\r \r

Aura for OpenClaw\r

\r Persistent memory and instant knowledge retrieval for your OpenClaw agent. Compile 60+ file formats into queryable .aura archives and give your agent a 3-tier memory that survives sessions, compaction, and restarts.\r \r

Memory OS v2.1 (auralith-aura>=0.2.3): Enhanced with temporal decay scoring, noise filtering, deduplication, bloom filters, SimHash fuzzy matching, and tiered priority scoring — zero RAM overhead.\r \r

Setup\r

\r

pip install auralith-aura\r
```\r
\r
For full document support (PDFs, DOCX, etc.):\r
```bash\r
pip install 'auralith-aura[all]'\r
```\r
\r
## What It Looks Like\r
\r
### Learn an entire codebase\r
\r
```\r
You: Learn my project in ~/repos/api-service\r
Agent: 🔥 Compiling ~/repos/api-service...\r
       ✅ Compiled 847 files → api-service.aura\r
       Indexed: Python (312), TypeScript (201), YAML (44), Markdown (73), JSON (217)\r
\r
You: How does the auth middleware handle token refresh?\r
Agent: Based on src/middleware/auth.ts (lines 45-89), the auth middleware\r
       intercepts expired JWTs by checking the `exp` claim. When expired,\r
       it calls `refreshTokenService.rotate()` which issues a new access\r
       token and sets it via the `X-Refreshed-Token` response header.\r
       The original request is then replayed with the new token.\r
```\r
\r
### Compile legal docs with PII masking\r
\r
```\r
You: Learn everything in ~/legal but mask any personal info\r
Agent: 🔥 Compiling with PII masking enabled...\r
       ✅ Compiled → legal.aura\r
       Masked: 23 emails, 8 phone numbers, 4 SSNs\r
\r
You: What are the termination conditions in the Series A agreement?\r
Agent: Per series_a_term_sheet.pdf, Section 7.2 — the agreement may be\r
       terminated by either party with 30 days written notice if:\r
       (a) material breach remains uncured after 15 business days,\r
       (b) insolvency or bankruptcy filing, or\r
       (c) failure to meet the minimum revenue threshold in Schedule B...\r
```\r
\r
### Remember facts across sessions\r
\r
```\r
You: Remember that our production database is on us-east-1, the staging\r
     API key rotates every 30 days, and I prefer verbose error logs.\r
\r
Agent: ✅ Written to /fact:\r
       • Production DB region: us-east-1\r
       • Staging API key rotation: 30-day cycle\r
       • Preference: verbose error logging\r
\r
--- next day, new session ---\r
\r
You: Set up the deployment config for staging\r
Agent: I remember your staging API key rotates every 30 days — I'll add\r
       a reminder comment in the config. Also setting error logging to\r
       verbose per your preference. Targeting us-east-1 since that's\r
       where your production DB lives...\r
```\r
\r
### Track decisions across a long project\r
\r
```\r
You: We decided to use PostgreSQL over MongoDB for the user service,\r
     and the team agreed on GraphQL for the public API.\r
\r
Agent: ✅ Written to /episodic:\r
       • Architecture decision: PostgreSQL for user service (over MongoDB)\r
       • API decision: GraphQL for public-facing API\r
\r
--- two weeks later ---\r
\r
You: Wait, should we use MongoDB for the new analytics service?\r
Agent: Based on your earlier decision log — you chose PostgreSQL over\r
       MongoDB for the user service citing ACID compliance needs. The\r
       analytics service has different requirements (write-heavy, schema\r
       flexibility), so MongoDB could make sense here. Want me to draft\r
       a comparison?\r
```\r
\r
## Commands\r
\r
### Compile a Knowledge Base\r
\r
When the user asks you to "learn", "compile", "index", or "remember" files from a directory:\r
\r
```bash\r
python {SKILL_DIR}/scripts/compile.py \x3Cinput_directory> \x3Coutput_file>\r
