Synapse Layer — Trust Infrastructure for AI Agents

Measured conclusion: Deep-tested. L2 outcome: tools_ok.

MCP-native Trust Infrastructure for AI Agents. Persistent encrypted memory with Trust Quotient.

Add to your agent

Copy this into your MCP client config:

{
  "mcpServers": {
    "Synapse Layer — Trust Infrastructure for AI Agents": {
      "url": "https://forge.synapselayer.org/api/mcp",
      "transport": "streamable-http"
    }
  }
}

Measured tools (13)

• recall read-only

  Retrieve relevant persisted memory using semantic, temporal, priority, or hybrid routing. Governance: requires reason (10–200 chars). Rate limit: 20/min.

  Parameters:

  • query (required) — string — What to recall — natural language query for memory retrieval.
  • reason (required) — string — REQUIRED (10–200 chars). Audit policy: MEMORY_EXPORT_GOVERNANCE_V1. Human-readable justification for retrieving PLAINTEXT memory. Without this parameter the call WILL fail with code -32602. Example: "User asked me to summarize last week decisions."
  • mode — string — Recall routing mode. "temporal" = pure chronological, "semantic" = FTS + TQ ranking, "priority" = critical/high-TQ first, "hybrid" = weighted blend, "auto" = detect from query (default).
  • limit — number — Maximum memories to return (1–50, default: 10).
  • scope — string — Explicit memory scope. "agent" (default, fail-closed) = limits recall to the given agent_id (or the token agent). "tenant" = cross-agent recall within the authenticated tenant. If omitted with agent_id present, "agent" is inferred; if both omitted, the token agent is used (verified-token scope).
  • subject — string — ESCOPO1 (optional): opaque 8-128 char token ([A-Za-z0-9._:-]) that isolates memory to a single person within a shared connect-token. Omit for token-scoped behavior. Never include PII/email.
  • agent_id — string — Agent identifier (canonical or alias). Used only when scope="agent" or scope is inferred.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world no

• save_to_synapse

  Persist memory with encryption at rest, sanitization, and deduplication controls.

  Parameters:

  • content (required) — string — The memory content to store securely.
  • tags — array — Tags for categorization.
  • type — string — Event type: [MILESTONE], [DECISION], [ALERT], [AUTO-STRAT], [AUTO-OP], [AUTO-INSIGHT], [AUTO-DECISION], [AUTO-CONTEXT], [MANUAL].
  • project — string — Project identifier (e.g., SYNAPSE_LAYER).
  • subject — string — ESCOPO1 (optional): opaque 8-128 char token ([A-Za-z0-9._:-]) that isolates memory to a single person within a shared connect-token. Omit for token-scoped behavior. Never include PII/email.
  • agent_id — string — Agent identifier for memory isolation. Defaults to "default".
  • importance — number — Accepted for compatibility; not enforced by the API.
  • force_intent — string — MANUAL OVERRIDE ONLY — use apenas quando você (humano) quer forçar explicitamente a classificação de intenção. Valores reconhecidos: preference | fact | procedural | bio | critical | operational | strategic | compliance | security | system_directive | clinical | general | ephemeral. Se não tiver certeza, NÃO preencha este campo — o sistema classifica automaticamente. Valores não reconhecidos são ignorados silenciosamente e retornados em warnings.
  • intent_reason — string — Auditable reason for forcing intent classification. Max 500 characters. Ignored if force_intent is absent.

  Hints: read-only no · destructive not declared · idempotent yes · open-world no

• process_text

  Extract candidate memories from free-form text with governance filters and sanitization.

  Parameters:

  • text (required) — string — Free-form text to scan for auto-save triggers.
  • source — string — Source identifier (default: mcp).
  • project — string — Force a specific project namespace. Auto-detected if omitted.
  • agent_id — string — Agent identifier. Defaults to "default".

  Hints: read-only no · destructive not declared · idempotent not declared · open-world no

• search read-only

  Search persisted memory across agent scopes using full-text matching. Governance: requires reason (10–200 chars) because results contain plaintext content.

  Parameters:

  • query (required) — string — Search query — natural language or keywords.
  • reason (required) — string — REQUIRED (10–200 chars). Audit policy: MEMORY_EXPORT_GOVERNANCE_V1. Human-readable justification for retrieving PLAINTEXT memory. Without this parameter the call WILL fail with code -32602. Example: "Checking prior migration decisions for the user."
  • limit — number — Maximum results to return (1–50, default: 20).
  • scope — string — Explicit search scope. "agent" (default, fail-closed) = limits search to the given agent_id (or the token agent). "tenant" = cross-agent search within the authenticated tenant. If omitted with agent_id present, "agent" is inferred; if both omitted, the token agent is used (verified-token scope).
  • subject — string — ESCOPO1 (optional): opaque 8-128 char token ([A-Za-z0-9._:-]) that isolates memory to a single person within a shared connect-token. Omit for token-scoped behavior. Never include PII/email.
  • agent_id — string — Agent identifier (canonical or alias). Used only when scope="agent" or scope is inferred.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world no

• health_check read-only

  Check service availability, engine version, and storage health. Public method.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world no

• initialize_context read-only

  Initialize a persistent memory context for a conversation or agent session. Returns REDACTED metadata by default (detail="summary"). Use detail="full" with a reason (10–200 chars) to retrieve plaintext content under MEMORY_EXPORT_GOVERNANCE_V1.

  Parameters:

  • detail — string — Detail level. "summary" (default) = redacted metadata only, no plaintext. "full" = plaintext content; REQUIRES the reason parameter and is audit-logged.
  • reason — string — REQUIRED only when detail="full" (10–200 chars). Audit policy: MEMORY_EXPORT_GOVERNANCE_V1. Human-readable justification for retrieving PLAINTEXT context. Without it, detail="full" WILL fail with code -32602.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world no

• save_memory

  Save a memory entry to the persistent store. Alias of save_to_synapse.

