Skip to content

monasprox/openclaw-memory

BranchesTags

Repository files navigation

Total Recall

Note (April 2026): This project is currently on hold. I'm focusing on other work for the foreseeable future. The code works, the docs are accurate, and issues/PRs are welcome, but response times will be slower than usual. See #22 for details.

Autonomous memory for OpenClaw-style agents.

Total Recall v2.0 keeps the existing v1.x stack intact:

  • Observer
  • Reflector
  • Session Recovery
  • Reactive Watcher
  • Dream Cycle

It also adds the Ambient Intelligence Engine (AIE): a configurable sensor and rumination pipeline that can watch external systems, think about what changed, maintain a preconscious buffer, and surface urgent items.

Prerequisites: python3, jq, curl, and PyYAML.

Install PyYAML with:

python3 -m pip install PyYAML
# or on Debian/Ubuntu:
sudo apt install python3-yaml

setup.sh will verify all dependencies are present before continuing.

Architecture

v1.x memory loop

Observer -> observations.md -> Reflector -> Dream Cycle -> session recovery / watcher

v2.0 Ambient Intelligence Engine

sensor-sweep
  -> enabled connectors emit events to memory/events/bus.jsonl
  -> rumination-engine reads new events and writes structured insights
  -> preconscious-select scores recent insights into memory/preconscious-buffer.md
  -> ambient-actions optionally enriches insights with read-only lookups
  -> emergency-surface pushes urgent alerts through configured channels
  -> buffer-inject sends the latest buffer back into the live session

The AIE flow is additive. None of the v1.x scripts are removed or replaced.

Quick Start

v1.x setup

git clone https://github.com/gavdalf/total-recall.git
cd total-recall
bash scripts/setup.sh

AIE setup

  1. Review config/aie.yaml.
  2. Set workspace if you do not want to use the repo root.
  3. Enable only the connectors and notification channels you actually use.
  4. Add credentials to ./.env or your shell environment.
  5. Run a dry pass:
bash scripts/sensor-sweep.sh --dry-run
bash scripts/rumination-engine.sh --dry-run
bash scripts/preconscious-select.sh --dry-run
bash scripts/ambient-actions.sh --dry-run-actions
  1. Schedule the AIE loop once the dry run looks correct:
*/15 * * * * OPENCLAW_WORKSPACE=/path/to/total-recall bash /path/to/total-recall/scripts/sensor-sweep.sh >> /path/to/total-recall/logs/sensor-sweep.log 2>&1
17,47 * * * * OPENCLAW_WORKSPACE=/path/to/total-recall bash /path/to/total-recall/scripts/preconscious-select.sh >> /path/to/total-recall/logs/rumination.log 2>&1
19,49 * * * * OPENCLAW_WORKSPACE=/path/to/total-recall bash /path/to/total-recall/scripts/emergency-surface.sh >> /path/to/total-recall/logs/emergency-surface.log 2>&1

Minimal first-run path:

  • leave every connector disabled except filewatch
  • keep notifications disabled
  • set only the model IDs you want under models
  • add the API key required by your chosen model provider

AIE Configuration

All new AIE behavior is controlled by config/aie.yaml.

Top-level sections

  • workspace: root directory for memory, logs, health data, and .env
  • paths: all file and directory locations used by AIE scripts
  • profile: assistant/user labels and timezone used in prompts and quiet hours
  • api: shared API metadata such as the outbound HTTP referer
  • models: model IDs for rumination, classification, enrichment, and ambient actions
  • connectors: per-connector enable flags, high-importance sender patterns, and connector-specific settings
  • notifications: quiet hours plus Telegram, Discord, and generic webhook delivery
  • thresholds: cooldowns, pruning windows, emergency thresholds
  • ambient_actions: global action toggle plus read-only tool settings

Connector toggles

Each connector now checks its own enabled flag before doing any work:

  • connectors.calendar.enabled
  • connectors.todoist.enabled
  • connectors.ionos.enabled
  • connectors.gmail.enabled
  • connectors.fitbit.enabled
  • connectors.filewatch.enabled

The orchestrator in scripts/sensor-sweep.sh also skips disabled connectors.

Model settings

Set your own model IDs here:

models:
  rumination: <model-id>
  classification: <model-id>
  enrichment: <model-id>
  ambient_actions: <model-id>

The AIE scripts no longer hardcode user-specific model selections.

Notification settings

notifications:
  quiet_hours:
    enabled: true
    timezone: UTC
    start_hour: 22
    end_hour: 7
  telegram:
    enabled: false
    bot_token: ""
    chat_id: ""
  discord:
    enabled: false
    webhook_url: ""
  webhook:
    enabled: false
    url: ""
    headers: {}

emergency-surface.sh will only send through enabled channels. If no channel is enabled, it exits without error.

