# Engram

## Docs

- [Create agent](https://docs.hakuya.ai/api-reference/agents/create-agent.md)
- [Delete agent](https://docs.hakuya.ai/api-reference/agents/delete-agent.md): Deletes the agent and all its associated memories, episodes, procedures, and schemas.
- [Get agent](https://docs.hakuya.ai/api-reference/agents/get-agent.md)
- [Get agent mind](https://docs.hakuya.ai/api-reference/agents/get-agent-mind.md): Returns the agent's full mental state — semantic memories, episodes, procedures, and schemas in one call.
- [Get hot-tier memories](https://docs.hakuya.ai/api-reference/agents/get-hot-tier-memories.md): Returns memories with confidence > 0.85. Use for prompt injection at session start.
- [Get learning statistics](https://docs.hakuya.ai/api-reference/agents/get-learning-statistics.md)
- [Get memory policies](https://docs.hakuya.ai/api-reference/agents/get-memory-policies.md)
- [Get tier statistics](https://docs.hakuya.ai/api-reference/agents/get-tier-statistics.md)
- [List agents](https://docs.hakuya.ai/api-reference/agents/list-agents.md)
- [Update memory policies](https://docs.hakuya.ai/api-reference/agents/update-memory-policies.md)
- [Bootstrap tenant](https://docs.hakuya.ai/api-reference/bootstrap-tenant.md): Creates a new tenant and returns a master API key. Protected by X-Setup-Token header.
- [Activate working memory](https://docs.hakuya.ai/api-reference/cognitive/activate-working-memory.md): Primes working memory for a task by retrieving and activating relevant memories.
- [Assess confidence on topic](https://docs.hakuya.ai/api-reference/cognitive/assess-confidence-on-topic.md)
- [Clear working memory session](https://docs.hakuya.ai/api-reference/cognitive/clear-working-memory-session.md)
- [Detect uncertainty](https://docs.hakuya.ai/api-reference/cognitive/detect-uncertainty.md): Surfaces areas of low confidence, active contradictions, and knowledge gaps.
- [Get confidence stats for a memory](https://docs.hakuya.ai/api-reference/cognitive/get-confidence-stats-for-a-memory.md)
- [Get memory health](https://docs.hakuya.ai/api-reference/cognitive/get-memory-health.md): Returns aggregate health statistics — confidence distribution, contradiction count, staleness, learning velocity.
- [Get working memory session](https://docs.hakuya.ai/api-reference/cognitive/get-working-memory-session.md)
- [Manually penalise a memory](https://docs.hakuya.ai/api-reference/cognitive/manually-penalise-a-memory.md)
- [Manually reinforce a memory](https://docs.hakuya.ai/api-reference/cognitive/manually-reinforce-a-memory.md)
- [Metacognitive reflection](https://docs.hakuya.ai/api-reference/cognitive/metacognitive-reflection.md): Triggers self-assessment of the agent's knowledge state.
- [Trigger consolidation](https://docs.hakuya.ai/api-reference/cognitive/trigger-consolidation.md): Runs the consolidation pipeline — episodes → beliefs → procedures → schemas.
- [Trigger memory decay](https://docs.hakuya.ai/api-reference/cognitive/trigger-memory-decay.md): Applies confidence decay to all memories for an agent. Archives memories that fall below the threshold.
- [Update working memory goal](https://docs.hakuya.ai/api-reference/cognitive/update-working-memory-goal.md)
- [Create episode](https://docs.hakuya.ai/api-reference/episodes/create-episode.md)
- [Get episode](https://docs.hakuya.ai/api-reference/episodes/get-episode.md)
- [Get episode associations](https://docs.hakuya.ai/api-reference/episodes/get-episode-associations.md)
- [Recall episodes](https://docs.hakuya.ai/api-reference/episodes/recall-episodes.md)
- [Record episode outcome](https://docs.hakuya.ai/api-reference/episodes/record-episode-outcome.md)
- [Get relationships](https://docs.hakuya.ai/api-reference/graph/get-relationships.md)
- [List entities](https://docs.hakuya.ai/api-reference/graph/list-entities.md)
- [Traverse graph](https://docs.hakuya.ai/api-reference/graph/traverse-graph.md)
- [Health check](https://docs.hakuya.ai/api-reference/health-check.md)
- [Create API key](https://docs.hakuya.ai/api-reference/keys/create-api-key.md)
- [List API keys](https://docs.hakuya.ai/api-reference/keys/list-api-keys.md)
- [Revoke API key](https://docs.hakuya.ai/api-reference/keys/revoke-api-key.md)
- [Detect implicit feedback](https://docs.hakuya.ai/api-reference/learning/detect-implicit-feedback.md): Analyses a conversation for implicit positive/negative signals about existing memories.
- [Record explicit feedback](https://docs.hakuya.ai/api-reference/learning/record-explicit-feedback.md)