  Parameters:

  • content (required) — string — The memory content to store securely.
  • tags — array — Tags for categorization.
  • subject — string — ESCOPO1 (optional): opaque 8-128 char token ([A-Za-z0-9._:-]) that isolates memory to a single person within a shared connect-token. Omit for token-scoped behavior. Never include PII/email.
  • agent_id — string — Agent identifier for memory isolation.
  • force_intent — string — MANUAL OVERRIDE ONLY — use apenas quando você (humano) quer forçar explicitamente a classificação de intenção. Valores reconhecidos: preference | fact | procedural | bio | critical | operational | strategic | compliance | security | system_directive | clinical | general | ephemeral. Se não tiver certeza, NÃO preencha este campo — o sistema classifica automaticamente. Valores não reconhecidos são ignorados silenciosamente e retornados em warnings.
  • intent_reason — string — Auditable reason for forcing intent classification. Max 500 characters. Ignored if force_intent is absent.

  Hints: read-only no · destructive not declared · idempotent yes · open-world no

• store_memory

  Store structured memory with metadata and trust scoring.

  Parameters:

  • content (required) — string — The memory content to store securely.
  • tags — array — Tags for categorization.
  • subject — string — ESCOPO1 (optional): opaque 8-128 char token ([A-Za-z0-9._:-]) that isolates memory to a single person within a shared connect-token. Omit for token-scoped behavior. Never include PII/email.
  • agent_id — string — Agent identifier for memory isolation. Defaults to "default".
  • force_intent — string — MANUAL OVERRIDE ONLY — use apenas quando você (humano) quer forçar explicitamente a classificação de intenção. Valores reconhecidos: preference | fact | procedural | bio | critical | operational | strategic | compliance | security | system_directive | clinical | general | ephemeral. Se não tiver certeza, NÃO preencha este campo — o sistema classifica automaticamente. Valores não reconhecidos são ignorados silenciosamente e retornados em warnings.
  • intent_reason — string — Auditable reason for forcing intent classification. Max 500 characters. Ignored if force_intent is absent.

  Hints: read-only no · destructive not declared · idempotent yes · open-world no

• recall_memory read-only

  Recall persisted memory by query. Alias of recall.

  Parameters:

  • query (required) — string — What to recall — natural language query for memory retrieval.
  • reason (required) — string — REQUIRED (10–200 chars). Audit policy: MEMORY_EXPORT_GOVERNANCE_V1. Without this the call WILL fail with code -32602. Example: "Checking user project status."
  • mode — string — Recall routing mode.
  • limit — number — Maximum memories to return (1–50, default: 10).
  • scope — string — Explicit memory scope. "agent" (default, fail-closed) = limits recall to the given agent_id (or the token agent). "tenant" = cross-agent recall within the authenticated tenant. If omitted with agent_id present, "agent" is inferred; if both omitted, the token agent is used (verified-token scope).
  • subject — string — ESCOPO1 (optional): opaque 8-128 char token ([A-Za-z0-9._:-]) that isolates memory to a single person within a shared connect-token. Omit for token-scoped behavior. Never include PII/email.
  • agent_id — string — Agent identifier (canonical or alias). Used only when scope="agent" or scope is inferred.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world no

• list_memories read-only

  List memory metadata with pagination and governance limits.

  Parameters:

  • limit — number — Maximum memory metadata rows to return. Hard-capped: default 5, max 10. Values above 10 are silently reduced to 10.
  • scope — string — Explicit memory scope. "agent" (default, fail-closed) = lists memories for the given agent_id (or the token agent). "tenant" = lists memories across all agents of the authenticated tenant.
  • subject — string — ESCOPO1 (optional): opaque 8-128 char token ([A-Za-z0-9._:-]) that isolates memory to a single person within a shared connect-token. Omit for token-scoped behavior. Never include PII/email.
  • agent_id — string — Agent identifier (canonical or alias). Used only when scope="agent" or scope is inferred.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world no

• memory_feedback

  Submit feedback to adjust trust scoring for stored memories.

  Parameters:

  • signal (required) — string — Feedback signal: used (auto), helpful (positive), irrelevant (negative), ignored.
  • memoryId — string — Deprecated camelCase alias of memory_id — still accepted for backward compatibility.
  • memory_id — string — ID of the memory to provide feedback on (canonical snake_case name).
  • sessionId — string — Deprecated camelCase alias of session_id — still accepted for backward compatibility.
  • session_id — string — Optional session identifier for tracking (canonical snake_case name).

  Hints: read-only no · destructive not declared · idempotent not declared · open-world no

• neural_handover

  Transfer contextual state between agents with continuity controls.

  Parameters:

  • token (required) — string — The handover token (64 hex chars). Generated by the source agent/user in Forge UI.
  • reason (required) — string — REQUIRED (10–200 chars). Human-readable justification for consuming this handover. Example: "Receiving project context from Claude session."
  • consuming_agent — string — Agent identifier consuming the handover (defaults to token agent).

  Hints: read-only no · destructive not declared · idempotent not declared · open-world no

• slo_report read-only

  Return uptime and SLO metrics for the MCP service.

  Parameters:

  • admin_token (required) — string — Admin authentication token. Must match ADMIN_TOKEN environment variable.
  • window_hours — array — Time windows in hours for the report. Default: [24, 168] (24h + 7d).

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world no

Publisher website Repository

Measured by MCPYP probing. Untested fields are omitted, never estimated. Listing is always free; MCPYP never proxies calls or stores user tokens.