Path settings

Everything that was previously tied to ~/clawd is now configurable under paths, including:

  • event bus
  • sensor state
  • rumination directory
  • preconscious buffer
  • health data
  • log directory
  • .env file
  • optional web search helper path

Threshold settings

Key defaults:

thresholds:
  rumination_cooldown_seconds: 1800
  rumination_staleness_seconds: 14400
  sensor_prune_hours: 48
  emergency:
    importance: 0.85
    expires_within_seconds: 14400
    max_alerts_per_day: 2

Connector reference

calendar

  • account
  • calendar_id
  • lookahead_days
  • max_events
  • keyring_password

gmail

  • account
  • query (default newer_than:3h)
  • max_messages
  • keyring_password

ionos

  • account
  • query (recent or unread)
  • max_messages

connectors.scoring

  • model
  • batch_size
  • cache_threshold
  • sender_cache_file

connectors.high_importance_senders

  • list of sender substrings that should raise email importance (default: empty)
  • add patterns for senders you care about most: your employer, accountant, school, doctor, bank, tax authority, etc.
  • example: ["mycompany", "school", "hospital", "accountant", "bank"]

fitbit

  • sleep_target_hours
  • short_sleep_minutes
  • great_sleep_minutes
  • watch_off_minutes
  • watch_uncertain_minutes
  • resting_hr_threshold
  • weight_target_lbs
  • steps_milestone

filewatch

  • watch_files

Ambient action reference

ambient_actions

  • enabled
  • max_actions
  • action_budget_seconds
  • weather_url
  • places.default_lat
  • places.default_lng
  • places.default_limit

ambient_actions.tool_settings

  • calendar_lookup.gog_account
  • calendar_lookup.gog_keyring_password
  • gmail_search.gog_account
  • gmail_search.gog_keyring_password
  • gmail_read.gog_account
  • gmail_read.gog_keyring_password
  • ionos_search.account
  • fitbit_data.enabled
  • openrouter_balance.enabled
  • web_search.enabled
  • web_search.script
  • places_lookup.enabled

Action resolution system

ambient-actions.sh reads enriched rumination insights and decides what to do with them. Five action types are supported, each with guardrails to prevent runaway behaviour:

Action Description Guardrails
ask Surfaces a question to the user via preconscious buffer Max 3/run
learn Stores a confirmed fact to learned-facts.json Importance >= 0.7, dedup, max 5/run
draft Prepares content for user review in drafts/ Importance >= 0.75, max 2/run
notify Sends an urgent alert via emergency-surface Importance >= 0.85, max 2/day, quiet hours respected
remind Creates a time-based nudge in reminders.jsonl Auto-surfaces when due, max 3/run

Runtime files

The AIE creates several files in memory/rumination/ during operation:

  • learned-facts.json — facts the engine has confirmed and stored autonomously
  • reminders.jsonl — pending time-based nudges
  • drafts/ — content prepared for user review (one .md file per draft)
  • cycle-state.json — working memory: tracks lookups between cycles with TTL-based dedup (4hr window)

Agent startup

For best results, have your agent read memory/rumination/learned-facts.json and check memory/rumination/drafts/ at session start. This surfaces what the AIE has learned and prepared between sessions.

Existing v1.x Notes

The original observer / reflector / dream-cycle flow still works as before. The v1.x scripts continue to read their existing environment variables and paths. The new AIE config does not remove that compatibility layer.

See:

Repository Layout

Path Purpose
scripts/observer-agent.sh v1.x observer
scripts/reflector-agent.sh v1.x reflector
scripts/dream-cycle.sh v1.x nightly consolidation helper
scripts/sensor-sweep.sh AIE connector orchestrator
scripts/rumination-engine.sh AIE reasoning pass
scripts/preconscious-select.sh buffer scorer/selector
scripts/ambient-actions.sh read-only enrichment loop
scripts/emergency-surface.sh urgent alert surfacing
scripts/buffer-inject.sh inject buffer back into active session
scripts/connectors/*.sh pluggable AIE sensors
config/aie.yaml AIE runtime configuration

License

MIT. See LICENSE.

About

Five-Layer Memory Protection for OpenClaw AI Agents — observer, reflector, reactive watcher, pre-compaction hook, session recovery

Resources

Readme

License

MIT license

Code of conduct

Code of conduct

Contributing

Contributing
Activity

Stars

1 star

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages