com/indaco/herald-md v0.3.0
-	github.com/mark3labs/mcp-go v0.47.0
+	github.com/mark3labs/mcp-go v0.47.1
 	github.com/spf13/cobra v1.10.2
 	github.com/spf13/viper v1.21.0
 	github.com/traefik/yaegi v0.16.1
@@ -31,7 +31,7 @@ require (
 
 require (
 	cloud.google.com/go v0.123.0 // indirect
-	cloud.google.com/go/auth v0.19.0 // indirect
+	cloud.google.com/go/auth v0.20.0 // indirect
 	cloud.google.com/go/auth/oauth2adapt v0.2.8 // indirect
 	cloud.google.com/go/compute/metadata v0.9.0 // indirect
 	github.com/Azure/azure-sdk-for-go/sdk/azcore v1.21.0 // indirect
@@ -58,9 +58,9 @@ require (
 	github.com/charmbracelet/harmonica v0.2.0 // indirect
 	github.com/charmbracelet/lipgloss v1.1.1-0.20250404203927-76690c660834 // indirect
 	github.com/charmbracelet/x/cellbuf v0.0.15 // indirect
-	github.com/charmbracelet/x/exp/charmtone v0.0.0-20260330094520-2dce04b6f8a4 // indirect
+	github.com/charmbracelet/x/exp/charmtone v0.0.0-20260406091427-a791e22d5143 // indirect
 	github.com/charmbracelet/x/exp/ordered v0.1.0 // indirect
-	github.com/charmbracelet/x/exp/slice v0.0.0-20260330094520-2dce04b6f8a4 // indirect
+	github.com/charmbracelet/x/exp/slice v0.0.0-20260406091427-a791e22d5143 // indirect
 	github.com/charmbracelet/x/exp/strings v0.1.0 // indirect
 	github.com/charmbracelet/x/json v0.2.0 // indirect
 	github.com/charmbracelet/x/termios v0.1.1 // indirect
@@ -81,7 +81,7 @@ require (
 	github.com/googleapis/enterprise-certificate-proxy v0.3.14 // indirect
 	github.com/googleapis/gax-go/v2 v2.21.0 // indirect
 	github.com/gorilla/websocket v1.5.3 // indirect
-	github.com/kaptinlin/go-i18n v0.3.0 // indirect
+	github.com/kaptinlin/go-i18n v0.3.1 // indirect
 	github.com/kaptinlin/jsonpointer v0.4.17 // indirect
 	github.com/kaptinlin/jsonschema v0.7.7 // indirect
 	github.com/kaptinlin/messageformat-go v0.4.19 // indirect
@@ -103,8 +103,8 @@ require (
 	github.com/yosida95/uritemplate/v3 v3.0.2 // indirect
 	github.com/yuin/goldmark v1.8.2 // indirect
 	go.opentelemetry.io/auto/sdk v1.2.1 // indirect
-	go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc v0.67.0 // indirect
-	go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.67.0 // indirect
+	go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc v0.68.0 // indirect
+	go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.68.0 // indirect
 	go.opentelemetry.io/otel v1.43.0 // indirect
 	go.opentelemetry.io/otel/metric v1.43.0 // indirect
 	go.opentelemetry.io/otel/trace v1.43.0 // indirect
@@ -114,9 +114,9 @@ require (
 	golang.org/x/net v0.52.0 // indirect
 	golang.org/x/oauth2 v0.36.0 // indirect
 	golang.org/x/time v0.15.0 // indirect
-	google.golang.org/api v0.274.0 // indirect
+	google.golang.org/api v0.275.0 // indirect
 	google.golang.org/genai v1.52.1 // indirect
-	google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20260406210006-6f92a3bedf2d // indirect
 	google.golang.org/grpc v1.80.0 // indirect
 	google.golang.org/protobuf v1.36.11 // indirect
 	gopkg.in/yaml.v2 v2.4.0 // indirect
@@ -128,13 +128,13 @@ require (
 	github.com/charmbracelet/x/term v0.2.2 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
 	github.com/lucasb-eyer/go-colorful v1.4.0 // indirect
-	github.com/mattn/go-isatty v0.0.20 // indirect
-	github.com/mattn/go-runewidth v0.0.22 // indirect
+	github.com/mattn/go-isatty v0.0.21 // indirect
+	github.com/mattn/go-runewidth v0.0.23 // indirect
 	github.com/muesli/cancelreader v0.2.2 // indirect
 	github.com/muesli/termenv v0.16.0 // indirect
 	github.com/rivo/uniseg v0.4.7 // indirect
 	github.com/spf13/pflag v1.0.10 // indirect
 	golang.org/x/sync v0.20.0 // indirect
-	golang.org/x/sys v0.42.0 // indirect
+	golang.org/x/sys v0.43.0 // indirect
 	golang.org/x/text v0.35.0
 )

