smrti
Long-term memory for AI agents in a single SQLite file. Your agent remembers what matters, forgets what doesn't, and never repeats a critical mistake — no vector database, no external services, no infrastructure.
Inspired by AtomSpace: memories are graph nodes with truth values, attention weights, and emotional valence. Retrieval fuses vector and lexical search, expands one hop through extracted relations, and ranks by relevance scaled by standing (attention, confidence, tone) under a personality that sets the weights, so nothing outranks a memory about the question because it is important elsewhere. Consolidation runs when the agent is used, not when the clock ticks.
- Why smrti
- Install
- Quick Start
- How It Works
- Server Modes (MCP, REST, proxy, visualizer)
- Configuration Reference
- Security & Monitoring
- Multi-Tenant / Space Model
- Personality System
- Architecture
- Data Model
- smrti-town
- Testing
- Benchmarks
- Upgrading
Why smrti
- Zero infrastructure — one SQLite file with sqlite-vec for KNN, and ONNX embeddings, NER, and decisions on CPU (no PyTorch anywhere).
pip installand go. - Error-avoidance memory — severe failures survive pruning and outrank recent trivia at recall; every result comes back classified as
critical_warning,known_antipattern, orcontext. - Knowledge graph — in the server modes, a GLiNER2 + LLM pipeline extracts entities and typed relations from what you store and resolves pronouns against the graph; no manual schema.
- Three integration paths — MCP server, REST API, or an OpenAI-compatible proxy that adds memory to an existing app by changing one base URL.
- Multilingual — 50+ languages end-to-end: multilingual embeddings, zero-shot NER, language-agnostic sentiment.
- Personality-driven — six presets (17 hyperparameters) shape what each agent notices, retains, and forgets.
Install
pip install smrti
Container
Same package as pip install smrti, with the embedding model bundled so a fresh container recalls offline instead of stalling on a first-run download.
docker run -d -p 8420:8420 -v smrti-data:/data ghcr.io/cyqlelabs/smrti
That serves the REST API. Any other command replaces it, and every environment variable in the configuration reference works as usual:
docker run -d -p 8421:8421 -v smrti-data:/data \
-e SMRTI_UPSTREAM_URL=http://host.docker.internal:11434/v1 \
-e SMRTI_API_KEY=your-key \
ghcr.io/cyqlelabs/smrti serve proxy --host 0.0.0.0 --port 8421
- Tags — every
v*release publisheslatest, the exact version, and a rollingMAJOR.MINOR; pin whichever you want to track. - Storage —
/dataholds the database, the weights that download on first use (NER on the first extraction, the decision model on the first decision), and the mapped copies of the embedding and NER models that the first load writes; mount a volume or all of them die with the container. - User — runs as non-root
smrti.
Quick Start
Python API
from smrti import Smrti
mem = Smrti(db_path="~/.smrti/memory.db", personality="balanced")
# Store memories
mem.remember("Alice prefers TypeScript", probability=0.9, valence=0.3)
mem.remember("The deploy pipeline is broken", probability=0.95, valence=-0.7)
# Recall by relevance and salience
results = mem.recall("programming languages")
for r in results:
print(f"{r.atom.label} (salience={r.salience:.2f}, confidence={r.atom.truth.confidence:.2f})")
# Report that the recalled memories were used
mem.reinforce([r.atom.id for r in results])
# Assert a belief with a reason, and read the reason back
atom_id = mem.believe("Python is the best language for ML", probability=0.85, evidence="Team survey results")
print(mem.evidence(atom_id)[0].text)
# Stop a memory from surfacing
mem.forget("deploy pipeline")
# Consolidate: revise evidence, decay, promote, prune
epoch = mem.reflect()
print(f"Updated {epoch.beliefs_updated} beliefs, pruned {epoch.atoms_pruned} atoms")
The constructor also takes tenant_id, write_space, read_spaces, ignore_patterns, and temporal (see Multi-Tenant / Space Model). The Python API stores, recalls, forgets, and consolidates; entity extraction and relative-date resolution run in the server modes (pass temporal=True for dates here). The embedding model downloads on first use, the NER weights on first extraction.
