Memory
A policy-carrying long-term memory store for agents: bond one to a triformer and its subconscious automatically learns facts, preferences, procedures, and corrections across conversations — recalled by meaning, by the situation a memory was formed in, and by what you were trying to do at the time. Per-person scoping keeps what one person told the agent from surfacing for anyone else.
Working with it
Selecting a Memory reveals its settings in the properties panel; it has no dedicated full-screen workbench.
How it appears
The same element type rendered as a definition, a circle instance, and a live workspace card.
When to use / not
When to use
- Give an agent durable memory across sessions — bond a private Memory and the subconscious stores/recalls automatically.
- Per-person memory for multi-user agents — enable subject_scoping so each member's, contact's, or guest's facts stay theirs; forget_subject handles right-to-be-forgotten.
- Team knowledge — bond the same Memory to several agents; what one learns, all recall.
- Deliberate memory management — set conscious_access to read or read_write to expose search/store/delete as agent tools.
When not to use
- Document/RAG corpora — vectorize documents into vector or document elements; Memory is for learned conversational knowledge, not ingested content.
- Structured business rows — use sql, contacts, or entity.
- Within-conversation state — the conversation itself carries short-term context; Memory is for what should survive it.
Topology
Created from the library and placed inside an app or circle. It is a top-level building block you compose with other elements.
Properties
descriptionstring- Human-readable description of what this memory store is for
scopestring- Declared sharing intent. `private` — one agent's own memory; `shared` — deliberately bonded by multiple agents (team knowledge). Documentation of intent for the panel and audits; the enforcement mechanism is which agents hold a bond to this element.
subject_scopingboolean- When true, memories are partitioned by WHO they are about: writes classify each memory as person-scoped (subject = the current interlocutor) or general (no subject), and recall only surfaces general memories plus those about the person currently speaking. When false, the subject columns are still populated when known but recall ignores them (everything is general).
conscious_accessstring- Whether the bonded agent's conscious mind gets memory tools. `none` — memory is fully involuntary (subconscious-managed only, the platform default stance). `read` — the agent gets memory_search / recall tools. `read_write` — the agent may also consciously store, update, and delete memories. Conscious writes flow through the same pipeline as subconscious ones (same embedding, categories, association capture, caps).
max_memoriesinteger- Maximum stored memories in this element. At the cap, storing a new memory evicts the least-accessed, oldest memory. Enforced by the executor at store time.
max_per_subjectinteger- Per-subject cap when subject_scoping is on — one chatty interlocutor cannot evict another person's memories or the general pool. Eviction within the subject's own rows only.
relevance_thresholdnumber- Minimum blended similarity score for recall (0 = return everything, 1 = exact match only). Enforced by the executor's search — replaces the legacy hardcoded MIN_VECTOR_SCORE.
association_weightsobject- Blend weights for the three association lanes at recall time. score = content·sim(query, memory content) + context·sim(current conversation window, memory's stored context) + intention·sim(current turn goal, memory's stored intention). Context and intention are captured at STORE time, so a memory can be found by the situation it was formed in even when its text shares nothing with the query. Weights are normalized if they don't sum to 1.
retentionobject- Optional per-category TTL (days). Memories older than the TTL are removed by the maintenance reaper (stats op / background sweep). Categories without an entry never expire. Sensible use: let `context` decay while `fact`/`procedure` persist.
Operations
- activityGET
- attachmentsGET
- batch_statsGET
- composePOST
- contextGET
- createPOST
- deletePOST
- disablePOST
- enablePOST
- export_bundleGET
- forget_subjectPOST
- getGET
- import_bundlePOST
- intentionGET
- listGET
- promotePOST
- readmeGET
- readme_updatePOST
- remove-modifierPOST
- restorePOST
- schemaGET
- searchPOST
- sourceGET
- source_branchesGET
- source_diff3GET
- source_fixturesPOST
- source_mergePOST
- source_promotePOST
- source_repairPOST
- source_statusGET
- source_validatePOST
- statsGET
- storePOST
- treeGET
- updatePOST
- update_metaPATCH
- versionGET
Composition
Errors / when it fails
- max_per_subject cannot exceed max_memories
- Fails unless:
spec.max_per_subject <= spec.max_memories
Validation rules
- conscious_access=read_write lets the agent consciously store and delete its own memories — the subconscious evaluator remains active and may update or remove conscious writes. Intended for agents that need deliberate memory management; the involuntary default is `none`.
