Sub-Agent Context Continuity

Signet has implemented live transcript persistence and sub-agent context inheritance to solve a gap where sub-agents spawned from parent sessions started with no memory or entity context. The engine now writes session transcripts on every user prompt and at session end, making active session content queryable, and automatically threads parent context into child sub-agents at spawn time.

Sub-Agent Context Continuity Engine-managed context threading from parent to sub-agent. “The engine manages memory deterministically. The agent doesn’t have to think about memory management — the engine just does the right thing.” — LCM paper, adapted Background Issue 315 surfaced a gap: when a parent agent spawns a sub-agent, the sub-agent starts cold. No memories. No entity context. No awareness of what the parent session has accumulated. The parent must manually summarize relevant context in the task description to compensate, which is fragile and burdens every sub-agent invocation. There is a deeper root cause underneath this: session transcripts are only persisted at session end. An active session is dark — its content is unavailable to any other session until it terminates. This makes the parent→child context problem unsolvable even if we wanted to solve it, because the parent’s transcript doesn’t exist yet at the time the sub-agent spawns. This spec addresses both: fix the persistence gap first, then build context inheritance on top of it. Phase 1: Live Transcript Persistence Status: complete. The Problem session transcripts migration 040, hardened by migration 047 stores the complete transcript and keeps it agent-scoped. Prompt-time upserts already keep active sessions visible before session end. If a session is interrupted or still running when a sub-agent spawns, the latest prompt-time snapshot remains queryable. The Fix Keep writing to session transcripts on every UserPromptSubmit hook call and at SessionEnd . The shared upsertSessionTranscript helper preserves created at , refreshes updated at , and updates the FTS surface through the table triggers. The active schema supports this: CREATE TABLE session transcripts session key TEXT NOT NULL, content TEXT NOT NULL, harness TEXT, project TEXT, agent id TEXT NOT NULL DEFAULT 'default', created at TEXT NOT NULL, updated at TEXT, PRIMARY KEY agent id, session key ; The upsert on each UserPromptSubmit keeps content current as the session grows. The created at column preserves the original write time and updated at records the latest active snapshot. session end retains its own write as a final snapshot to ensure the terminal state is captured even if the last few turns missed a hook. What This Enables Immediately Any session — sub-agent or otherwise — can now query a running parent session’s transcript via the existing session transcripts table. A new session search MCP tool and /api/sessions/search endpoint expose this. Cross-session transcript availability also improves debugging, diagnostics, and the session summary DAG LCM Pattern 4 which currently can only process completed sessions. Implementation platform/daemon/src/hooks.ts : handleUserPromptSubmit reads the transcript path or inline transcript, normalizes it, and calls upsertSessionTranscript . platform/daemon/src/hooks.ts : handleSessionEnd calls the same helper for the terminal write. platform/daemon/src/session-transcripts.ts : owns the scoped upsert, read, and FTS-backed search helpers. Phase 2: Sub-Agent Context Inheritance The Principle When a sub-agent spawns, the daemon should thread the parent’s context to it automatically. No prepare/claim tokens. No explicit tool calls. The engine observes the parent-child relationship through harness-native signals and injects at session-start. This is the LCM scope-reduction invariant applied to Signet: the parent delegates a scoped task and the engine ensures the sub-agent has the context it needs to perform it. Harness Detection Matrix Each harness surfaces parent-child relationships differently. Signet handles each natively: Claude Code — agent id is present on all hook payloads inside a sub-agent session added CC v2.1.69, March 5 2026 . No SubagentStart hook needed. Detection: - Sub-agent’s SessionStart arrives with agent id present. - Daemon queries session transcripts for the most recently updated row matching the same project , harness , and Signet agent id where session key = current : SELECT FROM session transcripts WHERE agent id = ? AND project = ? AND harness = ? ORDER BY updated at DESC LIMIT 1 - That row is the parent. No TTL, no pending-spawn slot, no extra agent action. If two parent sessions exist for the same project, the most recently active one is the least surprising choice. OpenClaw — session keys self-describe lineage: agent:{id}:subagent:{uuid} vs agent:{id}:main . The resolveAgentId function already parses this format Multi-Agent Phase 8 . Detection: - In handleSessionStart , if isSubagentSessionKey sessionKey is true, extract the parent session key from the format and query session transcripts directly. No pending-spawn slot needed. OpenCode — session.created SSE event includes parentID on child sessions. Detection: - The plugin records the child session’s parentID and passes it through the next session-start hook body as parentSessionKey . Daemon uses it for direct lookup. Codex — sub-agent sessions are indistinguishable from external hook payloads. No parent info is available. Graceful degradation: no context block injected. The agent can still call session search manually if the parent session key is known. The Context Block The inherited context block is assembled deterministically from data already in the DB. No LLM call. This keeps the session-start latency budget intact and avoids the cost-at-spawn problem LCM’s lcm expand query sub-agent isolates. Content ordered by usefulness, subject to token