CLI
smrti init --db ~/.smrti/memory.db --personality balanced # create a database
smrti status # inspect it
smrti serve mcp # MCP stdio server (Claude, etc.)
smrti serve rest # REST API on :8420
smrti serve viz # REST API + memory visualizer in the browser
smrti serve proxy # OpenAI-compatible proxy on :8421
smrti serve town # city-builder simulation demo on :8430
smrti stop # gracefully stop all servers started by `smrti serve`
smrti stop rest # stop one mode (rest, viz, proxy, town); --port to narrow further
How It Works
remember()
Embeds and stores text as a typed atom (episode, concept, belief, or goal) with a truth value, attention weight, and valence.
valenceunset is estimated from the text. Set it yourself and the memory is a deliberate report, the only kind recall can raise to a behavioral constraint.intensityis how strongly the tone is felt; unset, it is|valence|.type="belief"isbelieve(): a belief starts at confidence 0.3 and earns more through evidence. Asserted at probability ≥ 0.95 it is permanent and exempt from decay. Theevidencereason is kept on the evidence log;evidence(atom_id)lists it.- In the server modes, relative dates are resolved against the write time ("the session is tomorrow" still names a day next week), and entities and relations are extracted into the graph. A claim that replaces an earlier one about the same subject (a new city, a new employer, a changed preference) is recorded as superseding it.
recall()
Runs a vector KNN and a BM25 search over the read spaces, fuses them by Reciprocal Rank Fusion, expands one hop through the graph, and ranks by salience:
S = similarity × ( w_sim + w_sti × sti + w_conf × confidence + w_lti × lti + w_val × |valence| × intensity )
Similarity multiplies the standing terms, so a memory that is not about the question cannot outrank one that is; among memories that are, standing decides the order. On top of that:
- When valence < −0.5, weight shifts from STI to valence, so old critical errors outrank recent trivia. Valence terms read the tone a memory was written with, not the mood it has absorbed.
- An episode that restates the query scores nothing.
agent_source_trustdiscounts an agent-authored memory's standing, never its similarity.- Episodes repeating one already chosen from the same minutes share
max(2, top_k // 6)slots; beliefs keep up to two. - Results below the personality's
min_confidence_to_surfaceare excluded unless you passmin_confidence; forgotten memories never return. - The core decision engine judges the top candidates as evidence for the question and blends that judgement with salience; each result then carries an
evidencescore. Passrerank=Falseto skip the judgement. Only what is returned gets the access boost;attend(atom_ids)boosts what an external reranker kept out of a widerrecall(boost=False).
Each result carries a severity: critical_warning (a valence you stated, on anything but a bare concept), known_antipattern (a belief whose probability fell below 0.3, where a superseded preference or constraint lands), or context.
forget()
Stops the memories matching a query from surfacing. They are excluded from every recall, no consolidation lifts them back, and the next epoch may prune them. Forgetting is final.
reinforce()
Reports that recalled memories were used; a cheap test is that distinctive words from a memory appeared in the reply it informed. Use is weak evidence: confidence climbs a little, capped per epoch and discounted for agent-authored memories, and probability does not move.
reflect()
One consolidation epoch: revise pending evidence, decay attention and confidence, propagate both to neighbors, heal orphaned episodes, promote high-STI atoms to long-term importance, resolve contradictions (a superseded claim loses), link similar high-LTI atoms (every tenth epoch), and prune what fell below the floors. The servers run one every SMRTI_REFLECT_INTERVAL seconds for each space used in that interval, so idle memory does not age. What you told the agent decays only to the surfacing floor and stays recallable unless you forget it; what it inferred keeps fading, faster for agent-authored atoms.
Semantic decisions
At five points a rule cannot tell what a sentence means. Instead of a generative LLM call, the engine asks Laya, a local multilingual model that answers typed questions (a probability, a choice among options, a score) and never writes prose. It runs as an int8 ONNX graph — no PyTorch, about 560MB resident. Every task is active by default; a task can run in shadow (ask and record, apply nothing) or be turned off. Any failure or timeout falls back to the deterministic path.