- subject_scoping is off — all memories are treated as general at recall time. Subject columns are still populated when the interlocutor is known, so scoping can be enabled later without losing attribution.
Memory (memory)
Category: data | Form: | Symbol: Me
Durable agent memory — learned facts, preferences, and context that survive across conversations
Long-term memory store for agents. Bond one to a triformer (subconscious.memory.element) and its subconscious automatically stores, updates, and forgets memories; recall blends content, conversation-context, and intention similarity. Supports per-person scoping (subject_scoping) so facts about one person never surface for another, and configurable conscious access so the agent itself can search or manage its own memory as tools. Two agents bonded to the same Memory share knowledge.
Relationships
- Attaches to: rate-limit, auth-policy
Capabilities
- durable-memory: Cross-session memory storage with categories (fact, preference, procedure, context, correction) and automatic eviction at capacity
- blended-association-recall: Recall scored across three lanes — content similarity, conversation-context similarity, and intention similarity — with element-configured weights
- subject-scoping: Per-person memory partitioning: memories about one interlocutor (circle member, entity, or guest) never surface for another; includes forget_subject right-to-be-forgotten
- conscious-access: Configurable exposure of memory ops as agent tools (none / read / read_write) — the involuntary-subconscious default can be selectively relaxed
- shared-memory: Multiple agents bonding the same element share its memories — team knowledge without a separate mechanism
- retention: Optional per-category TTL reaping
Properties
| Property | Type | Default | Description |
|---|---|---|---|
description | string | — | Human-readable description of what this memory store is for |
scope | string | "private" | Declared sharing intent. private — one agent’s own memory; shared — deliberately bonded by multiple agents (team knowledge). Documentation of intent for the panel and audits; the enforcement mechanism is which agents hold a bond to this element. |
subject_scoping | boolean | false | When true, memories are partitioned by WHO they are about: writes classify each memory as person-scoped (subject = the current interlocutor) or general (no subject), and recall only surfaces general memories plus those about the person currently speaking. When false, the subject columns are still populated when known but recall ignores them (everything is general). |
conscious_access | string | "none" | Whether the bonded agent’s conscious mind gets memory tools. none — memory is fully involuntary (subconscious-managed only, the platform default stance). read — the agent gets memory_search / recall tools. read_write — the agent may also consciously store, update, and delete memories. Conscious writes flow through the same pipeline as subconscious ones (same embedding, categories, association capture, caps). |
max_memories | integer | 1000 | Maximum stored memories in this element. At the cap, storing a new memory evicts the least-accessed, oldest memory. Enforced by the executor at store time. |
max_per_subject | integer | 200 | Per-subject cap when subject_scoping is on — one chatty interlocutor cannot evict another person’s memories or the general pool. Eviction within the subject’s own rows only. |
relevance_threshold | number | 0.4 | Minimum blended similarity score for recall (0 = return everything, 1 = exact match only). Enforced by the executor’s search — replaces the legacy hardcoded MIN_VECTOR_SCORE. |
association_weights | object | — | Blend weights for the three association lanes at recall time. score = content·sim(query, memory content) + context·sim(current conversation window, memory’s stored context) + intention·sim(current turn goal, memory’s stored intention). Context and intention are captured at STORE time, so a memory can be found by the situation it was formed in even when its text shares nothing with the query. Weights are normalized if they don’t sum to 1. |
retention | object | — | Optional per-category TTL (days). Memories older than the TTL are removed by the maintenance reaper (stats op / background sweep). Categories without an entry never expire. Sensible use: let context decay while fact/procedure persist. |
Operations
activity
Get /ops/activity | Auth: Read
Get activity events for this element
Scope depends on element capabilities: individual elements query by element_id, project-form elements with activity-scope-members include member activities, circle-level elements with activity-scope-all query the entire circle. Gracefully returns empty list if activities table is missing (old circles).
attachments
Get /ops/attachments | Auth: Read
List all modifiers and resources attached to this element
Returns both modifiers (policy enforcement) and resources (data injection) with is_modifier flag to distinguish. Items in the generated MODIFIER_TYPES list are modifiers; everything else is a resource. Includes cascade_policy and version pin info.
batch_stats
Get /ops/batch_stats | Auth: Read
Get per-element statistics for all children of this element
Returns per-child stats plus an aggregate. Most meaningful on compound or manifest form elements (repositories, circles, projects); atoms have no children so the result is an empty children array with a zeroed aggregate. Uses efficient GROUP BY SQL. Weighted averages for eval scores.
compose
Post /ops/compose | Auth: Execute
Batch add and remove modifiers on this element in a single call
Declarative composition: add modifiers by ref path (slug or path@version) and remove by attachment ID, all in one atomic call on the target element. Each ‘add’ entry resolves the source element, validates topology, attaches with optional priority and cascade policy. Each ‘remove’ entry deletes the attachment row. Returns a summary of what was added and removed. Example: compose({ add: [{ref: “my-prompt”}, {ref: “rate-limit/api@v2”, priority: 50}], remove: [{attachment_id: “uuid”}] })
context
Get /ops/context | Auth: Read
Get connected elements (graph traversal)
Graph traversal showing all connected elements with their relationship type (contains, contained_by, references, referenced_by, attaches, etc.). Use ?depth=N to control traversal depth (default 1) and ?types=actor,data to filter by element types.
create
Post /ops/create | Auth: Write
Create child element
POST to the parent path — element_type goes in the request body, NOT the URL. Both element_type and slug are required and must be non-empty. Name is derived from slug if omitted. Writes to both Git and PostgreSQL. All elements are stored flat under the circle — no intermediate library wrapper rows.
delete
Post /ops/memories/{memory_id}/delete | Auth: Write
Delete a specific memory
Permanently removes one memory. Administrative or subconscious-evaluator action; with conscious_access read_write the agent itself may call it.
disable
Post /ops/disable | Auth: Admin
Disable element (hides and prevents use)
Idempotent — safe to call on already-disabled elements. Optionally pass a reason string. Disabled elements cannot be invoked or executed. Inverse of enable.
enable
Post /ops/enable | Auth: Admin
Enable element (makes usable and visible)
Idempotent — safe to call on already-enabled elements. Transitions element to ready/enabled state. Cannot enable deleted elements. Inverse of disable.
export_bundle
Get /ops/export/bundle | Auth: Read
Export element as downloadable git bundle
On non-root-namespace elements, returns a binary git bundle. On root-namespace (circle) elements, dispatch hands off to the circle’s own export_bundle op, which returns a multi-element JSON envelope with one base64 bundle per child element — this is intentional, not an error.
forget_subject
Post /ops/forget_subject | Auth: Write
Delete ALL memories about a subject (right-to-be-forgotten)
Permanently deletes every memory whose subject matches — the right-to-be-forgotten surface. Destructive and irreversible; requires the X-Confirm: destructive header. General memories are untouched.
get
Get /ops/memories/{memory_id} | Auth: Read
Get a specific memory with metadata and association info
import_bundle
Post /ops/import/bundle | Auth: Write
Import git bundle into element
Accepts a base64-encoded git bundle in the JSON bundle_base64 field. Use overwrite=true to replace existing elements with same slug (default skips duplicates). Imported elements get new UUIDs. Returns counts of imported/skipped elements and any errors.
intention
Get /ops/intention | Auth: Read
Get element intention with full inheritance chain
Returns three levels: direct (this element’s intention), inherited (from category and root), and resolved (final merged intention). Useful for understanding an element’s purpose in context of its hierarchy.
list
Get /ops/memories | Auth: Read
List memories with filters
promote
Post /ops/promote | Auth: Admin
Promote element configuration to a target environment
Only for manifest-form elements (projects). Environments advance: dev → demo → live. dev→demo requires member+ role, demo→live requires admin. Freezes member versions at promotion time (creates snapshot). Persists environment config to spec.environments.
readme
Get /ops/readme | Auth: Read
Get element README.md content
Reads README.md from the element’s git repository. Returns empty content (not an error) if no README exists. Always returns markdown format.
readme_update
Post /ops/readme_update | Auth: Write
Update element README.md content
Creates or overwrites README.md in the element’s git repo. Commits to the draft branch. Content must be provided as a markdown string.
remove-modifier
Post /ops/remove-modifier | Auth: Execute
Remove an attached modifier from this element by attachment ID
Removes a modifier/resource attachment by its row ID. The ID comes from the attachments or context API. This is the reverse of attach — called on the target element, not the source.
restore
Post /ops/restore | Auth: Admin
Restore element to a specific version
Automatically snapshots the current state before restoring (creates a ‘Before restore to vN’ version entry). Writes restored spec to git as .triform/spec.yaml. Git failures warn but don’t fail the operation — DB state is authoritative. Cannot restore deleted elements.
schema
Get /ops/schema | Auth: Read
Get element input/output schema (MCP tools/list compatible)
Returns type-level port schemas from the TypeRegistry — not instance-specific overrides. Includes direction (input/output), required flag, and JSON schema per port. Useful for understanding what data an element accepts and produces.
search
Post /ops/search | Auth: Read
Blended-association memory recall (content + context + intention similarity)
Recall memories by meaning, situation, and purpose. The query text is embedded and scored against each memory’s content vector, plus (when provided) the current conversation window against stored context vectors and the current goal against stored intention vectors — weights come from this element’s association_weights. When subject_scoping is on, results are general memories plus those about the given subject only. Results below relevance_threshold are dropped.
source
Get /ops/source | Auth: Read
Get any file’s content from the element’s git repository
Reads an arbitrary file from the element’s CAS-backed git tree by its relative path. Same store as
readme, just generalized. Path safety: rejects..traversal, leading/, and null bytes. Use this to viewmain.pyfor action elements, asset files for SPAs, etc. Returns empty content (not an error) if the file doesn’t exist.
source_branches
Get /ops/source/branches | Auth: Read
List Source branches for this element
Returns the standard draft/demo/live Source branches, their current commits, and promotion relationships. Use GET /api/{element_path}/ops/source/branches.
source_diff3
Get /ops/source/diff3 | Auth: Read
Preview a three-way merge of two commits without writing
Read-only. Computes what a merge of
theirs(default: your base_commit) intoours(default: the draft tip) would produce: per-file conflicts with ours/theirs/base content, plus clean/fast_forward flags. Use after a 409 to see exactly what moved before re-splicing your change. GET /api/{element_path}/ops/source/diff3?theirs={sha}.
source_fixtures
Post /ops/source/fixtures | Auth: Write
Dry-run or apply approved Source seed fixtures
Scans
.triform/fixtures/manifests from the addressed data element Source repo. Defaults to dry_run=true and never imports live runtime data. Apply requires dry_run=false plus confirm=true and dispatches approved records through existing generated element ops.
source_merge
Post /ops/source/merge | Auth: Write
Three-way merge a commit into the draft branch
Merges
theirs(default: your base_commit) into the draft branch. Clean (non-overlapping edits) → commits and advances draft, returns merged: true + the new commit. Conflicts → writes nothing and returns the conflict set to resolve and retry. Refused inside an active version-set scope (use version_sets_merge). POST /api/{element_path}/ops/source/merge {“theirs”: “{sha}”}.
source_promote
Post /ops/source/promote | Auth: Write
Promote Source branch forward
Promotes draft to demo or demo to live through the generated element op path. Direct Git pushes to demo/live are blocked by Source policy.