go.sum

@@ -10,8 +10,8 @@ charm.land/lipgloss/v2 v2.0.2 h1:xFolbF8JdpNkM2cEPTfXEcW1p6NRzOWTSamRfYEw8cs=
 charm.land/lipgloss/v2 v2.0.2/go.mod h1:KjPle2Qd3YmvP1KL5OMHiHysGcNwq6u83MUjYkFvEkM=
 cloud.google.com/go v0.123.0 h1:2NAUJwPR47q+E35uaJeYoNhuNEM9kM8SjgRgdeOJUSE=
 cloud.google.com/go v0.123.0/go.mod h1:xBoMV08QcqUGuPW65Qfm1o9Y4zKZBpGS+7bImXLTAZU=
-cloud.google.com/go/auth v0.19.0 h1:DGYwtbcsGsT1ywuxsIoWi1u/vlks0moIblQHgSDgQkQ=
-cloud.google.com/go/auth v0.19.0/go.mod h1:2Aph7BT2KnaSFOM0JDPyiYgNh6PL9vGMiP8CUIXZ+IY=
+cloud.google.com/go/auth v0.20.0 h1:kXTssoVb4azsVDoUiF8KvxAqrsQcQtB53DcSgta74CA=
+cloud.google.com/go/auth v0.20.0/go.mod h1:942/yi/itH1SsmpyrbnTMDgGfdy2BUqIKyd0cyYLc5Q=
 cloud.google.com/go/auth/oauth2adapt v0.2.8 h1:keo8NaayQZ6wimpNSmW5OPc283g65QNIiLpZnkHRbnc=
 cloud.google.com/go/auth/oauth2adapt v0.2.8/go.mod h1:XQ9y31RkqZCcwJWNSx2Xvric3RrU88hAYYbjDWYDL+c=
 cloud.google.com/go/compute/metadata v0.9.0 h1:pDUj4QMoPejqq20dK0Pg2N4yG9zIkYGdBtwLoEkH9Zs=
@@ -96,14 +96,14 @@ github.com/charmbracelet/x/conpty v0.1.1 h1:s1bUxjoi7EpqiXysVtC+a8RrvPPNcNvAjfi4
 github.com/charmbracelet/x/conpty v0.1.1/go.mod h1:OmtR77VODEFbiTzGE9G1XiRJAga6011PIm4u5fTNZpk=
 github.com/charmbracelet/x/errors v0.0.0-20240508181413-e8d8b6e2de86 h1:JSt3B+U9iqk37QUU2Rvb6DSBYRLtWqFqfxf8l5hOZUA=
 github.com/charmbracelet/x/errors v0.0.0-20240508181413-e8d8b6e2de86/go.mod h1:2P0UgXMEa6TsToMSuFqKFQR+fZTO9CNGUNokkPatT/0=
-github.com/charmbracelet/x/exp/charmtone v0.0.0-20260330094520-2dce04b6f8a4 h1:pIj18ZCZO4WOVj7jwjLoUb1lC7rS/I8oC3fZWXugNaY=
-github.com/charmbracelet/x/exp/charmtone v0.0.0-20260330094520-2dce04b6f8a4/go.mod h1:nsExn0DGyX0lh9LwLHTn2Gg+hafdzfSXnC+QmEJTZFY=
+github.com/charmbracelet/x/exp/charmtone v0.0.0-20260406091427-a791e22d5143 h1:zmBor0ftFNqVFp9U59ZoEDRUCIYSGOGSIfGGkNZRufs=
+github.com/charmbracelet/x/exp/charmtone v0.0.0-20260406091427-a791e22d5143/go.mod h1:nsExn0DGyX0lh9LwLHTn2Gg+hafdzfSXnC+QmEJTZFY=
 github.com/charmbracelet/x/exp/golden v0.0.0-20250806222409-83e3a29d542f h1:pk6gmGpCE7F3FcjaOEKYriCvpmIN4+6OS/RD0vm4uIA=
 github.com/charmbracelet/x/exp/golden v0.0.0-20250806222409-83e3a29d542f/go.mod h1:IfZAMTHB6XkZSeXUqriemErjAWCCzT0LwjKFYCZyw0I=
 github.com/charmbracelet/x/exp/ordered v0.1.0 h1:55/qLwjIh0gL0Vni+QAWk7T/qRVP6sBf+2agPBgnOFE=
 github.com/charmbracelet/x/exp/ordered v0.1.0/go.mod h1:5UHwmG+is5THxMyCJHNPCn2/ecI07aKNrW+LcResjJ8=
-github.com/charmbracelet/x/exp/slice v0.0.0-20260330094520-2dce04b6f8a4 h1:VSd4zShIAf/4FgEDFJpapEcAPrc7h3dyyN7V9JlJpQw=
-github.com/charmbracelet/x/exp/slice v0.0.0-20260330094520-2dce04b6f8a4/go.mod h1:vqEfX6xzqW1pKKZUUiFOKg0OQ7bCh54Q2vR/tserrRA=
+github.com/charmbracelet/x/exp/slice v0.0.0-20260406091427-a791e22d5143 h1:aEppolah2k9c0LzKX2fk5ryuyQ0Lq8kCOjkvMw1b8o4=
+github.com/charmbracelet/x/exp/slice v0.0.0-20260406091427-a791e22d5143/go.mod h1:vqEfX6xzqW1pKKZUUiFOKg0OQ7bCh54Q2vR/tserrRA=
 github.com/charmbracelet/x/exp/strings v0.1.0 h1:i69S2XI7uG1u4NLGeJPSYU++Nmjvpo9nwd6aoEm7gkA=
 github.com/charmbracelet/x/exp/strings v0.1.0/go.mod h1:/ehtMPNh9K4odGFkqYJKpIYyePhdp1hLBRvyY4bWkH8=
 github.com/charmbracelet/x/json v0.2.0 h1:DqB+ZGx2h+Z+1s98HOuOyli+i97wsFQIxP2ZQANTPrQ=
@@ -185,8 +185,8 @@ github.com/indaco/herald v0.13.0 h1:+xVG9Fx5NpuWhwku/9IlRL6I009NnX4VUGKvlZHTRxU=
 github.com/indaco/herald v0.13.0/go.mod h1:T5g1+XLYvpjouhzAGHnAHDCKizhESkoV6+QPZ3DhgWA=
 github.com/indaco/herald-md v0.3.0 h1:hN1cKyrexPPM9PeHBsKuaWvIizSi/iYvM9yzRgtdb8M=
 github.com/indaco/herald-md v0.3.0/go.mod h1:RUHVaDSG45ymJjKyxpDwBocLXrZo93FB4OeYMsw9B9s=
-github.com/kaptinlin/go-i18n v0.3.0 h1:wP76dvYg04bvwTb+8NB+CmdZ2kL7lSSCQ9B/kFv7QHo=
-github.com/kaptinlin/go-i18n v0.3.0/go.mod h1:pVcu9qsW5pOIOoZFJXesRYmLos1vMQrby70JPAoWmJU=
+github.com/kaptinlin/go-i18n v0.3.1 h1:plXi3XQE1aYamFi8TU0K6actODmw2+5FSobmhTkfQ/0=
+github.com/kaptinlin/go-i18n v0.3.1/go.mod h1:ZRoAHj7elWYamfbv7wev7Ajch6LOzjtBaq8nWe8HIVk=
 github.com/kaptinlin/jsonpointer v0.4.17 h1:mY9k8ciWncxbsECyaxKnR0MdmxamNdp2tLQkAKVrtSk=
 github.com/kaptinlin/jsonpointer v0.4.17/go.mod h1:SsfsjqnHG5zuKo1DTBzk1VknaHlL4osHw+X9kZKukpU=
 github.com/kaptinlin/jsonschema v0.7.7 h1:41BlQJ9dskH0oE5DSzBUrl/w4JQYIr6N6L0B5GNyDoM=
@@ -201,12 +201,12 @@ github.com/kylelemons/godebug v1.1.0 h1:RPNrshWIDI6G2gRW9EHilWtl7Z6Sb1BR0xunSBf0
 github.com/kylelemons/godebug v1.1.0/go.mod h1:9/0rRGxNHcop5bhtWyNeEfOS8JIWk580+fNqagV/RAw=
 github.com/lucasb-eyer/go-colorful v1.4.0 h1:UtrWVfLdarDgc44HcS7pYloGHJUjHV/4FwW4TvVgFr4=
 github.com/lucasb-eyer/go-colorful v1.4.0/go.mod h1:R4dSotOR9KMtayYi1e77YzuveK+i7ruzyGqttikkLy0=
-github.com/mark3labs/mcp-go v0.47.0 h1:h44yeM3DduDyQgzImYWu4pt6VRkqP/0p/95AGhWngnA=
-github.com/mark3labs/mcp-go v0.47.0/go.mod h1:JKTC7R2LLVagkEWK7Kwu7DbmA6iIvnNAod6yrHiQMag=
-github.com/mattn/go-isatty v0.0.20 h1:xfD0iDuEKnDkl03q4limB+vH+GxLEtL/jb4xVJSWWEY=
-github.com/mattn/go-isatty v0.0.20/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
-github.com/mattn/go-runewidth v0.0.22 h1:76lXsPn6FyHtTY+jt2fTTvsMUCZq1k0qwRsAMuxzKAk=
-github.com/mattn/go-runewidth v0.0.22/go.mod h1:XBkDxAl56ILZc9knddidhrOlY5R/pDhgLpndooCuJAs=
+github.com/mark3labs/mcp-go v0.47.1 h1:A9sJJ20mscl/ssLYHjodfaoBmq6uuhMG7pAPNYaQymQ=
+github.com/mark3labs/mcp-go v0.47.1/go.mod h1:JKTC7R2LLVagkEWK7Kwu7DbmA6iIvnNAod6yrHiQMag=
+github.com/mattn/go-isatty v0.0.21 h1:xYae+lCNBP7QuW4PUnNG61ffM4hVIfm+zUzDuSzYLGs=
+github.com/mattn/go-isatty v0.0.21/go.mod h1:ZXfXG4SQHsB/w3ZeOYbR0PrPwLy+n6xiMrJlRFqopa4=
+github.com/mattn/go-runewidth v0.0.23 h1:7ykA0T0jkPpzSvMS5i9uoNn2Xy3R383f9HDx3RybWcw=
+github.com/mattn/go-runewidth v0.0.23/go.mod h1:XBkDxAl56ILZc9knddidhrOlY5R/pDhgLpndooCuJAs=
 github.com/mitchellh/hashstructure/v2 v2.0.2 h1:vGKWl0YJqUNxE8d+h8f6NJLcCJrgbhC4NcD46KavDd4=
 github.com/mitchellh/hashstructure/v2 v2.0.2/go.mod h1:MG3aRVU/N29oo/V/IhBX8GR/zz4kQkprJgF2EVszyDE=
 github.com/muesli/cancelreader v0.2.2 h1:3I4Kt4BQjOR54NavqnDogx/MIoWBFa0StPA8ELUXHmA=
@@ -272,18 +272,18 @@ github.com/yuin/goldmark v1.8.2 h1:kEGpgqJXdgbkhcOgBxkC0X0PmoPG1ZyoZ117rDVp4zE=
 github.com/yuin/goldmark v1.8.2/go.mod h1:ip/1k0VRfGynBgxOz0yCqHrbZXhcjxyuS66Brc7iBKg=
 go.opentelemetry.io/auto/sdk v1.2.1 h1:jXsnJ4Lmnqd11kwkBV2LgLoFMZKizbCi5fNZ/ipaZ64=
 go.opentelemetry.io/auto/sdk v1.2.1/go.mod h1:KRTj+aOaElaLi+wW1kO/DZRXwkF4C5xPbEe3ZiIhN7Y=
-go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc v0.67.0 h1:yI1/OhfEPy7J9eoa6Sj051C7n5dvpj0QX8g4sRchg04=
-go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc v0.67.0/go.mod h1:NoUCKYWK+3ecatC4HjkRktREheMeEtrXoQxrqYFeHSc=
-go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.67.0 h1:OyrsyzuttWTSur2qN/Lm0m2a8yqyIjUVBZcxFPuXq2o=
-go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.67.0/go.mod h1:C2NGBr+kAB4bk3xtMXfZ94gqFDtg/GkI7e9zqGh5Beg=
+go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc v0.68.0 h1:0Qx7VGBacMm9ZENQ7TnNObTYI4ShC+lHI16seduaxZo=
+go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc v0.68.0/go.mod h1:Sje3i3MjSPKTSPvVWCaL8ugBzJwik3u4smCjUeuupqg=
+go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.68.0 h1:CqXxU8VOmDefoh0+ztfGaymYbhdB/tT3zs79QaZTNGY=
+go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.68.0/go.mod h1:BuhAPThV8PBHBvg8ZzZ/Ok3idOdhWIodywz2xEcRbJo=
 go.opentelemetry.io/otel v1.43.0 h1:mYIM03dnh5zfN7HautFE4ieIig9amkNANT+xcVxAj9I=
 go.opentelemetry.io/otel v1.43.0/go.mod h1:JuG+u74mvjvcm8vj8pI5XiHy1zDeoCS2LB1spIq7Ay0=
 go.opentelemetry.io/otel/metric v1.43.0 h1:d7638QeInOnuwOONPp4JAOGfbCEpYb+K6DVWvdxGzgM=
 go.opentelemetry.io/otel/metric v1.43.0/go.mod h1:RDnPtIxvqlgO8GRW18W6Z/4P462ldprJtfxHxyKd2PY=
-go.opentelemetry.io/otel/sdk v1.42.0 h1:LyC8+jqk6UJwdrI/8VydAq/hvkFKNHZVIWuslJXYsDo=
-go.opentelemetry.io/otel/sdk v1.42.0/go.mod h1:rGHCAxd9DAph0joO4W6OPwxjNTYWghRWmkHuGbayMts=
-go.opentelemetry.io/otel/sdk/metric v1.42.0 h1:D/1QR46Clz6ajyZ3G8SgNlTJKBdGp84q9RKCAZ3YGuA=
-go.opentelemetry.io/otel/sdk/metric v1.42.0/go.mod h1:Ua6AAlDKdZ7tdvaQKfSmnFTdHx37+J4ba8MwVCYM5hc=
+go.opentelemetry.io/otel/sdk v1.43.0 h1:pi5mE86i5rTeLXqoF/hhiBtUNcrAGHLKQdhg4h4V9Dg=
+go.opentelemetry.io/otel/sdk v1.43.0/go.mod h1:P+IkVU3iWukmiit/Yf9AWvpyRDlUeBaRg6Y+C58QHzg=
+go.opentelemetry.io/otel/sdk/metric v1.43.0 h1:S88dyqXjJkuBNLeMcVPRFXpRw2fuwdvfCGLEo89fDkw=
+go.opentelemetry.io/otel/sdk/metric v1.43.0/go.mod h1:C/RJtwSEJ5hzTiUz5pXF1kILHStzb9zFlIEe85bhj6A=
 go.opentelemetry.io/otel/trace v1.43.0 h1:BkNrHpup+4k4w+ZZ86CZoHHEkohws8AY+WTX09nk+3A=
 go.opentelemetry.io/otel/trace v1.43.0/go.mod h1:/QJhyVBUUswCphDVxq+8mld+AvhXZLhe+8WVFxiFff0=
 go.yaml.in/yaml/v3 v3.0.4 h1:tfq32ie2Jv2UxXFdLJdh3jXuOzWiL1fo0bu/FbuKpbc=
@@ -298,9 +298,8 @@ golang.org/x/oauth2 v0.36.0 h1:peZ/1z27fi9hUOFCAZaHyrpWG5lwe0RJEEEeH0ThlIs=
 golang.org/x/oauth2 v0.36.0/go.mod h1:YDBUJMTkDnJS+A4BP4eZBjCqtokkg1hODuPjwiGPO7Q=
 golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
 golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
-golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.42.0 h1:omrd2nAlyT5ESRdCLYdm3+fMfNFE/+Rf4bDIQImRJeo=
-golang.org/x/sys v0.42.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/sys v0.43.0 h1:Rlag2XtaFTxp19wS8MXlJwTvoh8ArU6ezoyFsMyCTNI=
+golang.org/x/sys v0.43.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
 golang.org/x/term v0.41.0 h1:QCgPso/Q3RTJx2Th4bDLqML4W6iJiaXFq2/ftQF13YU=
 golang.org/x/term v0.41.0/go.mod h1:3pfBgksrReYfZ5lvYM0kSO0LIkAl4Yl2bXOkKP7Ec2A=
 golang.org/x/text v0.35.0 h1:JOVx6vVDFokkpaq1AEptVzLTpDe9KGpj5tR4/X+ybL8=
@@ -309,16 +308,16 @@ golang.org/x/time v0.15.0 h1:bbrp8t3bGUeFOx08pvsMYRTCVSMk89u4tKbNOZbp88U=
 golang.org/x/time v0.15.0/go.mod h1:Y4YMaQmXwGQZoFaVFk4YpCt4FLQMYKZe9oeV/f4MSno=
 gonum.org/v1/gonum v0.17.0 h1:VbpOemQlsSMrYmn7T2OUvQ4dqxQXU+ouZFQsZOx50z4=
 gonum.org/v1/gonum v0.17.0/go.mod h1:El3tOrEuMpv2UdMrbNlKEh9vd86bmQ6vqIcDwxEOc1E=
-google.golang.org/api v0.274.0 h1:aYhycS5QQCwxHLwfEHRRLf9yNsfvp1JadKKWBE54RFA=
-google.golang.org/api v0.274.0/go.mod h1:JbAt7mF+XVmWu6xNP8/+CTiGH30ofmCmk9nM8d8fHew=
+google.golang.org/api v0.275.0 h1:vfY5d9vFVJeWEZT65QDd9hbndr7FyZ2+6mIzGAh71NI=
+google.golang.org/api v0.275.0/go.mod h1:Fnag/EWUPIcJXuIkP1pjoTgS5vdxlk3eeemL7Do6bvw=
 google.golang.org/genai v1.52.1 h1:dYoljKtLDXMiBdVaClSJ/ZPwZ7j1N0lGjMhwOKOQUlk=
 google.golang.org/genai v1.52.1/go.mod h1:A3kkl0nyBjyFlNjgxIwKq70julKbIxpSxqKO5gw/gmk=
 google.golang.org/genproto v0.0.0-20260319201613-d00831a3d3e7 h1:XzmzkmB14QhVhgnawEVsOn6OFsnpyxNPRY9QV01dNB0=
 google.golang.org/genproto v0.0.0-20260319201613-d00831a3d3e7/go.mod h1:L43LFes82YgSonw6iTXTxXUX1OlULt4AQtkik4ULL/I=
 google.golang.org/genproto/googleapis/api v0.0.0-20260319201613-d00831a3d3e7 h1:41r6JMbpzBMen0R/4TZeeAmGXSJC7DftGINUodzTkPI=
 google.golang.org/genproto/googleapis/api v0.0.0-20260319201613-d00831a3d3e7/go.mod h1:EIQZ5bFCfRQDV4MhRle7+OgjNtZ6P1PiZBgAKuxXu/Y=
-google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9 h1:m8qni9SQFH0tJc1X0vmnpw/0t+AImlSvp30sEupozUg=
-google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9/go.mod h1:4Hqkh8ycfw05ld/3BWL7rJOSfebL2Q+DVDeRgYgxUU8=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260406210006-6f92a3bedf2d h1:wT2n40TBqFY6wiwazVK9/iTWbsQrgk5ZfCSVFLO9LQA=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260406210006-6f92a3bedf2d/go.mod h1:4Hqkh8ycfw05ld/3BWL7rJOSfebL2Q+DVDeRgYgxUU8=
 google.golang.org/grpc v1.80.0 h1:Xr6m2WmWZLETvUNvIUmeD5OAagMw3FiKmMlTdViWsHM=
 google.golang.org/grpc v1.80.0/go.mod h1:ho/dLnxwi3EDJA4Zghp7k2Ec1+c2jqup0bFkw07bwF4=
 google.golang.org/protobuf v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE=

internal/acpserver/agent.go

@@ -23,18 +23,6 @@ import (
 // Version is injected at build time; fallback to "dev".
 var Version = "dev"
 