The runtime installs with Smrti; the weights do not. The first decision fetches them once (250MB, checked against a checksum) into ~/.smrti/decision-model, or reuses ~/.factor/decision-model when Factor already holds a copy. Fetch and load run on their own thread, so recall answers from the deterministic path until the model is ready. SMRTI_DECISIONS_MODEL points at a directory you unpacked yourself; SMRTI_DECISIONS_URL asks a server that already holds the model (Factor's EdgeJev on loopback, or any POST /v1/systemone) and loads nothing here — one copy of 560 MB is what a small machine can hold, not two; SMRTI_DECISIONS=off downloads nothing at all.
| Task | Decides |
|---|---|
routing |
Whether an episode needs the LLM claim extraction. Chatter skips it; a durable fact that names no entity ("never deploy on Fridays") forces it. |
rerank |
Whether each shortlisted memory answers the question, links to the answer, describes a replaced state, or contradicts the premise; blended with salience. |
supersession |
Whether a new claim replaces an older one. Ambiguity keeps both. |
entity |
Which existing atom an uncertain mention names. Under the confidence line, a duplicate is made rather than a wrong merge. |
tone |
Whose an estimated negative valence is: about the thing the memory describes, or the speaker's own mood. Mood is damped so an apology does not rank and persist like the failure it is about; a stated valence is never questioned. |
A decision never restores a forgotten atom, confers permanence, changes tenant or space, or mints a critical warning. Its confidence is logged, never written into a truth value.
export SMRTI_DECISIONS=off # deterministic only
export SMRTI_DECISIONS_RERANK=shadow # per task: off | shadow | active
export SMRTI_DECISIONS_MODEL=/models/laya-int8 # a directory you unpacked; no download
export SMRTI_DECISIONS_THREADS=2 # cap the cores inference may hold
GET /decisions lists every decision, /metrics counts them as smrti_decisions_total, make bench-decisions scores the routing gate and make bench-tone the tone check. Thresholds are in the Configuration Reference.
Server Modes
MCP Server
Exposes 8 tools over stdio for direct LLM integration.
Claude Code:
claude mcp add smrti -- smrti serve mcp
Claude Desktop (or any MCP client) — add to your MCP config:
{
"mcpServers": {
"smrti": {
"command": "smrti",
"args": ["serve", "mcp"],
"env": { "SMRTI_DB": "~/.smrti/memory.db" }
}
}
}
| Tool | Description |
|---|---|
smrti_remember |
Store an episode, goal, or belief (use type=belief + evidence to assert a probabilistic fact) |
smrti_recall |
Vector and BM25 search fused, ranked by salience, each result classified by severity |
smrti_reflect |
Run a consolidation epoch |
smrti_forget |
Stop memories matching a query from surfacing; the next epoch may prune them |
smrti_status |
Memory statistics and the tenant's spaces |
smrti_personality |
Get or set the personality preset |
smrti_space_query |
Compare two spaces: op=overlap (Jaccard), op=intersection, op=diff |
smrti_space_merge |
Materialize a bridge space from the overlap between two spaces |
REST API
Full CRUD over HTTP on port 8420:
smrti serve rest --host 0.0.0.0 --port 8420
# Store a memory
curl -X POST http://localhost:8420/remember \
-H "Content-Type: application/json" \
-d '{"content": "Alice prefers TypeScript", "probability": 0.9}'
# Recall
curl -X POST http://localhost:8420/recall \
-d '{"query": "programming languages", "top_k": 5}'
# Report that recalled memories shaped the reply — being used builds confidence
curl -X POST localhost:8420/reinforce \
-H 'Content-Type: application/json' \
-d '{"atom_ids": ["4f2c…", "9ab1…"]}'
# Run consolidation
curl -X POST http://localhost:8420/reflect
# Get status
curl http://localhost:8420/status
# Compare two spaces (op = overlap | intersection | diff)
curl -X POST http://localhost:8420/space_query \
-d '{"op": "overlap", "other_space": "personal"}'
# Grow a bridge space from what two spaces share
curl -X POST http://localhost:8420/space_merge \
-d '{"other_space": "personal", "min_jaccard": 0.1}'
# Boost only what you kept from a wide, unboosted recall you reranked yourself
curl -X POST http://localhost:8420/attend \
-d '{"atom_ids": ["4f2c…"]}'
# The semantic decision log (see "Semantic decisions"), newest first
curl http://localhost:8420/decisions
Every endpoint takes an optional space to route the call; /space_query and
/space_merge compare that space with other_space and refuse a self-compare.
