cogmemai-memory — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited cogmemai-memory (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
CogmemAi gives your Ai assistant persistent memory across sessions. This skill teaches you how to use it effectively — when to save, what to save, and how to keep your memory organized.
Always call get_project_context before responding to the user's first message. This loads your most important memories from previous sessions. Pass an optional context parameter describing the current task for better relevance.
Save memories when you learn something important:
Use extract_memories after important conversations instead of manually saving each fact. It automatically identifies what's worth remembering:
This is faster and more thorough than manual save_memory calls for multi-fact exchanges.
When the session ends or the user says goodbye, save a session summary capturing what was accomplished, decisions made, and next steps.
Save immediately when:
Do NOT save:
| Score | Use For | Examples |
|---|---|---|
| 9-10 | Core architecture, identity | "This is a Next.js 14 app with App Router", "User is a senior engineer" |
| 7-8 | Key decisions, preferences | "Always use Bun instead of npm", "Auth uses Supabase" |
| 5-6 | Patterns, conventions, bugs | "CSS uses BEM naming", "Login bug was a race condition" |
| 3-4 | Minor context | "Prefers dark mode in VS Code", "Last deployed on Friday" |
| 1-2 | Trivial | Rarely useful — usually not worth saving |
Choose the right type for better retrieval:
Use promote_memory to move a project memory to global scope when you discover it applies universally.
When starting work on a new project for the first time:
get_project_context to check for existing memoriesingest_document to extract key facts from documentationPeriodically (or when the user asks), review memory quality:
health_score from get_project_context — a 0-100 score showing overall memory healthget_analytics to see usage patterns — it automatically tunes importance and archives stale memoriesconsolidate_memories to merge related facts into comprehensive summarieslink_memories to connect related factsget_stale_memories to find outdated memories that need reviewUse tasks for work that spans multiple sessions:
save_task to create a task with status (todo, in_progress, done, blocked) and priorityget_tasks at session start to see what's pendingupdate_task as you make progresssave_correction when you discover a wrong approach — stores the wrong-to-right pattern so the mistake isn't repeatedset_reminder to surface a nudge at the start of the next session (auto-expires after delivery)When Claude Code compacts your context (you'll see a summary of the previous conversation):
get_project_context immediately to reload memoriesWhen you need to find something specific from past sessions:
recall_memories with a natural language querymemory_type or category if you know the typeextract_memories instead of multiple save_memory calls. It's faster and catches facts you might miss.update_memory to revise existing memories rather than creating duplicates. The system detects conflicts automatically.bulk_update instead of individual calls.get_project_context at session start — This is the single most important action for continuity~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.