openclaw/docs/reference/session-management-compaction.md

summary, read_when, title

summary	read_when	title
Deep dive: session store + transcripts, lifecycle, and (auto)compaction internals	You need to debug session ids, SQLite session rows/events, or doctor migration of legacy sessions.json/JSONL files You are changing auto-compaction behavior or adding "pre-compaction" housekeeping You want to implement memory flushes or silent system turns	Session management deep dive

OpenClaw manages sessions end-to-end across these areas:

• Session routing (how inbound messages map to a sessionKey)
• Session store and what it tracks
• Transcript persistence (SQLite event streams, doctor-only JSONL import, explicit debug export) and its structure
• Transcript hygiene (provider-specific fixups before runs)
• Context limits (context window vs tracked tokens)
• Compaction (manual and auto-compaction) and where to hook pre-compaction work
• Silent housekeeping (memory writes that should not produce user-visible output)

If you want a higher-level overview first, start with:

• Session management
• Compaction
• Memory overview
• Memory search
• Session pruning
• Transcript hygiene

---

Source of truth: the Gateway

OpenClaw is designed around a single Gateway process that owns session state.

• UIs (macOS app, web Control UI, TUI) should query the Gateway for session lists and token counts.
• In remote mode, session databases are on the remote host; "checking your local Mac files" won't reflect what the Gateway is using.

---

Two persistence layers

OpenClaw persists sessions in two layers:

1. Session store

   • Key/value map: sessionKey -> SessionEntry
   • SQLite-backed by default; legacy JSON import is doctor-only and support export is explicit
   • Tracks session metadata (current session id, last activity, toggles, token counters, etc.)

2. Transcript (agentId, sessionId)

   • SQLite-backed transcript event stream with tree structure (entries have id + parentId)
   • Stores the actual conversation + tool calls + compaction summaries
   • Used to rebuild the model context for future turns
   • Stored in SQLite for OpenClaw-owned runtime paths; JSONL files are legacy doctor-import inputs or explicit support artifacts, not runtime compatibility sidecars

• Runtime code passes structured agent/session scope. There is no active transcript file, URI, or locator layer.
• Scoped latest/tail assistant-text lookups, session exports, before_reset hook payloads, silent session rotations, chat history, TUI history, recovery, managed media indexing, token estimation, title/preview/usage helpers, and bounded session inspection read the scoped SQLite transcript.
• Pre-compaction checkpoints are SQLite transcript snapshots. OpenClaw does not create .checkpoint.*.jsonl copies on the runtime path.

Gateway history readers should avoid materializing the whole transcript unless the surface explicitly needs arbitrary historical access. First-page history, embedded chat history, restart recovery, and token/usage checks use bounded tail reads. Full transcript scans are keyed by SQLite agent/session scope, not by a file path.

---

On-disk locations

Per agent, on the Gateway host:

• Global store: ~/.openclaw/state/openclaw.sqlite by default. It stores shared registry, migration, plugin, task, and backup metadata.
• Agent store: ~/.openclaw/agents/<agentId>/agent/openclaw-agent.sqlite. It stores canonical session rows, transcript events, snapshots, VFS entries, artifacts, and agent-local cache rows.
• Legacy imports: openclaw doctor --fix imports ~/.openclaw/agents/<agentId>/sessions/sessions.json indexes and JSONL transcripts into the agent SQLite database, then removes imported legacy sources after durable verification. Gateway startup leaves legacy indexes alone.
• Transcripts: runtime transcript events live in the per-agent database (transcript_events and transcript_event_identities). The canonical identity is structured scope: agentId plus sessionId. Legacy JSONL files are doctor migration inputs or explicit support artifacts, never runtime sidecars or compatibility handles.

OpenClaw resolves these via src/config/sessions/*.

---

Store Cleanup

SQLite is the canonical per-agent session backend. sessions.json is a legacy doctor-import input, not a parallel export/debug store. Runtime code should read and write explicit { agentId, sessionKey } rows.

Runtime writes normalize and persist only; they do not prune, cap, import, archive, or run disk-budget cleanup. Session store reads also do not import, prune, or cap entries during Gateway startup. Use openclaw doctor --fix for legacy JSON/JSONL import.

OpenClaw no longer creates automatic sessions.json.bak.* rotation backups during Gateway writes. Legacy session.maintenance.* and session.writeLock.* settings are doctor-migrated raw config only, and openclaw doctor --fix removes them from older configs.

Transcript mutations are serialized through SQLite transactions plus the per-session append queue. Runtime bootstrap and manual compaction repair write SQLite transcript rows directly. Any retained JSONL shape is an explicit doctor/import/export/debug boundary, not a runtime lookup or persistence path.