OpenAI-Compatible Proxy
The fastest way to add memory to an existing app: point your OpenAI client at the proxy and keep everything else the same. It intercepts each chat request, injects relevant memories into the system prompt, and stores the exchange afterward.
smrti serve proxy --host 0.0.0.0 --port 8421 --upstream https://api.openai.com
from openai import OpenAI
client = OpenAI(
base_url="http://localhost:8421/v1",
api_key="sk-..." # forwarded to upstream
)
response = client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": "What do you know about Alice?"}],
extra_headers={
"X-Smrti-Tenant-Id": "user_123",
"X-Smrti-Write-Space": "work",
"X-Smrti-Read-Spaces": "work,personal",
}
)
On every request the proxy:
- Recalls relevant memories from the read spaces, building the query from recent conversation context (not just the last message)
- Injects them into the system prompt in two sections — behavioral constraints (
YOU MUST NOT/AVOID) forcritical_warningandknown_antipatternmemories, and background context (Note:) for the rest — each with a confidence qualifier - Stores the user message and assistant response as episodes (identical episodes are deduplicated per tenant/space)
- Extracts entities and claims into concept nodes and typed relation edges
Works with any OpenAI-compatible upstream — including local llama.cpp, vLLM, or Ollama endpoints.
Memory Visualizer
smrti serve viz opens a browser-based graph explorer to inspect atoms, relations, and attention weights, plus an LLM Calls debug tab showing every extraction request with full request/response, timing, and recalled memories.
Configuration Reference
All server modes read the same environment variables. Everything works with zero configuration; set these to customize.
Core (all modes):
| Variable | Default | Purpose |
|---|---|---|
SMRTI_DB |
~/.smrti/memory.db |
Database file path |
SMRTI_HOME |
~/.smrti |
Where the decision model is unpacked |
SMRTI_PERSONALITY |
balanced |
Personality preset |
SMRTI_TENANT_ID |
default |
Tenant partition (hard isolation) |
SMRTI_SPACE |
default |
Write space |
SMRTI_READ_SPACES |
write space | Comma-separated spaces to read from |
SMRTI_REFLECT_INTERVAL |
60 |
Seconds between consolidation epochs (0 = off); only spaces used during the interval get one |
SMRTI_RUN_DIR |
~/.smrti/run |
Where smrti serve writes PID files so smrti stop can find its servers |
SMRTI_IGNORE_PATTERNS |
— | Newline-separated regexes; matching content is dropped before storage (see below) |
Security (REST / proxy / viz):
| Variable | Default | Purpose |
|---|---|---|
SMRTI_API_KEY |
— | When set, every request must send Authorization: Bearer <key> or X-Api-Key: <key> |
SMRTI_CORS_ORIGINS |
— | Comma-separated allowed origins for the proxy; CORS middleware is only added when set |
SMRTI_VIZ_DBS |
— | :-separated extra SQLite paths the visualizer may open (default: the server's own DB only) |
Proxy:
| Variable | Default | Purpose |
|---|---|---|
SMRTI_UPSTREAM_URL |
https://api.openai.com |
Upstream OpenAI-compatible API |
SMRTI_RECALL_TOP_K |
5 |
Memories to inject per request |
SMRTI_RECALL_MIN_CONFIDENCE |
personality floor | Confidence floor for injected memories |
SMRTI_QUERY_MODE |
concat |
Recall query source: concat recent context or last message only |
SMRTI_QUERY_CONTEXT_MSGS |
5 |
Recent messages included in the recall query |
SMRTI_QUERY_MAX_CHARS |
500 |
Max characters of the recall query |
SMRTI_INJECT_MAX_CHARS |
500 |
Max characters per injected memory |
SMRTI_INJECT_BUDGET_CHARS |
0 |
Total characters of injected memory per request, constraints first, then context in rank order (0 = no budget) |
Extraction (all modes):
To get the knowledge graph from serve rest or serve mcp, point SMRTI_EXTRACT_URL (and SMRTI_EXTRACT_MODEL) at an OpenAI-compatible endpoint; the proxy uses its upstream. Unset, extraction runs local NER only.
