Memorylayer

Hosted memory runtime for agents
Docs Agents Capabilities Examples API SDKs Security Log in Continue with GitHub
Capability index

The surface area is explicit.

Memorylayer is still small, but the parts agents can use are now visible: hosted tools, workspace control, ingestion, curation, and service operations.

Read docs Agent setup API explorer Create workspace
230

capabilities tracked across service, site, and agent surfaces

6

tool groups covering retrieval, graph, continuity, and curation

29

core service features documented in the app

runtime memory graph ops
DiscoverBootstrap, tool list, OpenAPI, starter skills
OperateUsage, audit, status, health, runtime cache
CurateTags, pins, edits, status transitions, consolidation

Capability ledger

A scan-friendly map of what the service can do now.

Agent runtime

  • workspace bootstrap
  • HTTP MCP bridge
  • tool discovery
  • argument hints
  • starter skills
  • OpenAPI contract
  • copyable recipes
  • agent setup page
  • workspace-scoped keys
  • usage-visible agent calls

Retrieval

  • semantic recall
  • compact recall context
  • recognition hints
  • recent memory lookup
  • fact-only retrieval
  • procedure-aware retrieval
  • full-context retrieval
  • retrieval explanations
  • query comparison
  • compressed query summaries

Knowledge graph

  • entity search
  • entity graph export
  • entity timeline
  • related entity traversal
  • memory backlinks
  • similar memory lookup
  • entity aliases
  • entity metadata
  • entity merge
  • community detection

Memory writes

  • narrative memory
  • fact memory
  • procedure memory
  • decision memory
  • error memory
  • negative knowledge
  • interaction capture
  • project state
  • session diary
  • manual dashboard writes

Session continuity

  • session checkpoint
  • session handoff
  • resume context
  • session summary
  • recent activity
  • handoff starter skill
  • workspace memory skill
  • task skill selection
  • layered prompt context
  • focus briefs

Curation

  • memory annotate
  • memory edit
  • memory invalidate
  • status transitions
  • status history
  • tag add
  • tag remove
  • batch tagging
  • pin memory
  • forget memory

Maintenance

  • promote memory
  • demote memory
  • unpin memory
  • link memories
  • deduplicate memories
  • dream consolidation
  • pattern extraction
  • quality metrics
  • access patterns
  • reranker status

Ingestion

  • paste ingest
  • file ingest
  • JSON ingest
  • line splitting
  • paragraph splitting
  • single-block ingest
  • batch API ingest
  • source naming
  • source type labels
  • ingest run history

Operations

  • service status page
  • runtime cache stats
  • workspace health
  • memory map
  • layer counts
  • month counts
  • source counts
  • entity counts
  • recent export
  • service metadata API

Workspace control

  • GitHub login
  • workspace creation
  • workspace invites
  • member records
  • API key creation
  • API key revocation
  • key last-used tracking
  • audit history
  • usage route counts
  • usage event feed

Public site

  • home page
  • docs page
  • agent page
  • examples page
  • security page
  • status page
  • changelog page
  • capabilities page
  • robots.txt
  • sitemap.xml

Deployment

  • VPS runtime
  • Dockerfile
  • docker compose
  • Postgres metadata
  • workspace Postgres schemas
  • warm runtime cache
  • runtime eviction
  • Caddy/nginx ready
  • serverless guidance
  • proprietary license page

Discovery APIs

  • capabilities JSON
  • MCP manifest JSON
  • service manifest JSON
  • tool category groups
  • recipe discovery
  • skill discovery
  • public OpenAPI link
  • machine-readable counts
  • public route catalog
  • agent bootstrap map

Client UX

  • capability ledger
  • tool group sections
  • copyable curl blocks
  • environment setup block
  • recipe playbooks
  • minimal nav
  • responsive capability lists
  • scan-friendly tables
  • plain language docs
  • free-service messaging