source_repair
Post /ops/source/repair | Auth: Write
Inspect or repair the element Source index
Runs Source repair through the element operation path. Defaults to dry_run=true; set dry_run=false only after reviewing a dry-run report.
source_status
Get /ops/source/status | Auth: Read
Get Source control status for this element
Returns the branch-aware clone URL, checkout commands, current draft commit, child source-link count, portable export summary, Source health, warnings, and auth hints for the addressed element. Use the element-first path: GET /api/{element_path}/ops/source/status.
source_validate
Post /ops/source/validate | Auth: Read
Validate Source branch contents
Validates a Source branch before accepting local Git workflow changes or promotion. Defaults to branch=draft and rejects runtime data, generated output, secret material, and unreadable CAS refs.
stats
Get /ops/stats | Auth: Read
Memory store statistics and maintenance
Returns counts by category and subject plus capacity info. Also runs the retention reaper: memories older than their category’s configured TTL are removed and reported in
expired.
store
Post /ops/store | Auth: Write
Store a new memory
Stores a durable memory with content, category, and optional subject (who it is about) and association texts (the situation and goal it was formed in — embedded and stored for associative recall). At max_memories the least-accessed oldest memory is evicted; with subject_scoping on, per-subject eviction respects max_per_subject. Normally called by the agent’s subconscious evaluator; direct calls are for pipelines/automations and conscious read_write agents.
tree
Get /ops/tree | Auth: Read
Get the element’s position in the graph — ancestors, children, references, and subtree statistics
Uses per-circle ElementGraph cache for O(1) lookups. Returns ancestors (containment chain), children (direct), members (references), referenced_by (reverse refs), attachments, and subtree stats. Default depth is 3, max is 10. Pass ?include_metadata=true for name/state on each node.
update
Post /ops/memories/{memory_id}/update | Auth: Write
Update a memory’s content (re-embeds)
Replaces the memory’s content and recomputes its content embedding. Category, subject, and association vectors are preserved unless explicitly changed. Used by the subconscious evaluator when new information supersedes an existing memory.
update_meta
Patch /ops/update_meta | Auth: Write
Update element metadata (lightweight merge — does NOT bump version or snapshot spec)
Shallow JSONB merge into element.meta. Top-level keys in the provided value replace existing meta values; other keys are preserved. Used for UI metadata like canvas positions, panel state, viewer preferences. Wire-shape op_name is
update_meta(distinct fromupdate) so SSE subscribers + the cache auto-invalidator can distinguish lightweight metadata changes from spec edits without inspecting the payload. The MutatingElementStore wrapper stamps this op_name on the lifecycle event emitted byupdate_element_metastorage calls.
version
Get /ops/version | Auth: Read
Get current version or full history
Returns current version by default. Pass ?history=true for full version history (up to ?limit=N, default 50). Versions are backed by the element_versions table. Every spec update creates a new version entry.
Error Codes
| Code | Class | Retryable | Description |
|---|---|---|---|
MEMORY_NOT_FOUND | not_found | no | memory_id does not exist in this element |
MEMORY_INVALID_CATEGORY | validation | no | category is not one of fact, preference, procedure, context, correction |
MEMORY_SUBJECT_INCOMPLETE | validation | no | subject_type and subject_id must be provided together |
MEMORY_CONFIRM_REQUIRED | validation | no | forget_subject requires the X-Confirm: destructive header |
MEMORY_EMBEDDING_FAILED | internal | yes | Embedding generation failed — memory stored without vectors, text-search fallback only (store) or ILIKE fallback used (search) |
MEMORY_STORE_FAILED | internal | yes | Database write failed — transient backend error |
Observability
Defined for this element
Metrics
- memory_store_count
- memory_recall_count
- memory_recall_latency_ms
- memory_eviction_count
- memory_forget_subject_count
- memory_retention_expired_count
Pricing / cost
Platform default
Operation costs
- create: free
- update: free
- delete: free
- get: free
- list: free
- invoke: 10000 micro-AU
- tool_use: free