| Variable | Default | Purpose |
|---|---|---|
SMRTI_EXTRACT |
1 |
Entity/claim extraction after every remember (0 = off) |
SMRTI_EXTRACT_MODE |
hybrid |
hybrid (GLiNER + LLM), llm (LLM-only), local (no LLM) |
SMRTI_EXTRACT_URL |
proxy upstream, else unset | LLM endpoint for extraction calls; unset with no upstream = local mode |
SMRTI_EXTRACT_MODEL |
request model | Model for extraction calls |
SMRTI_EXTRACT_THINKING |
disabled |
disabled / auto / enabled; disabled is faster and avoids token-budget exhaustion on thinking models |
SMRTI_EXTRACT_TIMEOUT |
60 |
Extraction request timeout in seconds |
SMRTI_NER_MODEL |
lmo3/gliner2-multi-v1-onnx |
GLiNER2 ONNX model for local zero-shot NER |
SMRTI_TEMPORAL |
1 |
Resolve relative dates against the write time (0 = store text verbatim); one NER pass per write, never in llm mode, where the extraction model resolves dates itself |
Semantic decisions (all modes, see above):
| Variable | Default | Purpose |
|---|---|---|
SMRTI_DECISIONS |
active |
Mode for every decision task: off, shadow (ask and record, apply nothing), active |
SMRTI_DECISIONS_ROUTING / _RERANK / _SUPERSESSION / _ENTITY / _TONE |
SMRTI_DECISIONS |
Per-task mode override |
SMRTI_DECISIONS_MODEL |
Factor's copy, else ~/.smrti/decision-model |
Model directory to load; unset, the weights are fetched there on first use |
SMRTI_DECISIONS_URL |
unset | A server already holding the model (EdgeJev's POST /v1/systemone, e.g. Factor's on http://127.0.0.1:8731); set, nothing is loaded here |
SMRTI_DECISIONS_DEVICE |
auto | Execution provider passed to the ONNX runtime; unset lets it choose |
SMRTI_DECISIONS_THREADS |
runtime default | Cores one decision may hold; cap it on a machine that has two of them |
SMRTI_DECISIONS_TIMEOUT |
30 |
Deadline per local decision in seconds; on expiry the deterministic path answers |
SMRTI_DECISIONS_COOLDOWN |
60 |
Seconds the engine stops asking a provider that just failed, so an unavailable model costs one deadline per window instead of one per recall |
SMRTI_DECISIONS_CACHE |
256 |
Answers kept in the LRU cache; a state names the atoms it was built from, so it invalidates itself (0 = no cache) |
SMRTI_DECISIONS_RERANK_SHORTLIST |
20 |
Candidates judged per recall (one request) |
SMRTI_DECISIONS_RERANK_WEIGHT |
0.5 |
Share of the final order the evidence judgement decides against salience |
SMRTI_DECISIONS_RERANK_MIN_EVIDENCE |
0 |
Drop judged candidates under this evidence score (0 = rerank only, never filter); a stated warning is never dropped |
SMRTI_DECISIONS_ROUTING_SKIP / _FORCE |
0.2 / 0.75 |
Lines under which every routing judgement means "skip the LLM", and over which a durable fact or correction forces it |
SMRTI_DECISIONS_SUPERSESSION_MIN_CONFIDENCE |
0.6 |
Choice confidence a supersession verdict needs before the older claim is marked replaced |
SMRTI_DECISIONS_ENTITY_MIN_CONFIDENCE |
0.6 |
Choice confidence an entity verdict needs before an uncertain match is accepted |
SMRTI_DECISIONS_ENTITY_CANDIDATES |
5 |
Matches each resolution tier offers for verification |
SMRTI_DECISIONS_TONE_MIN_CONFIDENCE |
0.4 |
Choice confidence a tone verdict needs before an estimated valence is damped; under it the estimate stands. Measured: make bench-tone |
SMRTI_DECISIONS_TONE_DAMPING |
0.25 |
What an estimated valence is multiplied by when the tone is judged the speaker's mood rather than the memory's; under every line the engine draws through a tone |
Ignoring Automated Messages
Agentic frameworks often produce periodic system messages (heartbeat checks, status pings, tool scaffolding) that should not pollute memory. Any remember() call matching SMRTI_IGNORE_PATTERNS is silently dropped before embedding or extraction:
export SMRTI_IGNORE_PATTERNS="^# Heartbeat Check
^HEARTBEAT_OK$"
Patterns are matched with re.search (anchors optional) and apply to all server modes.