-// thinkingTagOpen and thinkingTagClose are the XML-style tags that some models
-// (Qwen, DeepSeek) wrap reasoning content in. We parse these to extract
-// reasoning/thinking content and send it as ACP thought updates.
-// Also support <think> format used by some models.
-const (
-	thinkingTagOpen    = "<thinking>"
-	thinkingTagClose   = "</thinking>"
-	shortThinkTagOpen  = "<think>"
-	shortThinkTagClose = "</think>"
-)
-
-// Agent implements the acp.Agent interface, delegating to Kit for LLM
 // execution, tool calls, and session management.
 type Agent struct {
 	conn     *acp.AgentSideConnection
@@ -42,10 +30,6 @@ type Agent struct {
 
 	// toolCallCounter provides unique IDs for tool calls within a turn.
 	toolCallCounter atomic.Int64
-
-	// inThinkingTag tracks whether we're currently inside a <thinking> tag
-	// when parsing streaming content from models that wrap reasoning in XML tags.
-	inThinkingTag bool
 }
 
 // NewAgent creates a new ACP agent backed by Kit.
@@ -144,9 +128,6 @@ func (a *Agent) Prompt(ctx context.Context, params acp.PromptRequest) (acp.Promp
 
 	log.Debug("acp: prompt", "session", sessionID, "prompt_len", len(promptText), "files", len(files))
 
-	// Reset thinking tag state for this new prompt turn
-	a.inThinkingTag = false
-
 	// Create a cancellable context for this prompt turn.
 	promptCtx, cancel := context.WithCancel(ctx)
 	sess.setCancel(cancel)
@@ -230,24 +211,8 @@ func (a *Agent) subscribeEvents(ctx context.Context, k *kit.Kit, sessionID acp.S
 		var update *acp.SessionUpdate
 		switch ev := e.(type) {
 		case kit.MessageUpdateEvent:
-			// Handle models that wrap reasoning in <thinking> tags (Qwen, DeepSeek)
-			// Parse the chunk and separate reasoning from regular text
-			reasoning, text := a.parseThinkingTags(ev.Chunk)
-
-			// Send reasoning update if we have reasoning content
-			if reasoning != "" {
-				u := acp.UpdateAgentThoughtText(reasoning)
-				_ = a.conn.SessionUpdate(ctx, acp.SessionNotification{
-					SessionId: sessionID,
-					Update:    u,
-				})
-			}
-
-			// Send text update if we have text content
-			if text != "" {
-				u := acp.UpdateAgentMessageText(text)
-				update = &u
-			}
+			u := acp.UpdateAgentMessageText(ev.Chunk)
+			update = &u
 
 		case kit.ReasoningDeltaEvent:
 			u := acp.UpdateAgentThoughtText(ev.Delta)
@@ -430,81 +395,6 @@ func extractPromptContent(blocks []acp.ContentBlock) (string, []kit.LLMFilePart)
 	return strings.Join(textParts, "\n"), files
 }
 
-// parseThinkingTags parses a text chunk for <thinking> or  tags and separates
-// reasoning content from regular text. This handles models (Qwen, DeepSeek)
-// that wrap reasoning in XML-style tags instead of using proper reasoning events.
-// Returns (reasoningContent, textContent).
-func (a *Agent) parseThinkingTags(chunk string) (reasoning string, text string) {
-	// Handle empty chunk
-	if chunk == "" {
-		return "", ""
-	}
-
-	// Determine which tag format to use (long or short)
-	openTag := thinkingTagOpen
-	closeTag := thinkingTagClose
-
-	if strings.Contains(chunk, shortThinkTagOpen) || strings.Contains(chunk, shortThinkTagClose) {
-		openTag = shortThinkTagOpen
-		closeTag = shortThinkTagClose
-	} else if !strings.Contains(chunk, thinkingTagOpen) && !strings.Contains(chunk, thinkingTagClose) && !a.inThinkingTag {
-		// No tags at all and not in thinking mode - return as text
-		return "", chunk
-	}
-
-	// Check for opening tag
-	if strings.Contains(chunk, openTag) {
-		parts := strings.SplitN(chunk, openTag, 2)
-
-		// Content before the opening tag is regular text
-		if !a.inThinkingTag && parts[0] != "" {
-			text = parts[0]
-		}
-
-		a.inThinkingTag = true
-
-		// Content after the opening tag is reasoning
-		if len(parts) > 1 {
-			// Check if the same chunk contains the closing tag
-			if strings.Contains(parts[1], closeTag) {
-				innerParts := strings.SplitN(parts[1], closeTag, 2)
-				reasoning = innerParts[0]
-				a.inThinkingTag = false
-
-				// Content after closing tag is regular text
-				if len(innerParts) > 1 && innerParts[1] != "" {
-					text += innerParts[1]
-				}
-			} else if parts[1] != "" {
-				// No closing tag yet, all remaining content is reasoning
-				reasoning = parts[1]
-			}
-		}
-		return reasoning, text
-	}
-
-	// Check for closing tag
-	if strings.Contains(chunk, closeTag) {
-		parts := strings.SplitN(chunk, closeTag, 2)
-		a.inThinkingTag = false
-
-		// Content before closing tag is reasoning
-		reasoning = parts[0]
-
-		// Content after closing tag is regular text
-		if len(parts) > 1 && parts[1] != "" {
-			text = parts[1]
-		}
-		return reasoning, text
-	}
-
-	// No tags found - content goes to current mode
-	if a.inThinkingTag {
-		return chunk, ""
-	}
-	return "", chunk
-}
-
 // isTextMimeType returns true if the MIME type indicates text content.
 func isTextMimeType(mimeType string) bool {
 	return strings.HasPrefix(mimeType, "text/") ||

internal/agent/agent.go

@@ -89,6 +89,14 @@ type ReasoningCompleteHandler func()
 // Note: This is an alias for core.ToolOutputCallback to avoid import cycles.
 type ToolOutputHandler = core.ToolOutputCallback
 
+// StepMessagesHandler is a function type for persisting messages after each
+// complete step in a multi-step agent turn. The handler receives the messages
+// produced by the step (typically an assistant message with tool calls followed
+// by a tool-role message with results, or a final assistant message with text).
+// This enables incremental session persistence so that progress is saved as
+// it happens rather than only at the end of the turn.
+type StepMessagesHandler func(stepMessages []fantasy.Message)
+
 // StepUsageHandler is a function type for handling token usage after each
 // complete step in a multi-step agent turn. This enables real-time cost
 // tracking during long-running tool-calling conversations.
@@ -141,6 +149,11 @@ type GenerateWithLoopResult struct {
 	TotalUsage fantasy.Usage
 	// StopReason is the LLM provider's finish reason for the final response.
 	StopReason string
+	// PersistedMessageCount is the number of new messages (beyond the original
+	// input) that were already persisted incrementally via OnStepMessages during
+	// generation. The caller should skip these when doing post-generation
+	// persistence to avoid duplicates.
+	PersistedMessageCount int
 }
 
 // NewAgent creates a new Agent with core tools and optional MCP tool integration.
@@ -377,7 +390,7 @@ func (a *Agent) GenerateWithLoop(ctx context.Context, messages []fantasy.Message
 	onResponse ResponseHandler, onToolCallContent ToolCallContentHandler,
 ) (*GenerateWithLoopResult, error) {
 	return a.GenerateWithLoopAndStreaming(ctx, messages, onToolCall, onToolExecution, onToolResult,
-		onResponse, onToolCallContent, nil, nil, nil, nil, nil)
+		onResponse, onToolCallContent, nil, nil, nil, nil, nil, nil)
 }
 
 // GenerateWithLoopAndStreaming processes messages using the agent with streaming and callbacks.
@@ -390,6 +403,7 @@ func (a *Agent) GenerateWithLoopAndStreaming(ctx context.Context, messages []fan
 	onReasoningDelta ReasoningDeltaHandler,
 	onReasoningComplete ReasoningCompleteHandler,
 	onToolOutput ToolOutputHandler,
+	onStepMessages StepMessagesHandler,
 	onStepUsage StepUsageHandler,
 ) (*GenerateWithLoopResult, error) {
 
@@ -429,6 +443,10 @@ func (a *Agent) GenerateWithLoopAndStreaming(ctx context.Context, messages []fan
 		// when it returns an error, but the OnStepFinish callback fires
 		// for every step that completed before the error occurred.
 		var completedStepMessages []fantasy.Message
+		// persistedCount tracks how many new messages (beyond the original
+		// input) were persisted incrementally via onStepMessages, so the
+		// caller can skip them during post-generation persistence.
+		var persistedCount int
 
 		// Use the streaming agent
 		streamCall := fantasy.AgentStreamCall{
@@ -514,6 +532,13 @@ func (a *Agent) GenerateWithLoopAndStreaming(ctx context.Context, messages []fan
 				// persisted even if a later step is cancelled.
 				completedStepMessages = append(completedStepMessages, step.Messages...)
 