budget : Checkpoint summary — content from the most recent session checkpoints row for the parent session key, if one exists. Checkpoints are already budgeted and formatted for injection; no re-processing needed. Transcript tail — last 3000 characters of session transcripts.content for turns since the last checkpoint or from the start if no checkpoint exists . Character budget, no turn parsing, no harness-specific format logic. Active constraints — entity attributes where kind = 'constraint' for focal entities in the checkpoint. Always included per Invariant 5. Focal entity names — entity names from the checkpoint’s structural snapshot, if present. If the parent has neither a checkpoint nor a transcript yet, the block is omitted rather than injected empty. The block is formatted as a named section in the inject: Inherited from Parent Session session title or "active session" Recent context: last N turns, truncated to budget Focal entities: entity names Active constraints: constraint text If the parent session has no transcript yet e.g., it just started , the block is omitted rather than injected empty. Sub-Agent Session Search With Phase 1 in place, sub-agents can query any session’s transcript, including the parent’s active session, via a new MCP tool: mcp signet session search query, sessionKey? query — natural language or keyword search applied to session transcripts.content via FTS5 sessionKey — optional; defaults to the parent session key if the current session is a sub-agent This gives sub-agents pull access to parent context on demand, complementing the push at session-start. Follows LCM’s On-Demand Expansion pattern Pattern 5 applied across session boundaries. The session search MCP tool is also useful for the parent agent to retrospectively search its own prior sessions, independent of the sub-agent use case. Configuration memory: pipelineV2: subagents: inheritContext: true default: true when parent is detected tailChars: 3000 chars from transcript tail since last checkpoint includeEntities and includeConstraints are always on — constraints surface unconditionally per Invariant 5. inheritContext: false disables the inject block entirely while leaving session search available. Visibility scoping is not configurable: sub-agents always inherit the parent’s agent id for all queries. An isolated parent’s sub-agents see only the parent’s data; a shared parent’s sub-agents see all visibility=global data. The existing agent scoping invariant handles this with no special-casing. Implementation platform/daemon/src/hooks.ts : add parent-lookup logic to handleSessionStart . For CC: if agent id is present in the payload, query session transcripts WHERE project = ? AND session key = ? ORDER BY updated at DESC LIMIT 1 . For OpenClaw: if isSubagentSessionKey sessionKey , extract parent key from session key format and query directly. For OpenCode: if parentSessionKey is present in the hook body, use it directly. If parent found, call assembleInheritBlock . platform/daemon/src/hooks.ts : add assembleInheritBlock parentKey, cfg — fetches latest session checkpoints row for parent, takes cfg.tailChars from tail of session transcripts.content , fetches active constraints for checkpoint’s focal entities. Pure DB reads, no LLM. Returns formatted block or null if no data exists yet. platform/daemon/src/mcp.ts : register session search tool. Queries session transcripts via FTS5; sessionKey param defaults to parent key when current session is a sub-agent. platform/daemon/src/daemon.ts : add GET /api/sessions/{key}/transcript and POST /api/sessions/search endpoints. platform/core/src/types.ts : extend AgentManifest with memory.pipelineV2.subagents config block.- No connector changes needed for CC — no SubagentStart hook required. Relationship to LCM Patterns This spec is a direct application of two LCM patterns from docs/specs/planning/LCM-PATTERNS.md : Pattern 1 Lossless Retention — Phase 1 applies lossless retention at the turn level. Active sessions are no longer dark. The immutable store LCM’s term is the session transcripts table; every turn is persisted verbatim as it arrives. Pattern 5 On-Demand Expansion — the session search tool is the pull path. The inherited context block is the push path. Together they cover both the predictable what the engine infers the sub-agent needs and the unpredictable what the sub-agent discovers it needs mid-task . The key deviation from LCM’s lcm expand query which spawns a sub-agent for expansion is that Signet’s expansion runs in-process. No sub-agent needed for expansion because the knowledge is structured in SQLite — a targeted query is cheaper and faster than a spawned agent. Decisions - Transcript slice format — character budget 3000 chars from tail , not discrete turn parsing. The inherit block is for orientation, not perfect reproduction. session search is the pull path for anything more specific. No harness-specific parsing needed. - CC parent detection — project-keyed recency query on session transcripts , no SubagentStart hook and no pending-spawn slot. ORDER BY updated at DESC LIMIT 1 where project, harness, and Signet agent scope match and session key differs. Simpler, no extra prepare step, and it follows the most recently active parent. - Visibility scoping — sub-agents inherit parent’s agent id automatically. Not configurable. The existing agent scoping invariant cross-cutting invariant 1 handles this with no special-casing. - Harness sub-agent IDs are not Signet agent IDs — Claude Code’s agent id hook field is treated as harness lineage metadata only. It must not be used as Signet’s persistence agent id or sub-agent sessions would silently fork user memory scope. Written by Nicholai and Ant. March 24, 2026. Informed by issue 315 and harness hook research Claude Code v2.1.69+, OpenClaw plugin hooks, OpenCode session.created parentID, Codex graceful degradation .