Security & Monitoring
API key auth — HTTP servers are open by default for local use. Set SMRTI_API_KEY to require a key on every REST, proxy, and visualizer request (Authorization: Bearer <key> or X-Api-Key: <key>). The CLI warns when you bind to a non-loopback host without a key set.
Prometheus metrics — REST and proxy expose GET /metrics in Prometheus text format, with zero extra dependencies. Gauges include smrti_atoms_total, smrti_atoms_by_type, smrti_epoch_count, and the active personality hyperparameters, all labeled by tenant and space.
Multi-Tenant / Space Model
Tenants are hard walls: atoms, embeddings, and attention weights never cross them. Spaces are permeable layers within a tenant: you write to one, read from many, and each has its own personality and consolidation cycle.
researcher = Smrti(tenant_id="team", write_space="researcher",
read_spaces=["researcher", "shared"], personality="curious")
deployer = Smrti(tenant_id="team", write_space="deployer",
read_spaces=["deployer", "shared"], personality="deterministic")
coordinator = Smrti(tenant_id="team", write_space="coordinator",
read_spaces=["coordinator", "shared", "researcher", "deployer"],
personality="analytical")
shared = Smrti(tenant_id="team", write_space="shared")
Each space consolidates independently: the researcher forgets fast, the deployer holds onto critical failures, and the coordinator reads everything through its own personality. This suits agent teams with private working memory and shared context, multi-agent simulations, and per-role views of one user.
Spaces also support set operations (overlap, intersection, difference, union, symmetric difference) and can materialize a bridge space from what two spaces share, via the space_query and space_merge MCP tools or the POST /space_query and POST /space_merge REST endpoints. Bridges are built only on request, never by the consolidation epoch.
Personality System
Six built-in presets control retrieval behavior, decay rates, and emotional dynamics:
| Preset | Bias | Use Case |
|---|---|---|
balanced |
Equal weights across all signals | General-purpose agents |
analytical |
High confidence weight, low valence | Logical reasoning, data-driven decisions |
curious |
High STI weight, fast decay | Exploration, novelty-seeking |
empathetic |
High valence weight, emotional propagation | Relationship-focused agents |
maverick |
Slow decay, high propagation | Independent, contrarian reasoning |
deterministic |
Fast learning, slow decay, laser focus | Agentic workflows, code gen, deployments |
Each preset tunes 17 hyperparameters. To create a custom personality, start from a preset and override individual values via the personality DB table or the /personality API endpoint.