+				// Persist step messages incrementally so progress is saved
+				// as it happens rather than only at the end of the turn.
+				if onStepMessages != nil && len(step.Messages) > 0 {
+					onStepMessages(step.Messages)
+					persistedCount += len(step.Messages)
+				}
+
 				if ctx.Err() != nil {
 					return ctx.Err()
 				}
@@ -592,7 +617,8 @@ func (a *Agent) GenerateWithLoopAndStreaming(ctx context.Context, messages []fan
 				partialMessages = append(partialMessages, messages...)
 				partialMessages = append(partialMessages, completedStepMessages...)
 				return &GenerateWithLoopResult{
-					ConversationMessages: partialMessages,
+					ConversationMessages:  partialMessages,
+					PersistedMessageCount: persistedCount,
 				}, err
 			}
 			return nil, err
@@ -607,7 +633,9 @@ func (a *Agent) GenerateWithLoopAndStreaming(ctx context.Context, messages []fan
 			onResponse(result.Response.Content.Text())
 		}
 
-		return convertAgentResult(result, messages), nil
+		r := convertAgentResult(result, messages)
+		r.PersistedMessageCount = persistedCount
+		return r, nil
 	}
 
 	// Non-streaming path with no callbacks — use the simpler Generate call.

internal/extensions/loader.go

@@ -2,11 +2,11 @@ package extensions
 
 import (
 	"fmt"
-	"log"
 	"os"
 	"path/filepath"
 	"strings"
 
+	"github.com/charmbracelet/log"
 	"github.com/traefik/yaegi/interp"
 	"github.com/traefik/yaegi/stdlib"
 	"github.com/traefik/yaegi/stdlib/unrestricted"
@@ -34,11 +34,10 @@ func LoadExtensions(extraPaths []string) ([]LoadedExtension, error) {
 	for _, p := range paths {
 		ext, err := loadSingleExtension(p)
 		if err != nil {
-			log.Printf("WARN skipping extension: path=%s err=%v", p, err)
 			continue
 		}
 		loaded = append(loaded, *ext)
-		log.Printf("DEBUG loaded extension: path=%s handlers=%d tools=%d commands=%d tool_renderers=%d", p, countHandlers(ext), len(ext.Tools), len(ext.Commands), len(ext.ToolRenderers))
+		log.Debug("loaded extension", "path", p, "handlers", countHandlers(ext), "tools", len(ext.Tools), "commands", len(ext.Commands), "tool_renderers", len(ext.ToolRenderers))
 	}
 	return loaded, nil
 }

internal/session/compaction_test.go

@@ -0,0 +1,317 @@
+package session
+
+import (
+	"slices"
+	"testing"
+
+	"charm.land/fantasy"
+	"github.com/mark3labs/kit/internal/message"
+)
+
+// TestCompactionCreatesNewLeaf verifies that after compaction, the compaction
+// entry has no parent (creating a new root), and BuildContext returns only
+// the summary and kept messages, not the old compacted messages.
+func TestCompactionCreatesNewLeaf(t *testing.T) {
+	tm := InMemoryTreeSession("/test")
+
+	// Add some messages: M1, M2 (old, will be compacted), M3, M4 (kept)
+	msg1 := message.Message{Role: message.RoleUser, Parts: []message.ContentPart{message.TextContent{Text: "Message 1 - old"}}}
+	msg2 := message.Message{Role: message.RoleAssistant, Parts: []message.ContentPart{message.TextContent{Text: "Message 2 - old"}}}
+	msg3 := message.Message{Role: message.RoleUser, Parts: []message.ContentPart{message.TextContent{Text: "Message 3 - kept"}}}
+	msg4 := message.Message{Role: message.RoleAssistant, Parts: []message.ContentPart{message.TextContent{Text: "Message 4 - kept"}}}
+
+	_, _ = tm.AppendMessage(msg1)
+	_, _ = tm.AppendMessage(msg2)
+	id3, _ := tm.AppendMessage(msg3)
+	id4, _ := tm.AppendMessage(msg4)
+
+	// Verify initial state - all messages should be in context
+	messages, _, _ := tm.BuildContext()
+	if len(messages) != 4 {
+		t.Fatalf("expected 4 messages before compaction, got %d", len(messages))
+	}
+
+	// Verify entry IDs
+	entryIDs := tm.GetContextEntryIDs()
+	if len(entryIDs) != 4 {
+		t.Fatalf("expected 4 entry IDs before compaction, got %d", len(entryIDs))
+	}
+
+	// Now add a compaction entry, simulating that M3 is the first kept entry
+	summary := "Summary of old messages"
+	compactionID, err := tm.AppendCompaction(summary, id3, 1000, 500, 2, []string{}, []string{})
+	if err != nil {
+		t.Fatalf("failed to append compaction: %v", err)
+	}
+
+	// Verify the compaction entry has no parent (empty ParentID)
+	compactionEntry := tm.GetEntry(compactionID).(*CompactionEntry)
+	if compactionEntry.ParentID != "" {
+		t.Errorf("compaction entry should have no parent, got %q", compactionEntry.ParentID)
+	}
+
+	// Verify the leaf is now the compaction entry
+	if tm.GetLeafID() != compactionID {
+		t.Errorf("leaf should be compaction entry %q, got %q", compactionID, tm.GetLeafID())
+	}
+
+	// Now BuildContext should return: [summary] + [M3, M4]
+	messages, _, _ = tm.BuildContext()
+	if len(messages) != 3 {
+		t.Fatalf("expected 3 messages after compaction (summary + 2 kept), got %d", len(messages))
+	}
+
+	// First message should be the summary
+	if messages[0].Role != fantasy.MessageRoleSystem {
+		t.Errorf("first message should be system summary, got %s", messages[0].Role)
+	}
+	summaryText := messages[0].Content[0].(fantasy.TextPart).Text
+	if summaryText != "[Conversation summary — earlier messages were compacted]\n\n"+summary {
+		t.Errorf("unexpected summary text: %s", summaryText)
+	}
+
+	// Second message should be M3 (kept)
+	if messages[1].Role != fantasy.MessageRoleUser {
+		t.Errorf("second message should be user (M3), got %s", messages[1].Role)
+	}
+	m3Text := messages[1].Content[0].(fantasy.TextPart).Text
+	if m3Text != "Message 3 - kept" {
+		t.Errorf("unexpected M3 text: %s", m3Text)
+	}
+
+	// Third message should be M4 (kept)
+	if messages[2].Role != fantasy.MessageRoleAssistant {
+		t.Errorf("third message should be assistant (M4), got %s", messages[2].Role)
+	}
+	m4Text := messages[2].Content[0].(fantasy.TextPart).Text
+	if m4Text != "Message 4 - kept" {
+		t.Errorf("unexpected M4 text: %s", m4Text)
+	}
+
+	// Verify GetContextEntryIDs returns correct IDs
+	entryIDs = tm.GetContextEntryIDs()
+	if len(entryIDs) != 3 {
+		t.Fatalf("expected 3 entry IDs after compaction (empty for summary + 2 kept), got %d: %v", len(entryIDs), entryIDs)
+	}
+
+	// First entry ID should be empty (summary has no entry)
+	if entryIDs[0] != "" {
+		t.Errorf("first entry ID should be empty (summary), got %q", entryIDs[0])
+	}
+
+	// Second and third should be id3 and id4 (the kept messages)
+	if entryIDs[1] != id3 {
+		t.Errorf("second entry ID should be %q (M3), got %q", id3, entryIDs[1])
+	}
+	if entryIDs[2] != id4 {
+		t.Errorf("third entry ID should be %q (M4), got %q", id4, entryIDs[2])
+	}
+}
+
+// TestCompactionWithNewMessagesAfterCompaction verifies that messages appended
+// after compaction are correctly included in the context.
+func TestCompactionWithNewMessagesAfterCompaction(t *testing.T) {
+	tm := InMemoryTreeSession("/test")
+
+	// Add initial messages
+	msg1 := message.Message{Role: message.RoleUser, Parts: []message.ContentPart{message.TextContent{Text: "Message 1"}}}
+	msg2 := message.Message{Role: message.RoleAssistant, Parts: []message.ContentPart{message.TextContent{Text: "Message 2"}}}
+	msg3 := message.Message{Role: message.RoleUser, Parts: []message.ContentPart{message.TextContent{Text: "Message 3 - kept"}}}
+
+	_, _ = tm.AppendMessage(msg1)
+	_, _ = tm.AppendMessage(msg2)
+	id3, _ := tm.AppendMessage(msg3)
+
+	// Compact, keeping only M3
+	_, _ = tm.AppendCompaction("Summary", id3, 1000, 500, 2, []string{}, []string{})
+
+	// Add a new message after compaction
+	msg4 := message.Message{Role: message.RoleAssistant, Parts: []message.ContentPart{message.TextContent{Text: "Message 4 - after compaction"}}}
+	_, _ = tm.AppendMessage(msg4)
+
+	// BuildContext should return: [summary] + [M4 (new after compaction)] + [M3 (kept)]
+	messages, _, _ := tm.BuildContext()
+	if len(messages) != 3 {
+		t.Fatalf("expected 3 messages (summary + M4 + M3), got %d: %+v", len(messages), messages)
+	}
+
+	// Verify order: summary, M4 (new), M3 (kept)
+	if messages[0].Role != fantasy.MessageRoleSystem {
+		t.Errorf("first message should be summary, got %s", messages[0].Role)
+	}
+	if messages[1].Role != fantasy.MessageRoleAssistant {
+		t.Errorf("second message should be assistant (M4), got %s", messages[1].Role)
+	}
+	m4Text := messages[1].Content[0].(fantasy.TextPart).Text
+	if m4Text != "Message 4 - after compaction" {
+		t.Errorf("unexpected M4 text: %s", m4Text)
+	}
+	if messages[2].Role != fantasy.MessageRoleUser {
+		t.Errorf("third message should be user (M3), got %s", messages[2].Role)
+	}
+
+	// Verify that M1 is NOT in the context
+	for i, msg := range messages {
+		if msg.Role == fantasy.MessageRoleUser {
+			text := msg.Content[0].(fantasy.TextPart).Text
+			if text == "Message 1" {
+				t.Errorf("Message 1 (compacted) should not be in context at index %d", i)
+			}
+		}
+	}
+}
+
+// TestCompactionWithNoKeptMessages verifies compaction when all messages are compacted.
+func TestCompactionWithNoKeptMessages(t *testing.T) {
+	tm := InMemoryTreeSession("/test")
+
+	// Add messages that will all be compacted
+	msg1 := message.Message{Role: message.RoleUser, Parts: []message.ContentPart{message.TextContent{Text: "Message 1"}}}
+	msg2 := message.Message{Role: message.RoleAssistant, Parts: []message.ContentPart{message.TextContent{Text: "Message 2"}}}
+
+	if _, err := tm.AppendMessage(msg1); err != nil {
+		t.Fatalf("failed to append message: %v", err)
+	}
+	if _, err := tm.AppendMessage(msg2); err != nil {
+		t.Fatalf("failed to append message: %v", err)
+	}
+
+	// Compact with no kept messages (empty firstKeptEntryID)
+	summary := "All messages summarized"
+	compactionID, _ := tm.AppendCompaction(summary, "", 1000, 100, 2, []string{}, []string{})
+
+	// Verify the compaction entry has no parent
+	compactionEntry := tm.GetEntry(compactionID).(*CompactionEntry)
+	if compactionEntry.ParentID != "" {
+		t.Errorf("compaction entry should have no parent, got %q", compactionEntry.ParentID)
+	}
+
+	// BuildContext should return only the summary
+	messages, _, _ := tm.BuildContext()
+	if len(messages) != 1 {
+		t.Fatalf("expected 1 message (summary only), got %d: %+v", len(messages), messages)
+	}
+	if messages[0].Role != fantasy.MessageRoleSystem {
+		t.Errorf("message should be system summary, got %s", messages[0].Role)
+	}
+}
+
+// TestMultipleCompactions verifies that multiple compactions work correctly.
+func TestMultipleCompactions(t *testing.T) {
+	tm := InMemoryTreeSession("/test")
+
+	// First batch of messages
+	msg1 := message.Message{Role: message.RoleUser, Parts: []message.ContentPart{message.TextContent{Text: "Batch 1 - User"}}}
+	msg2 := message.Message{Role: message.RoleAssistant, Parts: []message.ContentPart{message.TextContent{Text: "Batch 1 - Assistant"}}}
+	id1, _ := tm.AppendMessage(msg1)
+	id2, _ := tm.AppendMessage(msg2)
+
+	// First compaction
+	_, _ = tm.AppendCompaction("Summary 1", id1, 1000, 500, 1, []string{}, []string{})
+
+	// Second batch
+	msg3 := message.Message{Role: message.RoleUser, Parts: []message.ContentPart{message.TextContent{Text: "Batch 2 - User"}}}
+	msg4 := message.Message{Role: message.RoleAssistant, Parts: []message.ContentPart{message.TextContent{Text: "Batch 2 - Assistant"}}}
+	id3, _ := tm.AppendMessage(msg3)
+	id4, _ := tm.AppendMessage(msg4)
+
+	// Second compaction (compacting the first compaction + batch 2)
+	// Note: id3 is the first kept entry, so id3 and id4 should be preserved
+	compactionID2, _ := tm.AppendCompaction("Summary 2", id3, 1000, 500, 3, []string{}, []string{})
+
+	// Verify second compaction has no parent
+	compactionEntry2 := tm.GetEntry(compactionID2).(*CompactionEntry)
+	if compactionEntry2.ParentID != "" {
+		t.Errorf("second compaction entry should have no parent, got %q", compactionEntry2.ParentID)
+	}
+
+	// Add final message
+	msg5 := message.Message{Role: message.RoleUser, Parts: []message.ContentPart{message.TextContent{Text: "Final message"}}}
+	id5, _ := tm.AppendMessage(msg5)
+
+	// BuildContext should include:
+	// - Summary 2 (from second compaction)
+	// - msg5 (final message)
+	// - msg3, msg4 (kept from second compaction)
+	// But NOT Summary 1 or msg1, msg2 (they're before the first kept entry of compaction 2)
+	messages, _, _ := tm.BuildContext()
+
+	// Should have: Summary 2 + msg5 + msg3 + msg4 = 4 messages
+	if len(messages) != 4 {
+		t.Fatalf("expected 4 messages (Summary 2 + msg5 + msg3 + msg4), got %d: %+v", len(messages), messages)
+	}
+
+	// First should be Summary 2
+	if messages[0].Role != fantasy.MessageRoleSystem {
+		t.Errorf("first message should be system (Summary 2), got %s", messages[0].Role)
+	}
+	summaryText := messages[0].Content[0].(fantasy.TextPart).Text
+	if summaryText != "[Conversation summary — earlier messages were compacted]\n\nSummary 2" {
+		t.Errorf("unexpected summary: %s", summaryText)
+	}
+
+	// Verify msg5 is included
+	foundFinal := false
+	for _, msg := range messages {
+		if msg.Role == fantasy.MessageRoleUser {
+			text := msg.Content[0].(fantasy.TextPart).Text
+			if text == "Final message" {
+				foundFinal = true
+				break
+			}
+		}
+	}
+	if !foundFinal {
+		t.Error("Final message (msg5) should be in context")
+	}
+
+	// Verify msg1, msg2 are NOT included (compacted by first compaction, then second)
+	for _, msg := range messages {
+		if msg.Role == fantasy.MessageRoleUser || msg.Role == fantasy.MessageRoleAssistant {
+			text := msg.Content[0].(fantasy.TextPart).Text
+			if text == "Batch 1 - User" || text == "Batch 1 - Assistant" {
+				t.Errorf("Batch 1 messages should not be in context, found: %s", text)
+			}
+		}
+	}
+
+	// Verify entry IDs
+	entryIDs := tm.GetContextEntryIDs()
+	if len(entryIDs) != 4 {
+		t.Fatalf("expected 4 entry IDs, got %d: %v", len(entryIDs), entryIDs)
+	}
+
+	// First should be empty (summary)
+	if entryIDs[0] != "" {
+		t.Errorf("first entry ID should be empty (summary), got %q", entryIDs[0])
+	}
+
+	// Check that id5 is in the list
+	if !slices.Contains(entryIDs, id5) {
+		t.Errorf("id5 (final message) should be in entry IDs, got %v", entryIDs)
+	}
+
+	// Verify id3 and id4 ARE in the list (they were kept)
+	foundID3, foundID4 := false, false
+	for _, id := range entryIDs {
+		if id == id3 {
+			foundID3 = true
+		}
+		if id == id4 {
+			foundID4 = true
+		}
+	}
+	if !foundID3 {
+		t.Errorf("id3 (kept message) should be in entry IDs, got %v", entryIDs)
+	}
+	if !foundID4 {
+		t.Errorf("id4 (kept message) should be in entry IDs, got %v", entryIDs)
+	}
+
+	// Verify id1 and id2 are NOT in the list (they were compacted away)
+	for _, id := range entryIDs {
+		if id == id1 || id == id2 {
+			t.Errorf("id1 or id2 (compacted) should not be in entry IDs, found %q in %v", id, entryIDs)
+		}
+	}
+}