Legacy session import belongs to openclaw doctor --fix. Runtime no longer has a session cleanup command that prunes missing transcript rows; after doctor runs, reset or delete any intentionally stale session explicitly.

---

Cron sessions and run logs

Isolated cron runs also create session entries/transcripts. Session rows use the same SQLite session tables as other rows:

• Legacy cron session imports happen through openclaw doctor --fix.
• cron.runLog.maxBytes + cron.runLog.keepLines prune SQLite cron run history (defaults: 2_000_000 approximate serialized bytes and 2000 rows per job).

When cron force-creates a new isolated run session, it sanitizes the previous cron:<jobId> session entry before writing the new row. It carries safe preferences such as thinking/fast/verbose settings, labels, and explicit user-selected model/auth overrides. It drops ambient conversation context such as channel/group routing, send or queue policy, elevation, origin, and ACP runtime binding so a fresh isolated run cannot inherit stale delivery or runtime authority from an older run.

---

Session keys (sessionKey)

A sessionKey identifies which conversation bucket you're in (routing + isolation).

Common patterns:

• Main/direct chat (per agent): agent:<agentId>:<mainKey> (default main)
• Group: agent:<agentId>:<channel>:group:<id>
• Room/channel (Discord/Slack): agent:<agentId>:<channel>:channel:<id> or ...:room:<id>
• Cron: cron:<job.id>
• Webhook: hook:<uuid> (unless overridden)

The canonical rules are documented at /concepts/session.

---

Session ids (sessionId)

Each sessionKey points at a current sessionId (the SQLite transcript identity that continues the conversation).

Rules of thumb:

• Reset (/new, /reset) creates a new sessionId for that sessionKey.
• Daily reset (default 4:00 AM local time on the gateway host) creates a new sessionId on the next message after the reset boundary.
• Idle expiry (session.reset.idleMinutes) creates a new sessionId when a message arrives after the idle window. When daily + idle are both configured, whichever expires first wins. openclaw doctor --fix migrates old session.idleMinutes configs into session.reset.idleMinutes.
• System events (heartbeat, cron wakeups, exec notifications, gateway bookkeeping) may mutate the session row but do not extend daily/idle reset freshness. Reset rollover discards queued system-event notices for the previous session before the fresh prompt is built.
• Parent fork policy uses PI's active branch when creating a thread or subagent fork. If that branch is too large, OpenClaw starts the child with isolated context instead of failing or inheriting unusable history. The sizing policy is automatic; legacy session.parentForkMaxTokens config is removed by openclaw doctor --fix.

Implementation detail: the decision happens in initSessionState() in src/auto-reply/reply/session.ts.

---

Session store schema

The store's value type is SessionEntry in src/config/sessions/types.ts.

Key fields (not exhaustive):

• sessionStartedAt: start timestamp for the current sessionId; daily reset freshness uses this. Legacy rows may derive it from the JSONL session header.
• lastInteractionAt: last real user/channel interaction timestamp; idle reset freshness uses this so heartbeat, cron, and exec events do not keep sessions alive. Legacy rows without this field fall back to the recovered session start time for idle freshness.
• updatedAt: last store-row mutation timestamp, used for listing and bookkeeping. It is not the authority for daily/idle reset freshness.
• sessionId: current SQLite transcript id; callers pass structured { agentId, sessionId } scope instead of a transcript path override
• chatType: direct | group | room (helps UIs and send policy)
• provider, subject, room, space, displayName: metadata for group/channel labeling
• Toggles:
  • thinkingLevel, verboseLevel, reasoningLevel, elevatedLevel
  • sendPolicy (per-session override)
• Model selection:
  • providerOverride, modelOverride, authProfileOverride
• Token counters (best-effort / provider-dependent):
  • inputTokens, outputTokens, totalTokens, contextTokens
• compactionCount: how often auto-compaction completed for this session key
• memoryFlushAt: timestamp for the last pre-compaction memory flush
• memoryFlushCompactionCount: compaction count when the last flush ran

The store is safe to edit, but the Gateway is the authority: it may rewrite or rehydrate entries as sessions run.

---

Transcript structure

Transcripts are stored as SQLite rows and opened by {agentId, sessionId}.

The event stream is stored in the per-agent transcript_events table:

• First event: session header (type: "session", includes id, cwd, timestamp, optional parentSession)
• Then: session entries with id + parentId (tree)

Doctor JSONL import uses the same event shape, one JSON object per line. User-facing exports may materialize support-bundle JSONL from SQLite rows, but runtime code does not read or write transcript JSONL files.

