diff --git a/.env.example b/.env.example index 4baf8d6635..b8d9e51ede 100644 --- a/.env.example +++ b/.env.example @@ -15,14 +15,16 @@ # # Defaults shown in comments. No provider is ever picked from a key alone: # agentmemory sends session content to an outside AI provider only after you -# name one with AGENTMEMORY_LLM_PROVIDER or EMBEDDING_PROVIDER. +# name one with AGENTMEMORY_LLM_PROVIDER or EMBEDDING_PROVIDER, or opt into the +# Claude subscription fallback with AGENTMEMORY_ALLOW_AGENT_SDK=true. # ----------------------------------------------------------------------------- # 1. LLM provider: off unless you choose one # ----------------------------------------------------------------------------- # -# Without AGENTMEMORY_LLM_PROVIDER, agentmemory runs in noop mode and sends no -# session content to an LLM, even when a provider key is in the environment. +# Without AGENTMEMORY_LLM_PROVIDER or AGENTMEMORY_ALLOW_AGENT_SDK=true, agentmemory +# runs in noop mode and sends no session content to an LLM, even when a provider +# key is in the environment. # In noop mode: observations are # indexed via zero-LLM synthetic compression and BM25 recall still works, # but LLM-backed summarisation / reflection / consolidation are disabled. @@ -76,7 +78,7 @@ # Stop-hook recursion when invoked from inside Claude Code. # AGENTMEMORY_ALLOW_AGENT_SDK=true -# FALLBACK_PROVIDERS=anthropic,gemini # Comma-separated chain tried after the chosen AGENTMEMORY_LLM_PROVIDER returns an error (e.g. rate limit); ignored when no provider is chosen +# FALLBACK_PROVIDERS=anthropic,gemini # Comma-separated chain tried after the chosen AGENTMEMORY_LLM_PROVIDER returns an error (e.g. rate limit); ignored when zero-LLM is active # ----------------------------------------------------------------------------- # 2. Embedding provider: off unless you set EMBEDDING_PROVIDER @@ -125,7 +127,7 @@ # AGENTMEMORY_GRAPH_WEIGHT=0.2 # Graph traversal bonus on smart-search ranking # AGENTMEMORY_GRAPH_COMPACT_ON_BOOT=false # Report boot compaction as off (otherwise deferred; manual maintenance only) # TOKEN_BUDGET=2000 # Max tokens injected via mem::context per session -# MAX_OBS_PER_SESSION=500 # Per-session observation cap before consolidation kicks in +# MAX_OBS_PER_SESSION=500 # Hard per-session cap: later observations of that session are refused; only capture-worker refusals are parked as dead letters (0 = no cap) # SUMMARIZE_CHUNK_SIZE=400 # When mem::summarize sees a session larger than this, it chunks observations and map-reduces (chunk-summarize → reduce-merge) to stay within the LLM's context window. Default 400 ≈ 50k tokens per chunk at ~110 tok/obs. Native sessions are capped by MAX_OBS_PER_SESSION; chunking primarily matters for bulk-imported jsonl sessions, which bypass that cap. # SUMMARIZE_CHUNK_CONCURRENCY=6 # Parallel chunk LLM calls during chunked summarize. Default 6 fits ~100-chunk sessions under iii's 180s function-invocation timeout at typical ~8s/call. High-throughput providers (Novita, DeepInfra, DeepSeek) commonly allow 100+ concurrent — bump this for very large imported sessions.