```\r
\r
Options:\r
```bash\r
# Mask PII before compilation\r
python {SKILL_DIR}/scripts/compile.py ./data knowledge.aura --pii-mask\r
\r
# Filter low-quality content\r
python {SKILL_DIR}/scripts/compile.py ./data knowledge.aura --min-quality 0.3\r
```\r
\r
### Query the Knowledge Base\r
\r
```bash\r
python {SKILL_DIR}/scripts/query.py knowledge.aura "search query here"\r
```\r
\r
### Agent Memory\r
\r
Write to memory tiers:\r
```bash\r
python {SKILL_DIR}/scripts/memory.py write pad "scratch note"\r
python {SKILL_DIR}/scripts/memory.py write fact "verified information"\r
python {SKILL_DIR}/scripts/memory.py write episodic "session event"\r
```\r
\r
Search and manage memory:\r
```bash\r
python {SKILL_DIR}/scripts/memory.py query "search query"\r
python {SKILL_DIR}/scripts/memory.py list\r
python {SKILL_DIR}/scripts/memory.py usage\r
python {SKILL_DIR}/scripts/memory.py prune --before 2026-01-01\r
python {SKILL_DIR}/scripts/memory.py end-session\r
```\r
\r
## Memory Tiers\r
\r
| Tier | What It Stores | Lifecycle |\r
|------|---------------|-----------|\r
| **`/pad`** | Working notes, scratch space, in-progress thinking | Transient — cleared between sessions |\r
| **`/episodic`** | Session transcripts, decisions, conversation history | Auto-archived — retained for reference |\r
| **`/fact`** | Verified facts, user preferences, learned rules | Persistent — survives indefinitely |\r
\r
## Supported File Types\r
\r
Documents: PDF, DOCX, DOC, RTF, ODT, EPUB, TXT, HTML, PPTX, EML\r
Data: CSV, TSV, XLSX, XLS, Parquet, JSON, JSONL, YAML, TOML\r
Code: Python, JavaScript, TypeScript, Rust, Go, Java, C/C++, and 20+ more\r
Markup: Markdown (.md), reStructuredText, LaTeX\r
\r
## External Endpoints\r
\r
| URL | Data Sent |\r
|-----|-----------|\r
| None | None |\r
\r
This skill makes **zero network requests**. All processing is local.\r
\r
## Data Provenance & Trust\r
\r
Every memory entry stores `source` (agent/user/system), `namespace`, `timestamp`, `session_id`, and a unique `entry_id`. Nothing is inferred or synthesized — memory contains only what was explicitly written. No hidden embeddings, no derived data.\r
\r
```python\r
memory.show_usage()                              # Inspect what's stored per tier\r
memory.prune_shards(before_date="2026-01-01")    # Prune by date\r
memory.prune_shards(shard_ids=["specific_id"])   # Delete specific shards\r
# Or delete ~/.aura/memory/ to wipe everything\r
```\r
\r
## Security & Privacy\r
\r
- **No data leaves your machine.** All compilation and memory operations run locally.\r
- The `.aura` format uses `safetensors` (no pickle) — no arbitrary code execution risk.\r
- Memory files are stored locally at `~/.aura/memory/`.\r
- No environment variables or API keys are required.\r
- No telemetry, analytics, or usage reporting.\r
\r
## Model Invocation Note\r
\r
This skill is autonomously invoked by the agent as part of its normal operation. The agent decides when to compile documents and manage memory based on user requests. You can disable autonomous invocation in your OpenClaw settings.\r
\r
## Trust Statement\r
\r
By using this skill, **no data is sent to any external service**. All processing happens on your local machine. Only install this skill if you trust [Rta Labs](https://rtalabs.org). Source code for the compiler and RAG components is available on [GitHub](https://github.com/Rtalabs-ai/aura-core).\r
\r
## Notes\r
\r
- Memory OS provides instant writes and background compilation to durable shards.\r
- Compiler and RAG components are open source (Apache 2.0). Memory OS is proprietary, free to use.\r
- For emphasis weighting and training features, see [OMNI Platform](https://omni.rtalabs.org).\r