Notable entry types:

• message: user/assistant/toolResult messages
• custom_message: extension-injected messages that do enter model context (can be hidden from UI)
• custom: extension state that does not enter model context
• compaction: persisted compaction summary with firstKeptEntryId and tokensBefore
• branch_summary: persisted summary when navigating a tree branch

Runtime transcript repair and compaction mutate SQLite rows through scoped transcript APIs. Legacy JSONL shape upgrades happen only in doctor import before rows are written.

---

Context windows vs tracked tokens

Two different concepts matter:

1. Model context window: hard cap per model (tokens visible to the model)
2. Session store counters: rolling stats written into the session store (used for /status and dashboards)

If you're tuning limits:

• The context window comes from the model catalog (and can be overridden via config).
• contextTokens in the store is a runtime estimate/reporting value; don't treat it as a strict guarantee.

For more, see /token-use.

---

Compaction: what it is

Compaction summarizes older conversation into a persisted compaction entry in the transcript and keeps recent messages intact.

After compaction, future turns see:

• The compaction summary
• Messages after firstKeptEntryId

Compaction is persistent (unlike session pruning). See /concepts/session-pruning.

Compaction chunk boundaries and tool pairing

When OpenClaw splits a long transcript into compaction chunks, it keeps assistant tool calls paired with their matching toolResult entries.

• If the token-share split lands between a tool call and its result, OpenClaw shifts the boundary to the assistant tool-call message instead of separating the pair.
• If a trailing tool-result block would otherwise push the chunk over target, OpenClaw preserves that pending tool block and keeps the unsummarized tail intact.
• Aborted/error tool-call blocks do not hold a pending split open.

---

When auto-compaction happens (Pi runtime)

In the embedded Pi agent, auto-compaction triggers in two cases:

1. Overflow recovery: the model returns a context overflow error (request_too_large, context length exceeded, input exceeds the maximum number of tokens, input token count exceeds the maximum number of input tokens, input is too long for the model, ollama error: context length exceeded, and similar provider-shaped variants) → compact → retry.
2. Threshold maintenance: after a successful turn, when:

contextTokens > contextWindow - reserveTokens

Where:

• contextWindow is the model's context window
• reserveTokens is headroom reserved for prompts + the next model output

These are Pi runtime semantics (OpenClaw consumes the events, but Pi decides when to compact).

OpenClaw can also trigger a preflight local compaction before opening the next run when agents.defaults.compaction.maxActiveTranscriptBytes is set and the active SQLite transcript reaches that size. This is a transcript-size guard for local reopen cost, not raw archival: OpenClaw still runs normal semantic compaction, and it requires rotateAfterCompaction so the compacted summary can become a new successor transcript.

For embedded Pi runs, agents.defaults.compaction.midTurnPrecheck.enabled: true adds an opt-in tool-loop guard. After a tool result is appended and before the next model call, OpenClaw estimates the prompt pressure using the same preflight budget logic used at turn start. If the context no longer fits, the guard does not compact inside Pi's transformContext hook. It raises a structured mid-turn precheck signal, stops the current prompt submission, and lets the outer run loop use the existing recovery path: truncate oversized tool results when that is enough, or trigger the configured compaction mode and retry. The option is disabled by default and works with both default and safeguard compaction modes, including provider-backed safeguard compaction. This is independent of maxActiveTranscriptBytes: the byte-size guard runs before a turn opens, while mid-turn precheck runs later in the embedded Pi tool loop after new tool results have been appended.

---

Compaction settings (reserveTokens, keepRecentTokens)

Pi's compaction settings live in Pi settings:

{
  compaction: {
    enabled: true,
    reserveTokens: 16384,
    keepRecentTokens: 20000,
  },
}

OpenClaw also enforces a safety floor for embedded runs:

• If compaction.reserveTokens < reserveTokensFloor, OpenClaw bumps it.
• Default floor is 20000 tokens.
• Set agents.defaults.compaction.reserveTokensFloor: 0 to disable the floor.
• If it's already higher, OpenClaw leaves it alone.
• Manual /compact honors an explicit agents.defaults.compaction.keepRecentTokens and keeps Pi's recent-tail cut point. Without an explicit keep budget, manual compaction remains a hard checkpoint and rebuilt context starts from the new summary.
• Set agents.defaults.compaction.midTurnPrecheck.enabled: true to run the optional tool-loop precheck after new tool results and before the next model call. This is a trigger only; summary generation still uses the configured compaction path. It is independent of maxActiveTranscriptBytes, which is a turn-start active-transcript byte-size guard.
• Set agents.defaults.compaction.maxActiveTranscriptBytes to a byte value or string such as "20mb" to run local compaction before a turn when the active transcript gets large. This guard is active only when rotateAfterCompaction is also enabled. Leave it unset or set 0 to disable.
• When agents.defaults.compaction.rotateAfterCompaction is enabled, OpenClaw rewrites the active SQLite transcript to the compacted successor after compaction. The old full transcript is available only through the SQLite pre-compaction checkpoint snapshot while retained.

