aops-core/skills/remember/SKILL.md
Unified memory skill: immediate mode (/remember) persists knowledge via PKB MCP; maintenance mode (/sleep, GHA cron) runs periodic consolidation — transcript mining, knowledge synthesis, data quality, brain sync.
npx skillsauth add nicsuzor/academicops rememberInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Provides domain instructions for knowledge capture, persistence, and maintenance in the Personal Knowledge Base (PKB) ($ACA_DATA).
/remember)Write or Edit on files under $ACA_DATA/. Use the PKB MCP toolset.mcp__pkb__search for semantic discovery. Do not use Glob or Grep on $ACA_DATA/.mcp__pkb__create: Create a document (projects, contexts, meeting notes).mcp__pkb__create_memory: Add a memory/note.mcp__pkb__append: Extend an existing document.mcp__pkb__get_document: Read a document.mcp__pkb__search: Search before creating new content — and before asserting a fact. Don't assume; look it up./sleep loop.| Content | Write Via | Target Location |
| :--------------------------- | :----------------------------------------------------------- | :------------------------------- |
| Epics/projects | mcp__pkb__create(type="epic"\|"project") | projects/ |
| Tasks/issues | gh issue create (GitHub primary) | Synced automatically |
| Durable knowledge | mcp__pkb__create or mcp__pkb__create_memory | knowledge/, context/ |
| Session findings | mcp__pkb__update_task on parent task | Task body |
| Axioms / framework rules | Invoke /framework skill — do NOT write to brain/context/ | skills/ via framework workflow |
| Content | Location | Notes |
| :--------------------- | :---------------------------------- | :------------------------------------ |
| Project metadata | projects/<name>.md | Hub files. Hub-doc type is epic. |
| Project details | projects/<name>/ | Subdirectory. |
| Goals | goals/ | Strategic objectives. |
| Context (about user) | context/ | User preferences, history. |
| Sessions/daily | sessions/ | Daily notes only, type: daily-note. |
| General knowledge | knowledge/<topic>/ | Durable facts (non-user). |
| Meeting/call notes | knowledge/<topic>/ or projects/ | type: meeting-note. |
| Maps of Content (MOCs) | knowledge/ or topic subdirs | Navigational hubs, type: moc. |
$ACA_DATA/.$ACA_DATA/ with correct type frontmatter.Maintain exactly one canonical note per first-class topic (tools, projects, skills, agents, concepts).
mcp__pkb__append in the relevant section.Overview, Usage, Known Issues) only if the topic does not exist.During updates, search for duplicate or peer notes. Merge unique content, retire weaker versions, and update wikilinks. For how to retire — delete the superseded file and trust git history; do not leave superseded_by: pointers — follow the single canonical rule in [[consolidation-procedure]] Step 3.
mcp__pkb__search(query="topic").mcp__pkb__create(
title="Descriptive Title",
body="Content with [[wikilinks]].",
type="note" | "project" | "epic" | "knowledge" | "moc" | "meeting-note",
tags=["tag1", "tag2"]
)
Or for lightweight facts:
mcp__pkb__create_memory(
title="descriptive title",
body="content",
tags=["tag1"]
)
[org/repo#123](url)).## Relationships section:
## Relationships
- [related] [[task-id]] — description
- [upstream-bug] [org/repo#123](url)
For framework tasks, capture generalizable patterns (e.g., "Configuration should be set once at initialization") rather than local implementation details.
Factual claims in episodic notes must be formatted as Obsidian callouts:
> [!observation] Factual statement
> Source: [[source-note]]
> Confidence: <numeric value 0.0 to 1.0>
established (>= 0.8): verified, multiple sources.provisional (0.4 - 0.79): limited evidence.speculative (< 0.4): inference.type: knowledge)Include these frontmatter fields:
sources: (YAML list)synthesized: (ISO date)last_reviewed: (ISO date)confidence: (numeric 0.0 - 1.0)maturity: seedling | budding | evergreenUse MOCs to index and curate links to related notes when a topic area grows to 5+ notes.
title: "MOC: Topic Name"
type: moc
tags: [moc, topic-area]
created: YYYY-MM-DD
last_reviewed: YYYY-MM-DD
To capture memories asynchronously:
Task(
subagent_type="general-purpose", model="haiku",
run_in_background=true,
description="Remember: [summary]",
prompt="Invoke Skill(skill='remember') to persist: [content]"
)
State the PKB write details directly and concisely:
/sleep)Runs periodic offline consolidation on a GHA workflow or manual trigger.
consolidated: YYYY-MM-DD in frontmatter).First match determines mode:
SLEEP_MODE overrides default (short-loop | full-session).LOOP_INTERVAL_MINUTES <= 30 -> short-loop.short-loop.full-session.See [[references/maintenance-phases]] for detailed sub-agent rules, pacing, and exit protocols.
| Phase | Name | | :---- | :-------------------------- | | 0 | Graph Health Baseline | | 1 | Session Backfill | | 2 | Transcript Mining | | 3 | Episode Replay | | 4 | Knowledge Consolidation | | 5 | Index Refresh | | 6 | Data Quality Reconciliation | | 7 | Staleness Sweep | | 8 | Refile Processing | | 9 | Graph Maintenance | | 10 | Consolidation Self-Check | | 11 | Brain Sync |
When running on GitHub Actions, no PKB MCP server is available. Direct file tool operations (Bash, Glob, Grep, Read, Write, Edit) on markdown files are allowed. Do not edit files outside $AOPS_SESSIONS/ (except git commits/push).
tools
Streamlit implementation of the analyst presentation layer. Use when building or updating a Streamlit dashboard that displays pre-computed research data. This is the Streamlit-specific HOW for the tech-agnostic principles in the aops-tools analyst skill — display only, never transform.
tools
Python plotting and statistical-modelling libraries (matplotlib, seaborn, statsmodels) for the analyst presentation and statistical-methodology layers. Use when producing publication-quality figures or fitting statistical models in Python. Library-specific HOW for the tech-agnostic principles in the aops-tools analyst skill.
tools
dbt (data build tool) implementation of the analyst transformation layer. Use when a project has a dbt/ directory or you need to build, test, or document SQL transformations as version-controlled, reproducible dbt models. This is the dbt-specific HOW for the tech-agnostic principles in the aops-tools analyst skill.
development
Core academicOps skill — institutional memory, strategic coordination, workflow routing, and framework governance. Merges butler (chief-of-staff) with framework development conventions.