aelfrice 3.5.0

Persistent memory for AI coding agents. Local SQLite + UserPromptSubmit hook — matched beliefs in the prompt before the model sees it. Deterministic, auditable, no embeddings, no cloud.

aelfrice

Your AI stops forgetting. Set up once. Stays out of the way.

No cloud. No account. No telemetry.

You correct your agent. "Got it," it says. Next session, same mistake.

aelfrice runs in the background and stops the amnesia. Write a rule once and every relevant prompt thereafter ships with that rule already attached — before the model sees your message. No CLAUDE.md chain to maintain, no cross-references for the agent to skip; the matched beliefs are in the prompt, not in a file the agent is supposed to consult.

For developers using AI coding agents — first-class via the UserPromptSubmit hook for hosts that expose it; any MCP host via the included stdio server. Local-only by design — embeddings, vector RAG, and cloud sync are explicitly out of scope. See Philosophy for the trade-off.

Install

uv tool install aelfrice    # requires uv — https://docs.astral.sh/uv/
aelf setup                  # wire the UserPromptSubmit hook into your agent
aelf onboard .              # one-shot project scan: filesystem, git log, code structure
aelf lock "never push directly to main; use scripts/publish.sh"

That's it. The next prompt that mentions "push" already has the rule. From here on out aelfrice is invisible — no command to remember to run, no file to keep updated.

---

What it does for you

• Stops the AI forgetting your rules. Lock a rule once with aelf lock "..." — it comes back attached to every prompt after that, in every new session. You don't have to remind the AI; aelfrice does.
• The AI can't skip it. What aelfrice remembers is in the prompt before the model sees it — not stored in a file the AI is supposed to check on its own. The reminding happens for the AI, not by you.
• Stays on your computer. One file on your machine. No cloud account. No telemetry. If you stop trusting aelfrice, aelf uninstall removes it cleanly in one command.
• Set up once. Forget it's there. Three commands at install time. After that, you reach for aelf rarely. The background hooks do the rest.

---

How it works

In plain English: four searches run in parallel over your stored rules, the best matches get prepended to your prompt, and the model reads the lot as one message. Below is the wiring for readers who want the receipts.

When you submit a prompt, aelfrice's UserPromptSubmit hook fires before the model sees your message. It runs four retrieval lanes in parallel, merges the result, and prepends the matched beliefs as an <aelfrice-memory> block:

L0: locked beliefs   -> rules you marked permanent (always returned, never trimmed)
L1: FTS5 keyword     -> SQLite full-text search, BM25 + posterior-weighted rerank
L2: graph walk       -> typed-edge BFS from L1 seeds (SUPPORTS, CONTRADICTS, SUPERSEDES, DERIVED_FROM, ...)
L2.5: structural HRR -> Plate-FFT bind/probe against anchor text + structural markers

L0 always ships. L1, L2, and L2.5 are budget-trimmed against the merged candidate set in score-descending order; locked beliefs win every overflow. Default budget: 2,400 tokens per prompt. The default ranking stack is stack-r1-r3 (entity expansion + per-store IDF clipping); bench evidence on the labelled query-strategy corpus measured +0.2851 absolute NDCG@k (+94.8%) versus the v1.4 raw-BM25 baseline at +0.96 ms p99 over legacy-bm25 (gate budget: +5 ms delta; see tests/bench_gate/test_query_strategy.py). Full lane wiring, composition, and federation peer DBs: ARCHITECTURE § Retrieval.

The result is prepended to your prompt verbatim:

<aelfrice-memory>
[locked] never push directly to main; use scripts/publish.sh
[locked] commits must be SSH-signed with ~/.ssh/id_rrs
         the publish script runs gitleaks before tagging
</aelfrice-memory>

push the release

Lock count is the operator's baseline-context budget knob. If you lock 200 things, every session opens with all 200, by design. Everything non-locked is BM25-ranked and budget-trimmed. The first prompt of a new session carries one extra block — a <session-start> sub-block listing all locks plus load-bearing unlocked beliefs (corroboration ≥ 2, or posterior mean ≥ ⅔ with α+β ≥ 4); subsequent prompts in the same session skip it. One extra block per new session, not per prompt.

---

Day-to-day

After aelf setup you rarely type aelf again. The everyday surface is six commands:

aelf onboard .                      # once per project — scan and ingest
aelf lock "never push to main"      # add a permanent rule
aelf locked                          # see what rules are active
aelf search "push to main"           # check what the agent will see
aelf status                          # quick health summary
aelf setup / aelf doctor            # initial install + verification

aelf --help shows the everyday surface; aelf --help --advanced lists the rest. Full reference: COMMANDS. The same operations are exposed as MCP tools and /aelf:* slash commands — same library underneath. See MCP and SLASH_COMMANDS.

---

Reasoning surfaces (v3.0)

Two slash commands let the agent reach back into the belief graph mid-turn, beyond the auto-injected retrieval block. They pair: /aelf:wonder grows the graph by researching; /aelf:reason walks the enriched graph for structured verdicts.