Hyperparameter reference (17 parameters, defaults from the balanced preset)
Salience weights — control how retrieval ranks results (should sum to ~1.0):
| Parameter | Default | Effect |
|---|---|---|
w_similarity |
0.35 | Weight of embedding cosine similarity |
w_sti |
0.25 | Weight of short-term importance (recency/access) |
w_confidence |
0.20 | Weight of truth value confidence |
w_lti |
0.10 | Weight of long-term importance |
w_valence |
0.10 | Weight of emotional intensity (dynamically boosted when valence < -0.5) |
Belief dynamics — govern how confidence evolves over time:
| Parameter | Default | Effect |
|---|---|---|
confidence_decay_rate |
0.02 | Per-epoch confidence decay. Higher = memories fade faster |
confidence_update_lr |
0.3 | How far one observation moves a belief. Higher = new evidence counts for more |
min_confidence_to_surface |
0.1 | Recall floor when the caller passes none; user testimony stops decaying here |
Attention dynamics — control what stays in focus:
| Parameter | Default | Effect |
|---|---|---|
sti_decay_rate |
0.1 | Per-epoch STI decay. Higher = faster attention loss |
sti_boost_on_access |
0.5 | STI added each time an atom is recalled. Higher = stronger recency bias |
sti_propagation_factor |
0.15 | Fraction of STI boost propagated to linked atoms. Higher = broader activation |
lti_promotion_threshold |
0.7 | STI above which an atom earns long-term importance. Higher = harder to earn |
lti_decay_rate |
0.01 | Per-epoch LTI decay. Higher = long-term importance erodes faster |
Provenance — weighs what the agent wrote against what the user said:
| Parameter | Default | Effect |
|---|---|---|
agent_source_trust |
0.5 | Standing of agent-authored memories relative to user-stated ones. Lower = model output fades faster while what the user said persists |
Emotional dynamics — shape how valence influences behavior:
| Parameter | Default | Effect |
|---|---|---|
valence_weight |
0.2 | Strength of the weight shift from STI to valence for severely negative memories |
valence_propagation |
0.1 | Fraction of valence spread to linked atoms each epoch. Moves the mood a recall result reports, not the tone ranking and severity judge on |
mood_inertia |
0.8 | Resistance to mood shifts (0 = reactive, 1 = stable) |
Architecture
graph TD
subgraph Facade
S["Smrti<br/><small>remember · recall · attend · believe · reinforce · reflect · forget · status</small>"]
end
subgraph Servers
MCP["mcp.py<br/><small>MCP stdio</small>"]
REST["rest.py<br/><small>FastAPI :8420</small>"]
PROXY["proxy.py<br/><small>OpenAI proxy :8421</small>"]
end
subgraph Core
AS["AtomSpace"]
DB["Database"]
EMB["Embedder"]
MOD["Models"]
end
subgraph Retrieval
FAN["fan_out<br/><small>vector + BM25, fused</small>"]
SAL["salience"]
DIV["diversify"]
CLS["classify"]
end
subgraph Evolution
EPO["epoch"]
TRU["truth"]
REI["reinforcement"]
CON["connections"]
HEA["healing"]
end
subgraph Spaces
SOP["set_ops"]
EMG["emergence"]
end
subgraph Extraction
EXT["extract"]
RES["resolve"]
ALI["aliases"]
TMP["temporal"]
end
subgraph Decisions
DEC["engine · policies · audit<br/><small>core · active by default</small>"]
LAYA["laya · model<br/><small>multilingual int8 ONNX · local typed judgements</small>"]
end
subgraph Storage
SQL["SQLite + sqlite-vec<br/><small>multilingual-MiniLM-L12-v2 · 384d · ONNX CPU</small>"]
end
MCP & REST & PROXY --> S
S --> Core & Retrieval & Evolution & Extraction & Spaces
Retrieval & Extraction -.->|"shortlist · route · verify"| DEC
DEC --> LAYA
Core & Retrieval & Evolution & Extraction & Spaces --> SQL
Retrieval: embed the query → vector + BM25 search, fused → 1-hop graph expansion → salience ranking → local evidence judgement → diversity cap → top-k → access boost on what was returned. Consolidation: the epoch steps under reflect().
Data Model
| Atom Type | Purpose | Example |
|---|---|---|
concept |
Reusable entities | "Alice", "Python", "OpenAI" |
belief |
Probabilistic facts | "Alice prefers TypeScript" |
episode |
Timestamped observations | "User asked about deployment" |
goal |
Desired states | "Finish the migration by Friday" |
relation |
Edges between atoms | Alice → works_at → Acme Corp |
Each atom carries:
- TruthValue —
probability[0,1] andconfidence[0,1]. Confidence grows with evidence: every stated reason, re-mention, report of use, or supersession revises it (PLN revision). For episodes and concepts, which hold no proposition, it is retention strength rather than truth. - AttentionValue —
sti(short-term importance, decays fast) andlti(long-term, accumulates). - Valence — emotional tone [-1,1] and intensity [0,1]. Ranking and severity read the tone a memory was written with; the
valencea recall result reports is its current mood, which drifts toward its neighbors during consolidation. - Evidence — an append-only log of observations: the probability observed, its weight, who reported it, and what was observed (
text).
smrti-town
A living demo: smrti-town is a city-builder where every citizen carries a persistent smrti memory graph. You place the Town Hall and choose a mayor; an LLM-generated council debates what to build, citizens immigrate, work, and petition — and every decision they make is driven by what they remember.
smrti serve town # simulation + frontend on :8430
Testing
pytest tests/ -v
Benchmarks
Two harnesses in bench/ ingest a published dataset as episodes and answer its questions through recall. Retrieval and answering are scored separately, so a strong answering model cannot hide a retrieval regression.