internal/session/tree_manager.go

@@ -509,11 +509,19 @@ func (tm *TreeManager) AppendExtensionData(extType, data string) (string, error)
 // AppendCompaction adds a compaction entry to the tree. The entry records
 // the summary and the ID of the first entry that should be preserved in the
 // LLM context. Messages before that entry are replaced by the summary.
+//
+// The compaction entry becomes a new "root" for the post-compaction branch
+// with no parent (empty ParentID). This breaks the parent chain so that old
+// compacted messages are no longer traversed when building context. The kept
+// messages are explicitly collected via FirstKeptEntryID in BuildContext.
 func (tm *TreeManager) AppendCompaction(summary, firstKeptEntryID string, tokensBefore, tokensAfter, messagesRemoved int, readFiles, modifiedFiles []string) (string, error) {
 	tm.mu.Lock()
 	defer tm.mu.Unlock()
 
-	entry := NewCompactionEntry(tm.leafID, summary, firstKeptEntryID, tokensBefore, tokensAfter, messagesRemoved, readFiles, modifiedFiles)
+	// The compaction entry has no parent, making it a new "root" for the
+	// post-compaction branch. This ensures old compacted messages are not
+	// traversed when walking from the current leaf.
+	entry := NewCompactionEntry("", summary, firstKeptEntryID, tokensBefore, tokensAfter, messagesRemoved, readFiles, modifiedFiles)
 	if err := tm.appendAndPersist(entry); err != nil {
 		return "", err
 	}
@@ -683,14 +691,18 @@ func (tm *TreeManager) BuildContext() (messages []fantasy.Message, provider stri
 	// Find the last compaction entry on this branch — it determines
 	// which older messages are replaced by the summary.
 	var lastCompaction *CompactionEntry
+	var compactionIndex = -1
 	for i := len(branch) - 1; i >= 0; i-- {
 		if c, ok := branch[i].(*CompactionEntry); ok {
 			lastCompaction = c
+			compactionIndex = i
 			break
 		}
 	}
 
-	// If there is a compaction, inject the summary first.
+	// If there is a compaction, inject the summary first and collect
+	// the kept messages starting from FirstKeptEntryID (since the
+	// compaction entry's parent chain doesn't include them).
 	if lastCompaction != nil {
 		messages = append(messages, fantasy.Message{
 			Role: fantasy.MessageRoleSystem,
@@ -700,21 +712,104 @@ func (tm *TreeManager) BuildContext() (messages []fantasy.Message, provider stri
 				},
 			},
 		})
-	}
 
-	// Determine whether to skip entries (everything before firstKeptEntryID).
-	skipping := lastCompaction != nil
-	for _, entry := range branch {
-		// Once we reach the first kept entry, stop skipping.
-		if skipping {
-			entryID := tm.EntryID(entry)
-			if entryID == lastCompaction.FirstKeptEntryID {
-				skipping = false
-			} else {
+		// Collect entries from the compaction entry itself (at compactionIndex)
+		// and any entries before it in the branch (newer messages).
+		for i := compactionIndex; i < len(branch); i++ {
+			entry := branch[i]
+			switch e := entry.(type) {
+			case *MessageEntry:
+				msg, err := e.ToMessage()
+				if err != nil {
+					continue // skip malformed entries
+				}
+				msgs := msg.ToLLMMessages()
+				messages = append(messages, msgs...)
+
+			case *BranchSummaryEntry:
+				// Convert branch summary to a user message for context.
+				if e.Summary != "" {
+					messages = append(messages, fantasy.Message{
+						Role: fantasy.MessageRoleUser,
+						Content: []fantasy.MessagePart{
+							fantasy.TextPart{
+								Text: fmt.Sprintf("[Branch context: %s]", e.Summary),
+							},
+						},
+					})
+				}
+
+			case *ModelChangeEntry:
+				provider = e.Provider
+				modelID = e.ModelID
+
+			case *CompactionEntry:
+				// Already handled above (summary injected).
 				continue
 			}
 		}
 
+		// Now collect the kept messages starting from FirstKeptEntryID.
+		// These are not in the current branch because the compaction entry
+		// is parented to the first kept entry's parent, not the first kept entry.
+		// We iterate through entries in order (not using getBranchLocked) to avoid
+		// walking back to old compacted messages.
+		// We stop when we reach the compaction entry to avoid double-counting
+		// messages that were added after the compaction.
+		if lastCompaction.FirstKeptEntryID != "" {
+			found := false
+			for _, entry := range tm.entries {
+				entryID := tm.EntryID(entry)
+
+				// Skip entries until we reach the first kept entry.
+				if !found {
+					if entryID == lastCompaction.FirstKeptEntryID {
+						found = true
+					} else {
+						continue
+					}
+				}
+
+				// Stop when we reach the compaction entry itself.
+				// Messages after the compaction are collected from the branch walk above.
+				if entryID == lastCompaction.ID {
+					break
+				}
+
+				// Process this kept entry.
+				switch e := entry.(type) {
+				case *MessageEntry:
+					msg, err := e.ToMessage()
+					if err != nil {
+						continue
+					}
+					msgs := msg.ToLLMMessages()
+					messages = append(messages, msgs...)
+
+				case *BranchSummaryEntry:
+					if e.Summary != "" {
+						messages = append(messages, fantasy.Message{
+							Role: fantasy.MessageRoleUser,
+							Content: []fantasy.MessagePart{
+								fantasy.TextPart{
+									Text: fmt.Sprintf("[Branch context: %s]", e.Summary),
+								},
+							},
+						})
+					}
+
+				case *ModelChangeEntry:
+					provider = e.Provider
+					modelID = e.ModelID
+				}
+			}
+		}
+
+		return messages, provider, modelID
+	}
+
+	// No compaction - process the entire branch normally.
+	for _, entry := range branch {
 		switch e := entry.(type) {
 		case *MessageEntry:
 			msg, err := e.ToMessage()
@@ -740,10 +835,6 @@ func (tm *TreeManager) BuildContext() (messages []fantasy.Message, provider stri
 		case *ModelChangeEntry:
 			provider = e.Provider
 			modelID = e.ModelID
-
-		case *CompactionEntry:
-			// Already handled above (the last one on the branch).
-			continue
 		}
 	}
 
@@ -853,31 +944,92 @@ func (tm *TreeManager) GetContextEntryIDs() []string {
 
 	// Find the last compaction entry for skip logic.
 	var lastCompaction *CompactionEntry
+	var compactionIndex = -1
 	for i := len(branch) - 1; i >= 0; i-- {
 		if c, ok := branch[i].(*CompactionEntry); ok {
 			lastCompaction = c
+			compactionIndex = i
 			break
 		}
 	}
 
 	var ids []string
 
-	// If there's a compaction summary injected, it has no entry ID.
+	// If there's a compaction, we need to collect IDs from:
+	// 1. Entries after the compaction entry in the branch (newer messages)
+	// 2. Entries from FirstKeptEntryID onwards (kept messages)
 	if lastCompaction != nil {
-		ids = append(ids, "") // placeholder for the summary system message
-	}
+		// Placeholder for the summary system message (no entry ID).
+		ids = append(ids, "")
 
-	skipping := lastCompaction != nil
-	for _, entry := range branch {
-		if skipping {
-			entryID := tm.EntryID(entry)
-			if entryID == lastCompaction.FirstKeptEntryID {
-				skipping = false
-			} else {
-				continue
+		// Collect IDs from entries after the compaction entry (newer messages).
+		for i := compactionIndex + 1; i < len(branch); i++ {
+			entry := branch[i]
+			switch e := entry.(type) {
+			case *MessageEntry:
+				msg, err := e.ToMessage()
+				if err != nil {
+					continue
+				}
+				msgs := msg.ToLLMMessages()
+				for range msgs {
+					ids = append(ids, e.ID)
+				}
+
+			case *BranchSummaryEntry:
+				if e.Summary != "" {
+					ids = append(ids, e.ID)
+				}
 			}
 		}
 