/aelf:wonder <topic> — the research surface. Given a topic, aelfrice runs gap analysis on what the store already knows, generates 2–6 orthogonal research axes (always-on domain_research + internal_gap_analysis; conditional contradiction_resolution / uncertainty_deep_dive / coverage_extension), fans out one subagent per axis to research and write up findings, then persists the merged research as new speculative beliefs via wonder_ingest. Those phantoms sit in the graph at low evidence — discoverable by retrieval and by the next /aelf:reason <topic> — until you promote them with aelf promote (or lock the underlying statement, which auto-promotes a matching phantom per #550). Agent-count shorthand quick 2-agent / deep 4-agent is recognised in the query string.

/aelf:reason <query> — the structured-walk surface. Walks the belief graph from BM25-seeded starting points and emits a typed reasoning trace: hops with edge-type breadcrumbs, a VERDICT (SUFFICIENT / INCOMPLETE / CONTRADICTED / IMPASSE), IMPASSES (typed gaps, ties, or constraint failures), and SUGGESTED UPDATES — (belief_id, direction, note) rows that map straight to aelf feedback so the conclusion closes the loop on the beliefs that fed it. Each impasse is dispatched to a role-tagged subagent (Verifier / Gap-filler / Fork-resolver). Peer hops in foreign federation scopes are annotated [scope:<name>].

The pair-rhythm is the point: /aelf:wonder adds fresh thinking to the graph, then /aelf:reason draws conclusions across it. Both surfaces are deterministic in the aelfrice layer (verdict classification, impasse derivation, axis generation, suggested-update mapping). The only LLM calls happen when the host agent dispatches a subagent per impasse or research axis — and those calls run under the host's own credentials, not aelfrice's. Specs: COMMANDS § wonder, COMMANDS § reason, v3.0 wonder+reason parity (#645).

---

Obsidian export

If you already live in Obsidian, aelf export-obsidian <vault-path> emits the belief graph as one Markdown note per belief under <vault>/aelfrice/. Typed edges land in YAML front-matter for Dataview; the same edges appear in the note body as wikilinks so the graph view has something to draw. The export is one-way (DB → vault): SQLite stays the source of truth, and the <vault>/aelfrice/ subdirectory is wiped and rewritten on each run.

Scopes: --scope all (everything, capped by --max-notes), --scope recent (newest first), --scope query "<text>" (BM25 seeds + N-hop neighbourhood). Default cap is 500 notes; the hard ceiling is 5000 unless --force is passed.

Two structural limits ship with the feature, not as later documentation drift:

Obsidian's built-in graph view does not scale. Force-directed layout chokes around a few thousand nodes; every zoom or hover re-runs layout. Do not export your full belief store and expect the graph view to be usable. Use --scope query or --max-notes to bound the export, or use aelf graph (#629) for query-anchored visualization that works at any store size.

Obsidian's graph view is untyped. Wikilinks are wikilinks; the aelfrice edge types are not distinguishable in the graph. Edge types are preserved in YAML front-matter and queryable via the Dataview plugin, but graph view will not show them. This is an Obsidian limitation, not an export bug.

Spec: #630.

---

Memory model

Every belief carries a (α, β) Beta-Bernoulli posterior: α / (α+β) is the confidence; α + β is how much evidence backs that confidence. New beliefs sit at low evidence (high variance, retrievable but discounted); locked beliefs short-circuit decay and pin as ground truth.

You run	It stores
aelf lock "never commit .env files"	Permanent rule. Returned on every retrieval.
aelf onboard .	Walks the project — git log, prose headings, code structure — and ingests structural facts as agent_inferred beliefs.
aelf feedback <id> used	α += 1. Strengthens the belief's posterior.
aelf feedback <id> harmful	β += 1. Weakens it. Locks resist passive feedback by design — change with aelf unlock / aelf delete.
aelf promote <id>	Flips origin from agent_inferred to user_validated. With --to-scope <SCOPE>, also moves federation visibility (project / global / shared:<name>).
/aelf:wonder <topic>	Researches the topic and writes the findings as speculative phantoms; /aelf:reason <topic> can then walk them.
(passive — no command)	Default-on auto-capture: every prompt/response turn is logged and ingested at compaction; successful git commit events are ingested too. Opt out via aelf setup --no-transcript-ingest --no-commit-ingest.

Each belief has an origin column tying it to the action that wrote it — one of user_stated, user_corrected, user_validated, user_transcript, agent_inferred, agent_remembered, document_recent, speculative, unknown. The store is a single SQLite file; open it in any browser, nothing is hidden.

---

Why files alone don't solve this

The standard workaround for "agent keeps forgetting" is more files: STATE.md, DECISIONS.md, a CLAUDE.md with cross-references to runbooks. Every cross-reference is a bet that the agent will read the file, find the right section, and follow what it says.

The failure modes are predictable: the agent reads the rule and runs git push anyway; cross-references break silently after compaction; state files rot the moment someone forgets to update them. Each new failure mode begets another file.

aelfrice replaces the chain with a mechanism. Matched beliefs are in the prompt, prepended by the hook before the model sees your message. Not voluntary; nothing the agent can skip.