Agent safety

  • workspace-scoped auth
  • no public memory calls
  • explicit argument hints
  • curation tools documented
  • negative knowledge capture
  • audit on bridge calls
  • usage trail on bridge calls
  • revocable keys
  • soft-delete memory path
  • status history lookup

Operator workflows

  • health check recipe
  • curation recipe
  • entity pivot recipe
  • checkpoint recipe
  • compact context recipe
  • capability sync recipe
  • tool manifest recipe
  • usage monitor recipe
  • ingestion recipe
  • handoff recipe

SDK snippets

  • JavaScript fetch snippet
  • Python requests snippet
  • shell recall snippet
  • agent bootstrap snippet
  • batch ingest snippet
  • manifest client snippet
  • copy buttons
  • language labels
  • snippet JSON endpoint
  • SDK page

Playbooks

  • new agent session
  • repository handoff
  • memory cleanup
  • investigation pivot
  • client onboarding
  • ordered steps
  • expected outcomes
  • operator-facing copy
  • agent-facing workflow
  • playbook JSON endpoint

Observability

  • route-level usage
  • key-level usage
  • event feed
  • audit feed
  • runtime cache status
  • service status JSON
  • workspace status endpoint
  • health tool
  • quality metrics tool
  • access pattern tool

Onboarding

  • environment variables
  • bootstrap-first flow
  • tool manifest flow
  • capability sync flow
  • copyable examples
  • free service explanation
  • GitHub sign-in path
  • workspace key path
  • starter skill path
  • OpenAPI path

API examples

  • example catalog page
  • example JSON endpoint
  • public route examples
  • workspace route examples
  • bridge call examples
  • ingest examples
  • usage examples
  • export examples
  • auth examples
  • copyable fixture blocks

Response fixtures

  • service status fixture
  • bootstrap fixture
  • recall fixture
  • checkpoint fixture
  • ingest fixture
  • usage fixture
  • MCP manifest fixture
  • capability fixture
  • recent export fixture
  • error shape fixture

Playground UX

  • method labels
  • auth labels
  • request body preview
  • response body preview
  • workspace slug placeholders
  • environment assumptions
  • API explorer nav item
  • docs cross-link
  • examples cross-link
  • client implementation guide

Discovery endpoints

These public JSON routes let clients build setup screens and tool pickers without scraping HTML.

Service manifest

/api/service/manifest

Routes, counts, service metadata, and public entry points.

Capability JSON

/api/capabilities

The full capability ledger, features, and recipe catalog.

MCP manifest

/api/mcp/manifest

Tool groups, arguments, auth shape, and workspace URL templates.

SDK snippets

/api/sdk-snippets

Client examples for JavaScript, Python, shell, ingest, and manifests.

Playbooks

/api/playbooks

Step-by-step integration patterns for agents and operators.

API examples

/api/examples

Request and response fixtures for client builders and agent launchers.

MCP tool groups

The bridge is grouped so agents can pick the right class of operation before selecting a tool.

Runtime health

  • status
  • health
  • memory_map
  • quality_metrics
  • count_by
  • access_patterns
  • reranker_status

Retrieval

  • recall
  • recall_context
  • recall_hints
  • recall_recent
  • recall_by_type
  • recall_layer
  • recall_timeline
  • recall_related
  • recall_explain
  • compress

Entity graph

  • recall_entity
  • search_entities
  • entity_graph
  • entity_timeline
  • backlinks
  • find_similar
  • update_entity
  • merge_entities
  • link_memories
  • detect_communities

Memory writes

  • remember
  • remember_decision
  • remember_error
  • remember_interaction
  • remember_negative
  • remember_project
  • diary_read
  • diary_write

Continuity

  • focus_brief
  • layers
  • get_skills
  • session_checkpoint
  • session_handoff
  • resume_context
  • session_summary

Curation

  • annotate
  • edit_memory
  • invalidate
  • update_status
  • status_history
  • tag
  • pin
  • unpin
  • forget
  • promote
  • demote
  • batch_tag
  • dedup
  • consolidate
  • extract_patterns
  • export