+		// Collect IDs from the kept messages starting at FirstKeptEntryID.
+		// We iterate through entries in order (not using getBranchLocked) to avoid
+		// walking back to old compacted messages.
+		// We stop when we reach the compaction entry to avoid double-counting.
+		if lastCompaction.FirstKeptEntryID != "" {
+			found := false
+			for _, entry := range tm.entries {
+				entryID := tm.EntryID(entry)
+
+				// Skip entries until we reach the first kept entry.
+				if !found {
+					if entryID == lastCompaction.FirstKeptEntryID {
+						found = true
+					} else {
+						continue
+					}
+				}
+
+				// Stop when we reach the compaction entry itself.
+				if entryID == lastCompaction.ID {
+					break
+				}
+
+				switch e := entry.(type) {
+				case *MessageEntry:
+					msg, err := e.ToMessage()
+					if err != nil {
+						continue
+					}
+					msgs := msg.ToLLMMessages()
+					for range msgs {
+						ids = append(ids, e.ID)
+					}
+
+				case *BranchSummaryEntry:
+					if e.Summary != "" {
+						ids = append(ids, e.ID)
+					}
+				}
+			}
+		}
+
+		return ids
+	}
+
+	// No compaction - collect IDs from the entire branch.
+	for _, entry := range branch {
 		switch e := entry.(type) {
 		case *MessageEntry:
 			msg, err := e.ToMessage()
@@ -893,9 +1045,6 @@ func (tm *TreeManager) GetContextEntryIDs() []string {
 			if e.Summary != "" {
 				ids = append(ids, e.ID)
 			}
-
-		case *CompactionEntry:
-			continue
 		}
 	}
 

internal/ui/input.go

@@ -69,7 +69,7 @@ type InputComponent struct {
 	hideHint bool
 
 	// agentBusy indicates the agent is currently working. When true, the
-	// hint text shows steering shortcut (Ctrl+S) instead of submit.
+	// hint text shows steering shortcut (Ctrl+X s) instead of submit.
 	agentBusy bool
 
 	// pendingImages holds clipboard images attached to the next submission.
@@ -109,7 +109,7 @@ func NewInputComponent(width int, title string, appCtrl AppController) *InputCom
 	ta.Placeholder = "Type your message..."
 	ta.ShowLineNumbers = false
 	ta.Prompt = ""
-	ta.CharLimit = 5000
+	ta.CharLimit = 0
 	ta.SetWidth(width - 8) // Account for container padding, border and internal padding
 	ta.SetHeight(3)        // Default to 3 lines like huh
 	ta.Focus()
@@ -514,12 +514,12 @@ func (s *InputComponent) View() tea.View {
 		availableHintWidth := s.width - 3
 		if s.agentBusy {
 			// When the agent is working, show steering shortcut.
-			if availableHintWidth >= 55 {
-				hint = "enter queue • ctrl+s steer • esc esc cancel"
-			} else if availableHintWidth >= 35 {
-				hint = "↵ queue • ^S steer • esc×2 cancel"
+			if availableHintWidth >= 60 {
+				hint = "enter queue • ctrl+x s steer • esc esc cancel"
+			} else if availableHintWidth >= 40 {
+				hint = "↵ queue • ^X s steer • esc×2 cancel"
 			} else {
-				hint = "^S steer"
+				hint = "^X s steer"
 			}
 		} else if availableHintWidth >= 67 {
 			hint = "enter submit • ctrl+j / shift+enter new line • ctrl+v paste image"

internal/ui/messages.go

@@ -152,7 +152,7 @@ func (r *MessageRenderer) SetWidth(width int) {
 
 // RenderUserMessage renders a user's input message using herald Tip alert
 func (r *MessageRenderer) RenderUserMessage(content string, timestamp time.Time) UIMessage {
-	rendered := render.UserBlock(content, r.ty, style.GetTheme())
+	rendered := render.UserBlock(content, r.width, r.ty, style.GetTheme())
 
 	return UIMessage{
 		Type:      UserMessage,

internal/ui/model.go

@@ -477,7 +477,7 @@ type AppModel struct {
 	queuedMessages []string
 
 	// steeringMessages stores the text of prompts that were sent as steer
-	// messages (injected mid-turn via Ctrl+S). Rendered with a "STEERING"
+	// messages (injected mid-turn via Ctrl+X s). Rendered with a "STEERING"
 	// badge above the input. Cleared when the steer is consumed.
 	steeringMessages []string
 
@@ -498,6 +498,11 @@ type AppModel struct {
 	// A second ESC within 2 seconds will cancel the current step.
 	canceling bool
 
+	// leaderKeyActive tracks whether the Ctrl+X leader key prefix has been
+	// pressed. The next keypress is interpreted as a chord suffix (e.g. "s"
+	// for steer). Cleared on any subsequent keypress.
+	leaderKeyActive bool
+
 	// providerName is the LLM provider for the startup message.
 	providerName string
 
@@ -1268,6 +1273,71 @@ func (m *AppModel) Update(msg tea.Msg) (tea.Model, tea.Cmd) {
 			return m, tea.Batch(cmds...)
 		}
 
+		// ── Leader key chord handling (Ctrl+X prefix) ──────────────
+		// If the leader key was previously pressed, the current key
+		// completes the chord. We consume it regardless of match so
+		// the prefix doesn't leak to child components.
+		if m.leaderKeyActive {
+			m.leaderKeyActive = false
+			switch msg.String() {
+			case "s":
+				// Ctrl+X s → Steer: inject the current input as a steering
+				// message into the running agent turn.
+				if m.state == stateWorking && m.appCtrl != nil {
+					var text string
+					if ic, ok := m.input.(*InputComponent); ok {
+						text = strings.TrimSpace(ic.textarea.Value())
+					}
+					if text != "" {
+						// Clear the input, collect pending images, and push to history.
+						var images []uicore.ImageAttachment
+						if ic, ok := m.input.(*InputComponent); ok {
+							ic.pushHistory(text)
+							ic.textarea.SetValue("")
+							images = ic.ClearPendingImages()
+						}
+
+						// Preprocess @file references.
+						processedText := text
+						if m.cwd != "" {
+							processedText = fileutil.ProcessFileAttachments(text, m.cwd)
+						}
+
+						// Convert image attachments to kit.LLMFilePart for the app layer.
+						var fileParts []kit.LLMFilePart
+						for _, img := range images {
+							fileParts = append(fileParts, kit.LLMFilePart{
+								Data:      img.Data,
+								MediaType: img.MediaType,
+							})
+						}
+
+						// Build display text (include image count if any).
+						displayText := text
+						if len(images) > 0 {
+							displayText = fmt.Sprintf("%s\n[%d image(s) attached]", text, len(images))
+						}
+
+						// Inject the steer message.
+						sLen := m.appCtrl.SteerWithFiles(processedText, fileParts)
+						if sLen > 0 {
+							m.steeringMessages = append(m.steeringMessages, displayText)
+							m.layoutDirty = true
+						} else {
+							// Started immediately (agent was idle).
+							m.pendingUserPrints = append(m.pendingUserPrints, displayText)
+							m.flushStreamAndPendingUserMessages()
+							if m.state != stateWorking {
+								m.state = stateWorking
+							}
+						}
+					}
+				}
+			}
+			// Chord consumed — don't propagate to children.
+			return m, tea.Batch(cmds...)
+		}
+
 		switch msg.String() {
 		case "esc":
 			if m.state == stateWorking {
@@ -1286,61 +1356,10 @@ func (m *AppModel) Update(msg tea.Msg) (tea.Model, tea.Cmd) {
 			}
 			// In other states pass ESC through to children below.
 
-		case "ctrl+s":
-			// Steer: inject the current input as a steering message into the
-			// running agent turn. Only active during stateWorking — in input
-			// state, Ctrl+S is passed through to children (no-op by default).
-			if m.state == stateWorking && m.appCtrl != nil {
-				var text string
-				if ic, ok := m.input.(*InputComponent); ok {
-					text = strings.TrimSpace(ic.textarea.Value())
-				}
-				if text != "" {
-					// Clear the input, collect pending images, and push to history.
-					var images []uicore.ImageAttachment
-					if ic, ok := m.input.(*InputComponent); ok {
-						ic.pushHistory(text)
-						ic.textarea.SetValue("")
-						images = ic.ClearPendingImages()
-					}
-
-					// Preprocess @file references.
-					processedText := text
-					if m.cwd != "" {
-						processedText = fileutil.ProcessFileAttachments(text, m.cwd)
-					}
-
-					// Convert image attachments to kit.LLMFilePart for the app layer.
-					var fileParts []kit.LLMFilePart
-					for _, img := range images {
-						fileParts = append(fileParts, kit.LLMFilePart{
-							Data:      img.Data,
-							MediaType: img.MediaType,
-						})
-					}
-
-					// Build display text (include image count if any).
-					displayText := text
-					if len(images) > 0 {
-						displayText = fmt.Sprintf("%s\n[%d image(s) attached]", text, len(images))
-					}
-
-					// Inject the steer message.
-					sLen := m.appCtrl.SteerWithFiles(processedText, fileParts)
-					if sLen > 0 {
-						m.steeringMessages = append(m.steeringMessages, displayText)
-						m.layoutDirty = true
-					} else {
-						// Started immediately (agent was idle).
-						m.pendingUserPrints = append(m.pendingUserPrints, displayText)
-						m.flushStreamAndPendingUserMessages()
-						if m.state != stateWorking {
-							m.state = stateWorking
-						}
-					}
-				}
-				return m, tea.Batch(cmds...)
-			}
+		case "ctrl+x":
+			// Activate leader key prefix — the next keypress completes the chord.
+			m.leaderKeyActive = true
+			return m, tea.Batch(cmds...)
 		}
 
 		// Route key events to the focused child. Check for editor
@@ -2462,22 +2481,34 @@ func (m *AppModel) renderHeaderFooter(getter func() *WidgetData) string {
 	return renderContentBlock(data.Text, m.width, opts...)
 }
 
+// maxQueuedMessageLines is the maximum number of visible content lines
+// rendered for each queued or steering message block. Messages exceeding
+// this limit are truncated with an ellipsis to prevent large pastes from
+// overflowing the screen and squeezing the stream region to zero.
+const maxQueuedMessageLines = 3
+
 // renderQueuedMessages renders queued and steering prompts as styled content
 // blocks with badges, anchored between the separator and input. Steering
 // messages use a distinct "STEERING" badge to differentiate from queued ones.
+// Long messages are visually truncated to maxQueuedMessageLines.
 func (m *AppModel) renderQueuedMessages() string {
 	if len(m.queuedMessages) == 0 && len(m.steeringMessages) == 0 {
 		return ""
 	}
 	theme := style.GetTheme()
 
+	// Available content width inside the block: container minus border (1)
+	// minus left padding (2). Used to estimate line wrapping for truncation.
+	contentWidth := max(m.width-3, 10)
+
 	var blocks []string
 
 	// Render steering messages first (higher priority).
 	if len(m.steeringMessages) > 0 {
 		badge := style.CreateBadge("STEERING", theme.Warning)
 		for _, msg := range m.steeringMessages {
-			content := msg + "\n" + badge
+			display := truncateMessageForBlock(msg, maxQueuedMessageLines, contentWidth)
+			content := display + "\n" + badge
 			rendered := renderContentBlock(
 				content,
 				m.width,
@@ -2492,7 +2523,8 @@ func (m *AppModel) renderQueuedMessages() string {
 	if len(m.queuedMessages) > 0 {
 		badge := style.CreateBadge("QUEUED", theme.Accent)
 		for _, msg := range m.queuedMessages {
-			content := msg + "\n" + badge
+			display := truncateMessageForBlock(msg, maxQueuedMessageLines, contentWidth)
+			content := display + "\n" + badge
 			rendered := renderContentBlock(
 				content,
 				m.width,
@@ -2506,6 +2538,58 @@ func (m *AppModel) renderQueuedMessages() string {
 	return strings.Join(blocks, "\n")
 }
 
