nbj-ob1-agent-memory-openclaw
This skill enables Claude to work with NBJ OB1 Agent Memory within OpenClaw tasks, providing structured recall and write-back of operational context. Use it when an OpenClaw environment has access to OB1 memory tools and you need to retrieve prior task context before work, apply that context according to its use policy, and then record decisions, outputs, and lessons back to memory with appropriate provenance labels. The system prioritizes project-scoped recall, honors user confirmation for binding memories, and avoids storing transcripts or sensitive values.
git clone --depth 1 https://github.com/NateBJones-Projects/OB1 /tmp/nbj-ob1-agent-memory-openclaw && cp -r /tmp/nbj-ob1-agent-memory-openclaw/skills/openclaw-agent-memory ~/.claude/skills/nbj-ob1-agent-memory-openclawSKILL.md
# NBJ OB1 Agent Memory for OpenClaw Use this skill when an OpenClaw task has access to NBJ OB1 Agent Memory tools. OB1 is the continuity layer. OpenClaw is the runtime that performs the work. ## Core Rule Recall before meaningful work. Write back only compact, provenance-labeled operational memory after the work is complete. ## Available Tools Use the OpenClaw plugin tools when available: - `openbrain_recall` - `openbrain_writeback` - `openbrain_report_usage` - `openbrain_inspect_memory` - `openbrain_list_review_queue` - `openbrain_review_memory` - `openbrain_get_recall_trace` If the tools are unavailable, continue the task normally and note that no OB1 recall or write-back occurred. Do not invent remembered facts. ## Pre-Task Recall Before meaningful work, call `openbrain_recall` with: - `task_type` - `query` - `entities` - `scope` - `limits` - `sensitivity` Prefer project-scoped recall when a project is known. Keep `project_only` true by default. Keep `include_unconfirmed` false unless the user explicitly asks for evidence-level context or the task is review/debugging. Do not pull personal or channel-only memory into team work unless the user explicitly shared it. Use returned memories according to `use_policy`: - `can_use_as_instruction`: the memory can guide behavior directly. - `can_use_as_evidence`: the memory can inform reasoning, but it is not binding. - `requires_user_confirmation`: surface the claim before relying on it. If two memories conflict, prefer user-confirmed or trusted imported memory over inferred or generated memory. If the conflict matters, ask for confirmation or proceed with the lower-risk assumption. ## Post-Task Write-Back After the task completes, call `openbrain_writeback` with compact categories: - `decisions` - `outputs` - `lessons` - `constraints` - `unresolved_questions` - `next_steps` - `failures` - `artifacts` Do not write raw transcripts, model reasoning traces, secret-like values, credential strings, large code blocks, or private customer data dumps. Store summaries and source references. Agent-written memory starts as evidence by default. It can become instruction only when a human confirms it or it is imported from a trusted source. Decision memories can become future instructions only when they are `user_confirmed` or imported from a trusted source. Model-generated lessons are evidence or suggestions until reviewed. ## Usage Reporting After recall, report which memory IDs were used or ignored with `openbrain_report_usage`. This preserves the recall trace so bad behavior can be debugged later. ## Code Review Memory Mode For pull request review tasks, recall: - repo conventions - prior review comments - recurring bug patterns - risky files or subsystems - test expectations - security-sensitive patterns - maintainer preferences Write back: - recurring issue patterns - maintainer corrections - false positives to avoid - tests that caught or missed the issue - fixes applied - reusable repo lessons Do not store the full diff. Store artifact references to PRs, commits, or files. ## TaskFlow Work Log Mode For long-running TaskFlows, recall: - prior task attempts - blockers - decisions - project constraints - owner and channel context Write back: - what was attempted - what changed - what failed - what remains open - what should be reviewed - what the next agent should know The write-back should let another agent continue without reading the full transcript.
Continuous learning system that extracts reusable knowledge from work sessions. Triggers: (1) /aiception command, (2) 'save this as a skill' or 'extract a skill from this', (3) 'what did we learn?', (4) after non-obvious debugging or trial-and-error discovery. Creates new skills when valuable reusable knowledge is identified. Integrates with Open Brain to prevent duplicates.
Morning digest of yesterday's Open Brain thoughts, drafted to Gmail
Generate infographic images from any research doc, Open Brain thoughts, or analysis. Auto-chunks content, writes prompts, generates images via Gemini API (free tier), and saves to media/. Use --premium for better text rendering.
|
Use when processing voice transcripts, brain dumps, stream-of-consciousness notes, or any raw multi-topic capture. Extracts every idea thread, then evaluates each one with deep brainstorming, then captures results to Open Brain. Trigger on transcripts, exports, "process this", "pan for gold", "brain dump", "what did I say", or multi-topic markdown files.
|
|