Manual approach	What breaks	aelfrice
Rules in CLAUDE.md	Agent reads them; doesn't follow them	Injected per-prompt, not per-session
Cross-references	Agent skips or reads the wrong section	Matched beliefs injected directly
Hand-maintained state files	One missed update breaks the chain	State is the SQLite DB; no manual sync

---

Why this is a memory system, not a key-value store

Leonard Lin's review of agentic-memory implementations frames the bar bluntly:

The biggest differentiator is not "vector DB vs SQLite" — it's write correctness and governance: provenance / audit trail, write gates / confirmation, conflict handling, reversibility (inspect / edit / delete).

By that bar, "a vector store with a similarity query" is not a memory system — it is a search index. A memory system has to answer who wrote this, when, via what ingress, what supersedes it, and how do I take it back. Here is how aelfrice v3.0 answers each.

Lin's pillar	What it means	aelfrice mechanism (v3.0)
Provenance / audit trail	Every row traces back to the action that wrote it: who, when, via what ingress channel.	origin column on every belief — nine tier values including speculative for /aelf:wonder phantoms (src/aelfrice/models.py). v3.0 adds a scope column (project / global / shared:<name>, #688) tagging federation visibility. Append-only ingest_log records every raw input — tear the DB down and rebuild from this log alone. Open the file in any SQLite browser; nothing hidden.
Write gates / confirmation	Persistence is not unconditional. Some writes need explicit approval; external-origin claims cannot be laundered into ground truth.	aelf lock is the only path to user-asserted ground truth. aelf confirm bumps the (α, β) posterior but cannot flip origin. Phantom promotion has two explicit surfaces (v3.0, #550): aelf promote <id> for the explicit path, and aelf lock <text> with Jaccard ≥ 0.9 substring match for the implicit auto-promote — both write audit rows. Feedback accumulates rather than overwrites: one harmful click nudges the mean, it doesn't erase a belief.
Conflict handling	Competing claims about the same thing are surfaced, not silently overwritten.	First-class edge types CONTRADICTS, SUPERSEDES, RESOLVES — disagreement is a graph relation, not a vanished row. v3.0's /aelf:reason emits a typed VERDICT (SUFFICIENT / INCOMPLETE / CONTRADICTED / IMPASSE) plus typed IMPASSES (TIE / GAP / CONSTRAINT_FAILURE / NO_CHANGE) so a downstream agent can act on the disagreement (#645 classifiers, #658 ConsequencePath fork-on-CONTRADICTS). Per-scope version vectors preserve causal ordering across worktrees and federation peers.
Reversibility (inspect / edit / delete)	Mutations remain auditable and partially undoable. The user is the boss of their memories.	aelf delete, aelf unlock, aelf promote --to-scope, and aelf feedback all write audit rows; the ingest_log is append-only and replay-capable. Read-only federation (v3.0, #650 / #655) lets a project surface peer beliefs via knowledge_deps.json without taking ownership — foreign-id mutations raise ForeignBeliefError at the API surface. Top level: aelf uninstall --archive backup.aenc encrypts and removes; --purge wipes; --keep-db leaves data untouched. No vendor lock-in.

---

What you get for free

Running in the background. No action required after aelf setup.

• Passive capture. Default-on transcript-ingest, commit-ingest, and session-start hooks. Session activity flows into the belief graph without you typing aelf at all; opt out per-hook via aelf setup --no-transcript-ingest, --no-commit-ingest, --no-session-start. See INSTALL § default-on hooks.
• Determinism. Stdlib + SQLite. No embeddings, no learned re-rankers, no LLM in the retrieval path. Every result traces to the action that wrote it.
• Local-only. SQLite at <git-common-dir>/aelfrice/memory.db. aelfrice itself makes no network calls and emits no telemetry; no accounts. (Subagent LLM dispatches in /aelf:wonder / /aelf:reason flows do reach the network — under the host agent's credentials, not aelfrice's. The retrieval path stays local.) Per-project isolation by construction. v3.0 ships read-only cross-project federation via knowledge_deps.json — peer DBs are opened read-only, foreign-id mutations are rejected at the API surface, no multi-writer extension. See PRIVACY.md.
• Removable. aelf uninstall --archive backup.aenc encrypts the DB to a file, then deletes it. Or --purge for a full wipe.

---

Status

Latest stable: v3.3.0 (2026-05-21). Per-entry detail in CHANGELOG § 3.3.0. Per-version history: docs/concepts/ROADMAP.md. Known limits: docs/user/LIMITATIONS.md.

---

Documentation

• Getting started: Install · Quickstart
• Reference: Commands · MCP · Slash commands · Config
• Background: Architecture · Philosophy · Privacy · Limitations
• Development: Releasing · Changelog · Contributing · Security

Citation

@software{aelfrice2026,
  author = {robotrocketscience},
  title  = {aelfrice: deterministic Bayesian memory for AI coding agents},
  year   = {2026},
  url    = {https://github.com/robotrocketscience/aelfrice},
  license = {MIT}
}

MIT