+// truncateMessageForBlock truncates a message to at most maxLines visible
+// lines, accounting for soft-wrapping at the given width. If the message is
+// truncated, the last visible line is replaced with an ellipsis ("…").
+func truncateMessageForBlock(msg string, maxLines, width int) string {
+	if width <= 0 {
+		width = 1
+	}
+
+	lines := strings.Split(msg, "\n")
+
+	// Count visible lines (each hard line may wrap into multiple visual lines).
+	var kept []string
+	visibleCount := 0
+	truncated := false
+
+	for _, line := range lines {
+		// Calculate how many visual lines this hard line occupies.
+		lineWidth := lipgloss.Width(line)
+		wrapped := 1
+		if lineWidth > width {
+			wrapped = (lineWidth + width - 1) / width // ceil division
+		}
+
+		if visibleCount+wrapped > maxLines {
+			// This line would exceed the limit. Keep a partial if we
+			// still have room for at least one more visual line.
+			remaining := maxLines - visibleCount
+			if remaining > 0 {
+				// Truncate the line to fit the remaining visual lines.
+				runes := []rune(line)
+				maxRunes := remaining * width
+				if maxRunes < len(runes) {
+					kept = append(kept, string(runes[:maxRunes]))
+				} else {
+					kept = append(kept, line)
+				}
+			}
+			truncated = true
+			break
+		}
+
+		kept = append(kept, line)
+		visibleCount += wrapped
+	}
+
+	if !truncated {
+		return msg
+	}
+
+	return strings.Join(kept, "\n") + "…"
+}
+
 // --------------------------------------------------------------------------
 // Print helpers — add content to ScrollList
 // --------------------------------------------------------------------------
@@ -2876,7 +2960,7 @@ func (m *AppModel) printHelpMessage() {
 		"**Keys:**\n" +
 		"- `Ctrl+C`: Exit at any time\n" +
 		"- `ESC` (x2): Cancel ongoing LLM generation\n" +
-		"- `Ctrl+S`: Steer — redirect the agent mid-turn (injected between tool calls)\n" +
+		"- `Ctrl+X s`: Steer — redirect the agent mid-turn (injected between tool calls)\n" +
 		"- `Enter` (while working): Queue message for after the agent finishes\n\n" +
 		"You can also just type your message to chat with the AI assistant."
 	m.printSystemMessage(help)

internal/ui/model_test.go

@@ -2,6 +2,7 @@ package ui
 
 import (
 	"errors"
+	"strings"
 	"testing"
 
 	tea "charm.land/bubbletea/v2"
@@ -892,3 +893,107 @@ func TestSubmit_duringWorking_stays(t *testing.T) {
 		t.Fatalf("expected Run('queued prompt') called, got %v", ctrl.runCalls)
 	}
 }
+
+// --------------------------------------------------------------------------
+// truncateMessageForBlock
+// --------------------------------------------------------------------------
+
+// TestTruncateMessageForBlock_shortMessage verifies that short messages are
+// returned unchanged.
+func TestTruncateMessageForBlock_shortMessage(t *testing.T) {
+	msg := "hello world"
+	got := truncateMessageForBlock(msg, 3, 80)
+	if got != msg {
+		t.Fatalf("expected unchanged message, got %q", got)
+	}
+}
+
+// TestTruncateMessageForBlock_exactLines verifies that a message with exactly
+// maxLines hard lines is returned unchanged.
+func TestTruncateMessageForBlock_exactLines(t *testing.T) {
+	msg := "line1\nline2\nline3"
+	got := truncateMessageForBlock(msg, 3, 80)
+	if got != msg {
+		t.Fatalf("expected unchanged message, got %q", got)
+	}
+}
+
+// TestTruncateMessageForBlock_tooManyLines verifies that messages exceeding
+// maxLines are truncated with an ellipsis.
+func TestTruncateMessageForBlock_tooManyLines(t *testing.T) {
+	msg := "line1\nline2\nline3\nline4\nline5"
+	got := truncateMessageForBlock(msg, 3, 80)
+	want := "line1\nline2\nline3…"
+	if got != want {
+		t.Fatalf("expected %q, got %q", want, got)
+	}
+}
+
+// TestTruncateMessageForBlock_longWrappingLine verifies that a single long
+// line that would wrap beyond maxLines is truncated.
+func TestTruncateMessageForBlock_longWrappingLine(t *testing.T) {
+	// 100 chars at width 20 = 5 visual lines, exceeds maxLines=3
+	msg := "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"
+	got := truncateMessageForBlock(msg, 3, 20)
+	// Should be truncated to 3*20=60 runes + "…"
+	if len([]rune(got)) != 61 { // 60 runes + "…"
+		t.Fatalf("expected 61 runes (60 + ellipsis), got %d runes: %q", len([]rune(got)), got)
+	}
+	if got[len(got)-3:] != "…" { // "…" is 3 bytes in UTF-8
+		t.Fatal("expected trailing ellipsis")
+	}
+}
+
+// TestTruncateMessageForBlock_emptyMessage verifies that empty messages are
+// returned unchanged.
+func TestTruncateMessageForBlock_emptyMessage(t *testing.T) {
+	got := truncateMessageForBlock("", 3, 80)
+	if got != "" {
+		t.Fatalf("expected empty string, got %q", got)
+	}
+}
+
+// TestTruncateMessageForBlock_mixedWrapAndHardLines verifies truncation when
+// some hard lines wrap and the total exceeds maxLines.
+func TestTruncateMessageForBlock_mixedWrapAndHardLines(t *testing.T) {
+	// First line: 40 chars at width 20 = 2 visual lines
+	// Second line: "short" = 1 visual line (total: 3, exactly at limit)
+	// Third line: would exceed
+	msg := "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\nshort\nextra"
+	got := truncateMessageForBlock(msg, 3, 20)
+	want := "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\nshort…"
+	if got != want {
+		t.Fatalf("expected %q, got %q", want, got)
+	}
+}
+
+// TestRenderQueuedMessages_truncatesLongMessages verifies that the rendered
+// queued message view truncates long messages instead of showing them in full.
+func TestRenderQueuedMessages_truncatesLongMessages(t *testing.T) {
+	ctrl := &stubAppController{}
+	m, _, _ := newTestAppModel(ctrl)
+	m.width = 80
+
+	// Queue a very long message (20 lines).
+	var b strings.Builder
+	for i := range 20 {
+		if i > 0 {
+			b.WriteByte('\n')
+		}
+		b.WriteString("This is a long line of text for testing purposes")
+	}
+	m.queuedMessages = []string{b.String()}
+
+	rendered := m.renderQueuedMessages()
+	if rendered == "" {
+		t.Fatal("expected non-empty rendered output")
+	}
+
+	// The full message would be ~20+ lines. With truncation to 3 content
+	// lines + badge + padding, it should be much shorter.
+	lines := len(strings.Split(rendered, "\n"))
+	// 3 content lines + 1 badge + 2 padding + border overhead ≈ ~7 lines max
+	if lines > 10 {
+		t.Fatalf("expected truncated output to be ≤10 lines, got %d lines", lines)
+	}
+}

internal/ui/prompt.go

@@ -78,7 +78,7 @@ func newInputPrompt(message, placeholder, defaultValue string, width, height int
 	ta.Placeholder = placeholder
 	ta.ShowLineNumbers = false
 	ta.Prompt = ""
-	ta.CharLimit = 1000
+	ta.CharLimit = 0
 	ta.SetWidth(width - 12) // account for border + padding
 	ta.SetHeight(1)
 	ta.Focus()

internal/ui/render/blocks.go

@@ -14,11 +14,19 @@ import (
 )
 
 // UserBlock renders a user message with herald Tip styling.
-func UserBlock(content string, ty *herald.Typography, theme style.Theme) string {
+// The width parameter controls line wrapping so long messages don't overflow.
+func UserBlock(content string, width int, ty *herald.Typography, theme style.Theme) string {
 	if strings.TrimSpace(content) == "" {
 		content = "(empty message)"
 	}
 
+	// Wrap content before passing to herald Alert so long lines break
+	// inside the alert box. Subtract 4 to account for the alert bar
+	// prefix ("│ ") and a small margin.
+	if width > 4 {
+		content = lipgloss.Wrap(content, width-4, "")
+	}
+
 	rendered := ty.Tip(content)
 	return styleMarginBottom(theme, rendered)
 }

internal/ui/style/styles.go

@@ -85,11 +85,13 @@ func GetMarkdownTypography() *herald.Typography {
 	return ty
 }
 
-// ToMarkdown renders markdown content using herald-md.
-// The width parameter is currently unused as herald handles wrapping
-// based on terminal width internally.
+// ToMarkdown renders markdown content using herald-md and wraps the result
+// to the given width so that long lines do not overflow the terminal.
 func ToMarkdown(content string, width int) string {
 	ty := GetMarkdownTypography()
 	rendered := heraldmd.Render(ty, []byte(content))
+	if width > 0 {
+		rendered = lipgloss.Wrap(rendered, width, "")
+	}
 	return rendered
 }

internal/ui/tool_approval_input.go

@@ -23,7 +23,7 @@ func NewToolApprovalInput(toolName, toolArgs string, width int) *ToolApprovalInp
 	ta := textarea.New()
 	ta.Placeholder = ""
 	ta.ShowLineNumbers = false
-	ta.CharLimit = 1000
+	ta.CharLimit = 0
 	ta.SetWidth(width - 8) // Account for container padding, border and internal padding
 	ta.SetHeight(4)        // Default to 3 lines like huh
 	ta.Focus()

pkg/kit/kit.go

@@ -1309,6 +1309,17 @@ func (m *Kit) generate(ctx context.Context, messages []fantasy.Message) (*agent.
 				IsStderr:   isStderr,
 			})
 		},
+		// Persist step messages incrementally so that progress survives
+		// crashes and long-running turns don't lose work. Each step's
+		// messages are persisted as a unit: for tool-calling steps this is
+		// the assistant message (with tool_use parts) + tool-role message
+		// (with tool_result parts) as a pair; for the final step it's the
+		// assistant text/reasoning message alone.
+		func(stepMessages []fantasy.Message) {
+			for _, msg := range stepMessages {
+				_, _ = m.session.AppendMessage(msg)
+			}
+		},
 		func(inputTokens, outputTokens, cacheReadTokens, cacheCreationTokens int64) {
 			// Emit step usage event for real-time cost tracking
 			if viper.GetBool("debug") {
@@ -1331,11 +1342,17 @@ func (m *Kit) generate(ctx context.Context, messages []fantasy.Message) (*agent.
 //  2. Persist pre-generation messages to the tree session.
 //  3. Build context from the tree (walks leaf-to-root for current branch).
 //  4. Emit turn/message start events.
-//  5. Run generation.
-//  6. Emit turn/message end events.
-//  7. Persist post-generation messages (tool calls, results, assistant).
+//  5. Run generation (messages are persisted incrementally per step).
+//  6. Persist any remaining messages not covered by incremental persistence.
+//  7. Emit turn/message end events.
 //  8. Run AfterTurn hooks.
 //
+// During generation, each completed step's messages are persisted immediately
+// via the onStepMessages callback. Tool calls are always persisted as
+// call/response pairs (assistant + tool messages together). Reasoning and
+// text-only assistant messages are persisted as soon as their step completes.
+// This ensures long-running turns don't lose progress on crash or cancellation.
+//
 // promptLabel is the human-readable label emitted in TurnStartEvent.Prompt.
 // prompt is the raw user text passed to BeforeTurn hooks.
 func (m *Kit) runTurn(ctx context.Context, promptLabel string, prompt string, preMessages []fantasy.Message) (*TurnResult, error) {
@@ -1405,16 +1422,18 @@ func (m *Kit) runTurn(ctx context.Context, promptLabel string, prompt string, pr
 