Config (2026-08-26): extraction off · no consolidation epochs · top_k=50 · deterministic preset · gemini-3.7-flash answering and judging. This measures retrieval alone, and it predates the current ranking formula; re-run make bench before reading the numbers against the current engine.
| Benchmark | Scope | Retrieval | Answers | Notes |
|---|---|---|---|---|
| LongMemEval-S | 40 questions, one per ability in turn | 0.975 hit · 0.912 evidence recall | 0.900 | 5 of 6 abilities retrieve without a miss |
| HaluMem | 3 personas, 180 questions | — | 0.517 correct | hallucination 0.394 · omission 0.089 |
make datasets # fetch both into data/ (265MB + 32MB, once)
make bench # fails if the retrieval hit rate drops
make bench-halumem # fails if the hallucination rate rises
# add --extract-url/--extract-model to either to build the entity graph
# add --epochs N to consolidate each history N times before querying it
# add --top-k 5 to measure at the proxy's injection budget (10 for the MCP tool)
Each benchmark locks its config beside a recorded baseline and refuses to compare across configs; --epochs and --top-k join the fingerprint when set. Neither is a CI gate: both need the datasets, the embedding model, and a judge key.
Where it is strong. LongMemEval retrieves the annotated evidence for five of six abilities without a miss; temporal reasoning and assistant's-own-words questions are answered perfectly. On HaluMem's memory boundary questions (things the user never said), smrti answers correctly 97% of the time and invents something 2.9% of the time.
Where it is weak. HaluMem's synthesis categories hallucinate heavily: dynamic update 71%, multi-hop inference 63%, generalization 62%. smrti rarely declines to answer (8.9% omission, where published systems omit 17–35%), so an unknown often comes out as an assertion.
Unmeasured. The entity graph (--extract-url, about 1.25 s and one LLM call per turn) and consolidation (--epochs). HaluMem's memory-extraction and memory-update tasks are not implemented.
These numbers are not directly comparable to published results: the subsets are small (40 to 180 questions, so one question moves a category by several points), a single judge grades them where published protocols average three, and the answering model differs. LongMemEval leaderboard figures for reference: MemOS 77.8, Memobase 72.4, Mem0 66.4, Zep 63.8.
Upgrading
An existing database is upgraded the first time a newer version opens it, with a .pre-migration.bak snapshot written beside it first; restore that file to downgrade. Migrations and data repairs are idempotent. The repairs rebuild the tone of entities extracted by earlier versions, drop the hub edges the old healing step drew, and let the pruner remove memories forgotten before this version. Bridge spaces an older epoch created on its own (a_x_b) are left in place; remove one with DELETE /spaces/current?space=a_x_b.
License
MIT
Release files for smrti 0.21.3
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| smrti-0.21.3.tar.gz | 4.4 MB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| smrti-0.21.3-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 8.9 MB
Release files / smrti-0.21.3.tar.gz
| Download URL | smrti-0.21.3.tar.gz |
|---|---|
| Size | 4.4 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
59543cbd9bd76b0999cb10a6d8bb02a8d7b5c1f1efbb2df6558af5aae1a4940f
|
|
BLAKE2b-256 checksum How to use checksums |
6bfe3f07ded68889a12732cbd9c9ddb46e10dc11b5300b050136a6832eaa0d91
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 23, 2026.
Transparency logRelease files / smrti-0.21.3-py3-none-any.whl
| Download URL | smrti-0.21.3-py3-none-any.whl |
|---|---|
| Size | 4.5 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
eef96aba23339c65fc900ab890ae298749fb558f924ca055bd4542be3888d9b6
|
|
BLAKE2b-256 checksum How to use checksums |
e8a92f2abdfeef3a1d9eab3b923421560d72bdbe7962dd4c9a7aadf328112640
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 23, 2026.
Transparency log