Why: leave enough headroom for multi-turn "housekeeping" (like memory writes) before compaction becomes unavoidable.

Implementation: ensurePiCompactionReserveTokens() in src/agents/pi-settings.ts (called from src/agents/pi-embedded-runner.ts).

---

Pluggable compaction providers

Plugins can register a compaction provider via registerCompactionProvider() on the plugin API. When agents.defaults.compaction.provider is set to a registered provider id, the safeguard extension delegates summarization to that provider instead of the built-in summarizeInStages pipeline.

• provider: id of a registered compaction provider plugin. Leave unset for default LLM summarization.
• Setting a provider forces mode: "safeguard".
• Providers receive the same compaction instructions and identifier-preservation policy as the built-in path.
• The safeguard still preserves recent-turn and split-turn suffix context after provider output.
• Built-in safeguard summarization re-distills prior summaries with new messages instead of preserving the full previous summary verbatim.
• Safeguard mode enables summary quality audits by default; set qualityGuard.enabled: false to skip retry-on-malformed-output behavior.
• If the provider fails or returns an empty result, OpenClaw falls back to built-in LLM summarization automatically.
• Abort/timeout signals are re-thrown (not swallowed) to respect caller cancellation.

Source: src/plugins/compaction-provider.ts, src/agents/pi-hooks/compaction-safeguard.ts.

---

User-visible surfaces

You can observe compaction and session state via:

• /status (in any chat session)
• openclaw status (CLI)
• openclaw sessions / sessions --json
• Gateway logs (pnpm gateway:watch or openclaw logs --follow): embedded run auto-compaction start + complete
• Verbose mode: 🧹 Auto-compaction complete + compaction count

---

Silent housekeeping (NO_REPLY)

OpenClaw supports "silent" turns for background tasks where the user should not see intermediate output.

Convention:

• The assistant starts its output with the exact silent token NO_REPLY / no_reply to indicate "do not deliver a reply to the user".
• OpenClaw strips/suppresses this in the delivery layer.
• Exact silent-token suppression is case-insensitive, so NO_REPLY and no_reply both count when the whole payload is just the silent token.
• This is for true background/no-delivery turns only; it is not a shortcut for ordinary actionable user requests.

As of 2026.1.10, OpenClaw also suppresses draft/typing streaming when a partial chunk begins with NO_REPLY, so silent operations don't leak partial output mid-turn.

---

Pre-compaction "memory flush" (implemented)

Goal: before auto-compaction happens, run a silent agentic turn that writes durable state to disk (e.g. memory/YYYY-MM-DD.md in the agent workspace) so compaction can't erase critical context.

OpenClaw uses the pre-threshold flush approach:

1. Monitor session context usage.
2. When it crosses a "soft threshold" (below Pi's compaction threshold), run a silent "write memory now" directive to the agent.
3. Use the exact silent token NO_REPLY / no_reply so the user sees nothing.

Config (agents.defaults.compaction.memoryFlush):

• enabled (default: true)
• model (optional exact provider/model override for the flush turn, for example ollama/qwen3:8b)
• softThresholdTokens (default: 4000)
• prompt (user message for the flush turn)
• systemPrompt (extra system prompt appended for the flush turn)

Notes:

• The default prompt/system prompt include a NO_REPLY hint to suppress delivery.
• When model is set, the flush turn uses that model without inheriting the active session fallback chain, so local-only housekeeping does not silently fall back to a paid conversation model.
• The flush runs once per compaction cycle (tracked in the session store).
• The flush runs only for embedded Pi sessions (CLI backends skip it).
• The flush is skipped when the session workspace is read-only (workspaceAccess: "ro" or "none").
• See Memory for the workspace file layout and write patterns.

Pi also exposes a session_before_compact hook in the extension API, but OpenClaw's flush logic lives on the Gateway side today.

---

Troubleshooting checklist

• Session key wrong? Start with /concepts/session and confirm the sessionKey in /status.
• Session metadata vs transcript mismatch? Confirm the Gateway host and agent database from openclaw status.
• Compaction spam? Check:
  • model context window (too small)
  • compaction settings (reserveTokens too high for the model window can cause earlier compaction)
  • tool-result bloat: review compaction thresholds and tool-result persistence
• Silent turns leaking? Confirm the reply starts with NO_REPLY (case-insensitive exact token) and you're on a build that includes the streaming suppression fix.

Related

• Session management
• Session pruning
• Context engine