 	result, err := m.generate(ctx, messages)
 	if err != nil {
-		// Persist any messages from completed steps (tool call/result
-		// pairs) so partial progress is not lost. The agent layer only
-		// includes fully-paired tool_use + tool_result messages in
-		// completedStepMessages, so there are no orphaned entries that
-		// would break subsequent API requests. The user message and any
-		// completed work remain in the session; only the in-progress
-		// (pending) message or tool call is discarded.
-		if result != nil && len(result.ConversationMessages) > sentCount {
-			for _, msg := range result.ConversationMessages[sentCount:] {
-				_, _ = m.session.AppendMessage(msg)
+		// Persist any messages from completed steps that were NOT already
+		// persisted incrementally by the onStepMessages callback. The agent
+		// layer only includes fully-paired tool_use + tool_result messages
+		// in completedStepMessages, so there are no orphaned entries that
+		// would break subsequent API requests.
+		if result != nil {
+			newMessages := result.ConversationMessages[sentCount:]
+			alreadyPersisted := result.PersistedMessageCount
+			if alreadyPersisted < len(newMessages) {
+				for _, msg := range newMessages[alreadyPersisted:] {
+					_, _ = m.session.AppendMessage(msg)
+				}
 			}
 		}
 		m.events.emit(TurnEndEvent{Error: err})
@@ -1425,12 +1444,17 @@ func (m *Kit) runTurn(ctx context.Context, promptLabel string, prompt string, pr
 
 	responseText := result.FinalResponse.Content.Text()
 
-	// Persist new messages (tool calls, tool results, assistant response)
-	// BEFORE emitting events so that extension handlers calling
-	// GetContextStats() see up-to-date token counts.
+	// Persist any new messages that were NOT already persisted incrementally
+	// by the onStepMessages callback during generation. This handles the
+	// non-streaming path (where onStepMessages is not called) and any edge
+	// cases where the final response messages weren't covered by step callbacks.
 	if len(result.ConversationMessages) > sentCount {
-		for _, msg := range result.ConversationMessages[sentCount:] {
-			_, _ = m.session.AppendMessage(msg)
+		newMessages := result.ConversationMessages[sentCount:]
+		alreadyPersisted := result.PersistedMessageCount
+		if alreadyPersisted < len(newMessages) {
+			for _, msg := range newMessages[alreadyPersisted:] {
+				_, _ = m.session.AppendMessage(msg)
+			}
 		}
 	}
 

pkg/kit/session.go

@@ -6,9 +6,21 @@ import (
 
 // SessionManager defines the contract for conversation storage backends.
 // Implementations can use files (default), databases, cloud storage, etc.
+//
+// Implementations must be safe for concurrent use. During generation,
+// AppendMessage is called incrementally from the agent's step-completion
+// callback while read methods (GetMessages, GetCurrentBranch, etc.) may be
+// called concurrently from the UI or extension goroutines.
 type SessionManager interface {
 	// AppendMessage adds a message to the current branch and returns its entry ID.
 	// The entry ID is used for tree navigation and must be unique within the session.
+	//
+	// During generation, AppendMessage is called incrementally after each
+	// completed agent step rather than in a batch at the end of the turn.
+	// For tool-calling steps, the assistant message (containing tool_use parts)
+	// and the tool-role message (containing tool_result parts) are appended
+	// together as a pair. This ensures the session never contains an orphaned
+	// tool call without its result, which would break subsequent LLM requests.
 	AppendMessage(msg LLMMessage) (entryID string, err error)
 
 	// GetMessages returns all messages on the current branch (from root to leaf),

www/public/session/index.html

@@ -901,6 +901,126 @@ a:hover { text-decoration: underline; }
   color: var(--text-muted);
 }
 
+/* ============================================================
+   Compaction Card
+   ============================================================ */
+.compaction-card {
+  margin: 16px 0;
+  border: 1px solid var(--border);
+  border-radius: var(--radius);
+  background: var(--surface);
+  overflow: hidden;
+}
+
+.compaction-header {
+  display: flex;
+  align-items: center;
+  gap: 10px;
+  padding: 12px 16px;
+  cursor: pointer;
+  user-select: none;
+  transition: background var(--transition);
+  background: var(--surface-raised);
+}
+
+.compaction-header:hover {
+  background: var(--surface-overlay);
+}
+
+.compaction-icon {
+  width: 18px;
+  height: 18px;
+  color: var(--yellow);
+  flex-shrink: 0;
+}
+
+.compaction-title {
+  font-size: 13px;
+  font-weight: 600;
+  color: var(--text-secondary);
+  flex: 1;
+}
+
+.compaction-badge {
+  font-size: 11px;
+  font-weight: 500;
+  color: var(--text-muted);
+  background: var(--surface);
+  padding: 2px 8px;
+  border-radius: 10px;
+  border: 1px solid var(--border);
+}
+
+.compaction-chevron {
+  width: 16px;
+  height: 16px;
+  color: var(--text-faint);
+  transition: transform var(--transition);
+  flex-shrink: 0;
+}
+
+.compaction-card.expanded .compaction-chevron {
+  transform: rotate(180deg);
+}
+
+.compaction-content {
+  max-height: 0;
+  overflow: hidden;
+  transition: max-height var(--transition);
+  border-top: 1px solid transparent;
+}
+
+.compaction-card.expanded .compaction-content {
+  max-height: 2000px;
+  overflow-y: auto;
+  border-top-color: var(--border);
+}
+
+.compaction-summary {
+  padding: 16px;
+  font-size: 13.5px;
+  line-height: 1.7;
+  color: var(--text-secondary);
+}
+
+.compaction-summary .md-content h1,
+.compaction-summary .md-content h2,
+.compaction-summary .md-content h3 {
+  color: var(--text);
+  margin: 16px 0 8px;
+}
+
+.compaction-summary .md-content h1 { font-size: 1.3em; }
+.compaction-summary .md-content h2 { font-size: 1.15em; }
+.compaction-summary .md-content h3 { font-size: 1.05em; }
+
+.compaction-summary .md-content ul,
+.compaction-summary .md-content ol {
+  padding-left: 20px;
+}
+
+.compaction-stats {
+  display: flex;
+  gap: 16px;
+  padding: 12px 16px;
+  background: var(--surface-raised);
+  border-top: 1px solid var(--border-subtle);
+  font-size: 11.5px;
+  color: var(--text-muted);
+  flex-wrap: wrap;
+}
+
+.compaction-stat {
+  display: flex;
+  align-items: center;
+  gap: 4px;
+}
+
+.compaction-stat strong {
+  color: var(--text-secondary);
+  font-weight: 600;
+}
+
 /* ============================================================
    System Prompt Display
    ============================================================ */
@@ -1460,6 +1580,7 @@ a:hover { text-decoration: underline; }
     let userMsgCount = 0;
     let assistantMsgCount = 0;
     let toolCallCount = 0;
+    let compactionCount = 0;
 
     // Render each entry
     for (const entry of path) {
@@ -1491,6 +1612,9 @@ a:hover { text-decoration: underline; }
         renderSystemNotice('Label', entry.label || '', 'label');
       } else if (entry.type === 'session_info') {
         // Already handled above for header
+      } else if (entry.type === 'compaction') {
+        compactionCount++;
+        renderCompaction(entry);
       }
     }
 
@@ -1501,6 +1625,7 @@ a:hover { text-decoration: underline; }
           <div class="stat-item"><strong>${userMsgCount}</strong> user message${userMsgCount !== 1 ? 's' : ''}</div>
           <div class="stat-item"><strong>${assistantMsgCount}</strong> assistant message${assistantMsgCount !== 1 ? 's' : ''}</div>
           ${toolCallCount > 0 ? `<div class="stat-item"><strong>${toolCallCount}</strong> tool call${toolCallCount !== 1 ? 's' : ''}</div>` : ''}
+          ${compactionCount > 0 ? `<div class="stat-item"><strong>${compactionCount}</strong> compaction${compactionCount !== 1 ? 's' : ''}</div>` : ''}
           ${header && header.cwd ? `<div class="stat-item">📁 ${escapeHtml(header.cwd)}</div>` : ''}
         </div>`;
       $conversation.insertAdjacentHTML('beforeend', statsHtml);
@@ -2030,6 +2155,60 @@ a:hover { text-decoration: underline; }
     $conversation.appendChild(el);
   }
 
+  // ============================================================
+  //  Compaction Display
+  // ============================================================
+  function renderCompaction(entry) {
+    const el = document.createElement('div');
+    el.className = 'compaction-card fade-in';
+    
+    const cardId = 'compaction-' + Math.random().toString(36).substr(2, 9);
+    
+    // Build stats
+    const stats = [];
+    if (entry.messages_removed > 0) {
+      stats.push(`<div class="compaction-stat"><strong>${entry.messages_removed}</strong> messages compacted</div>`);
+    }
+    if (entry.tokens_before > 0 && entry.tokens_after > 0) {
+      const saved = entry.tokens_before - entry.tokens_after;
+      stats.push(`<div class="compaction-stat"><strong>${saved}</strong> tokens saved</div>`);
+    }
+    
+    // Format timestamp
+    const timeStr = formatTime(entry.timestamp);
+    
+    el.innerHTML = `
+      <div class="compaction-header" onclick="toggleCompaction('${cardId}')">
+        <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16" fill="currentColor" class="compaction-icon">
+          <path d="M2.5 3.5a.5.5 0 0 1 .5-.5h10a.5.5 0 0 1 .5.5v1a.5.5 0 0 1-.5.5h-10a.5.5 0 0 1-.5-.5v-1Zm0 4a.5.5 0 0 1 .5-.5h10a.5.5 0 0 1 .5.5v1a.5.5 0 0 1-.5.5h-10a.5.5 0 0 1-.5-.5v-1Zm0 4a.5.5 0 0 1 .5-.5h10a.5.5 0 0 1 .5.5v1a.5.5 0 0 1-.5.5h-10a.5.5 0 0 1-.5-.5v-1Z"/>
+        </svg>
+        <span class="compaction-title">Context Compacted</span>
+        ${timeStr ? `<span class="compaction-badge">${timeStr}</span>` : ''}
+        <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16" fill="currentColor" class="compaction-chevron" id="${cardId}-chevron">
+          <path d="M4.427 9.427a.25.25 0 0 0 0 .353l3 3a.25.25 0 0 0 .353 0l3-3a.25.25 0 0 0-.353-.353L8 11.646V4.75a.75.75 0 0 0-1.5 0v6.896L4.78 9.427a.25.25 0 0 0-.353 0Z"/>
+        </svg>
+      </div>
+      <div class="compaction-content" id="${cardId}">
+        ${entry.summary ? `<div class="compaction-summary"><div class="md-content">${renderMarkdown(entry.summary)}</div></div>` : ''}
+        ${stats.length > 0 ? `<div class="compaction-stats">${stats.join('')}</div>` : ''}
+      </div>`;
+    
+    $conversation.appendChild(el);
+  }
+
+  // Toggle compaction card expansion
+  window.toggleCompaction = function(cardId) {
+    const card = document.getElementById(cardId).closest('.compaction-card');
+    const chevron = document.getElementById(cardId + '-chevron');
+    const isExpanded = card.classList.contains('expanded');
+    
+    if (isExpanded) {
+      card.classList.remove('expanded');
+    } else {
+      card.classList.add('expanded');
+    }
+  };
+
   // ============================================================
   //  System Prompt Display (collapsible)
   // ============================================================