- [Record learning outcome](https://docs.hakuya.ai/api-reference/learning/record-learning-outcome.md)
- [Delete memory](https://docs.hakuya.ai/api-reference/memories/delete-memory.md)
- [Extract memories from conversation](https://docs.hakuya.ai/api-reference/memories/extract-memories-from-conversation.md): Uses the LLM to extract and optionally store discrete memories from a conversation. Requires LLM provider.
- [Get memory](https://docs.hakuya.ai/api-reference/memories/get-memory.md)
- [Get mutation history](https://docs.hakuya.ai/api-reference/memories/get-mutation-history.md): Returns the full audit trail of confidence changes for a memory.
- [Recall memories](https://docs.hakuya.ai/api-reference/memories/recall-memories.md): Hybrid vector + graph recall. Returns the most relevant memories for the query.
- [Restore archived memory](https://docs.hakuya.ai/api-reference/memories/restore-archived-memory.md)
- [Store memory](https://docs.hakuya.ai/api-reference/memories/store-memory.md)
- [Get procedure](https://docs.hakuya.ai/api-reference/procedures/get-procedure.md)
- [Learn procedure from episode](https://docs.hakuya.ai/api-reference/procedures/learn-procedure-from-episode.md)
- [Match procedures](https://docs.hakuya.ai/api-reference/procedures/match-procedures.md): Finds procedural memories whose trigger pattern matches the input.
- [Record procedure outcome](https://docs.hakuya.ai/api-reference/procedures/record-procedure-outcome.md)
- [Contradict schema](https://docs.hakuya.ai/api-reference/schemas/contradict-schema.md)
- [Delete schema](https://docs.hakuya.ai/api-reference/schemas/delete-schema.md)
- [Detect schema patterns](https://docs.hakuya.ai/api-reference/schemas/detect-schema-patterns.md): Analyses the agent's memories and derives higher-order mental models.
- [Get schema](https://docs.hakuya.ai/api-reference/schemas/get-schema.md)
- [List schemas](https://docs.hakuya.ai/api-reference/schemas/list-schemas.md)
- [Match schemas](https://docs.hakuya.ai/api-reference/schemas/match-schemas.md)
- [Validate schema](https://docs.hakuya.ai/api-reference/schemas/validate-schema.md)
- [Benchmarks](https://docs.hakuya.ai/benchmarks.md): How Engram performs on LongMemEval — the ICLR 2025 benchmark for long-term conversational memory. Full per-task results and methodology.
- [Confidence Scoring](https://docs.hakuya.ai/concepts/confidence.md): How Engram tracks and updates belief confidence using log-odds arithmetic.
- [Contradiction Detection](https://docs.hakuya.ai/concepts/contradiction.md): How Engram handles conflicting beliefs — demoting, archiving, or allowing coexistence.
- [Graph Retrieval](https://docs.hakuya.ai/concepts/graph-retrieval.md): How Engram combines vector similarity with graph traversal for more accurate recall.
- [Memory Lifecycle](https://docs.hakuya.ai/concepts/lifecycle.md): How memories decay, consolidate, and mature over time.
- [Memory Types](https://docs.hakuya.ai/concepts/memory-types.md): Engram implements four cognitive memory types modelled on human memory science.
- [Subjects, Sessions & Canon](https://docs.hakuya.ai/concepts/scopes.md): Give every memory something to hold onto — who it's about (anchor), which conversation it came from (session), and what's shared across the whole org (canon). One agent, many isolated subjects.
- [LangChain Integration](https://docs.hakuya.ai/guides/langchain.md): Use Engram as a production-grade memory backend for your LangChain agents.
- [MCP Server](https://docs.hakuya.ai/guides/mcp.md): Use Engram as a memory tool in Claude Desktop, Cursor, Windsurf, and any MCP-compatible host.
- [Installation](https://docs.hakuya.ai/installation.md): Install Engram via Docker or build from source.
- [Introduction](https://docs.hakuya.ai/introduction.md): Memory infrastructure for AI agents — with calibrated confidence, contradiction detection, and lifecycle management.
- [Quickstart](https://docs.hakuya.ai/quickstart.md): Store your first memory and run your first recall in 5 minutes.
- [Python SDK](https://docs.hakuya.ai/sdk/overview.md): The official Python SDK for Engram. Sync and async, full type safety, Pydantic v2.
- [Docker Compose](https://docs.hakuya.ai/self-hosting/docker.md): Run Engram on your own infrastructure in one command.
- [Production Deployment](https://docs.hakuya.ai/self-hosting/production.md): Hardening checklist, connection pooling, monitoring, and HA Postgres for production Engram deployments.

## OpenAPI Specs

- [openapi](https://docs.hakuya.ai/openapi.yaml)