memory place says 'memory is off' on the default engine road while memory is on
Mantenedores costumam responder em até 1 dia
@santoshkumarradha já está trabalhando nisso.
Desde 27/9/2026.
Avaliação
- Dificuldade
- 4/5
- Tempo estimado
- 3-5 dias
- Facilidade para iniciantes
- 64/100
Direção de pesquisa
Start with internal/tui3/place_memory.go, internal/tui3/memory.go, and internal/tui3/app.go, then trace the ListMemories path and host adapter from cmd/codeaf/chatv3_host.go. Run the deterministic demo-home replication to compare the engine and --no-host roads. Done means both roads show the same shelves and counts, while memory off shows the off message and row edits are handled or clearly reported.
Escrita pelo modelo de indexação a partir do texto da issue.
Descrição
What happened
On dev@c34a3a76d (2026-09-26), the memory place on the default engine road says memory is off, even though memory is on and the store holds memories. The same home on the in-process road shows them. The memory place is reached with alt+8 from a plain codeaf launch.
memory
what it has learned about you and this machine · /remember adds a line
─ memory is off for this session · turn it on under /settings ──────────
On the same home, /settings reads memory on, and codeaf chat --no-host draws 11 held · 3 shelves on the same page.
This is not new: the guard below predates v0.4.1, so stable has it too.
Replication
Deterministic (no model). Seed the repository's demo home, which ships 12 memories with memory on, and compare the two roads:
make build
go build -o bin/codeaf-demo-home ./cmd/codeaf-demo-home
bin/codeaf-demo-home --into /tmp/cd-demo # prints "12 memories"
# use a short state root: the engine's unix socket lives under it
ln -s /tmp/cd-demo/.codeaf /tmp/cdm
cd /tmp/cd-demo/codeaf
HOME=/tmp/cd-demo CODEAF_HOME=/tmp/cdm /path/to/bin/codeaf # esc past setup, alt+8 → "memory is off for this session"
HOME=/tmp/cd-demo CODEAF_HOME=/tmp/cdm /path/to/bin/codeaf chat --no-host # alt+8 → "11 held · 3 shelves"
No model call is needed to see it. The page draws before anything is sent.
Where
internal/tui3/place_memory.gomemoryReady()requiresa.brain(). Ininternal/tui3/memory.go,brain()only accepts an agent that implements the in-processmemoryAgent(Remembers/Remember/Forget/Memories). The engine road's remote agent does not implement it, so the store reader that the host already supplies (Memory: memoryfromnewHostMemoryincmd/codeaf/chatv3_host.go) is never consulted./memories(internal/tui3/app.go,memoryReady()) takes the same wrong branch.- A first attempt made the page read
a.memoryon the engine road, and its unit tests passed with an injected store. The real repro above still printed "memory is off". So something on the engine side also answers "off": the wire'sListMemoriespath, or the host adapter's error mapping. That is the part left to find.
The fix
The memory place and /memories on the engine road read the engine's store through the host's memory reader. The "off" line appears only when the setting really is off. Row edits either go over the wire or say plainly where they can be done.
Acceptance
End to end, through the real door: the replication above, on the engine road, shows the same shelves and counts as --no-host (11 held · 3 shelves on the demo home) and never prints memory is off for this session. With memory off in /settings, both roads print that line. A tmux e2e subtest in internal/e2e would pin it (a seeded home, alt+8, assert on the shelf count line).
- Linguagem predominante
- Go
- Estrelas
- 115
- Forks
- 14
- Merge médio
- 9h 44min
- PRs com merge (30d)
- 766
Preparar o ambiente
Ainda não verificamos os arquivos de configuração deste projeto. Comece pelo README e veja nosso guia da primeira contribuição para os passos gerais.
Primeiros passos
- Leia a issue inteira e depois o guia de contribuição do projeto.
- Comente na issue dizendo que vai assumir — evita que duas pessoas façam o mesmo trabalho.
- Faça um fork do repositório e trabalhe em uma branch.
- Abra um pull request que referencie o número da issue.
Mais de Agent-Field/CodeAF
-
area:chat bug sev:papercut
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 86/100
Agent-Field/CodeAF#1592 ·
Mantenedores costumam responder em até 1 dia
-
area:headless bug sev:critical
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 88/100
Agent-Field/CodeAF#1566 · 1 responsável ·
Mantenedores costumam responder em até 1 dia
-
area:chat feature
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 88/100
Agent-Field/CodeAF#1510 ·
Mantenedores costumam responder em até 1 dia
-
area:tests bug
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 82/100
Agent-Field/CodeAF#1489 ·
Mantenedores costumam responder em até 1 dia
-
area:chat bug good first issue sev:papercut
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 78/100
Agent-Field/CodeAF#1470 ·
Mantenedores costumam responder em até 1 dia
Todas as issues de Agent-Field/CodeAF
Issues semelhantes
-
automation models
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 78/100
Mantenedores costumam responder em até 1 dia
-
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 72/100
txn2/mcp-data-platform#1984 ·
Mantenedores costumam responder em até 1 dia
-
agentic-workflows
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 68/100
Mantenedores costumam responder em até 1 dia
-
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 88/100
Mantenedores costumam responder em até 1 dia
-
kind/docs prio/P2
Dificuldade 1/5 Menos de uma hora Facilidade para iniciantes 95/100
agent-substrate/substrate#1986 ·
Mantenedores costumam responder em até 1 dia