Connect your OpenClaw agent to your Konteks account (konteks.app) for persistent memory, task management, and context sharing. Use when you need to store agent memories, create or read tasks/notes, check projects and folders, read daily plans, or sync context between conversations. Requires a Konteks API key from konteks.app/dashboard/settings.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
Connect your OpenClaw agent to your Konteks account (konteks.app) for persistent memory, task management, and context sharing. Use when you need to store agent memories, create or read tasks/notes, check projects and folders, read daily plans, or sync context between conversations. Requires a Konteks API key from konteks.app/dashboard/settings.
when
User asks to create/manage tasks, store memories, check projects, read daily plans, or manage notes in Konteks
examples
["Create a task to review the PR","What's on my agenda today","Remember that I prefer dark mode","Check my inbox for new items","What projects do I have"]
skills:konteks:apiKey:"sk_..."url:"https://konteks.app"# optional, defaults to thisagentId:"my-agent"# optional, defaults to "default"
API Base
All endpoints: {url}/api/agent/...
Auth header: Authorization: Bearer {apiKey}
Hybrid Memory Architecture
When this skill is installed, use a hybrid approach to memory — Konteks DB for shared structured data, local files for fast scratchpad notes.
Konteks DB (shared source of truth)
Use Konteks for anything your human should also see or that needs to persist across devices:
Tasks & items — created by human or agent, visible in the iOS and web apps
Projects & areas — the organizational structure
Decisions — why something was chosen (write to agent_contexts with category decision)
Preferences — user preferences you learn over time (category preference)
Learnings — things you discover that matter long-term (category learning)
Project notes — context about specific projects (category project_note)
Local markdown (agent scratchpad)
Keep using local files (memory/*.md, MEMORY.md) for:
Daily session logs — raw notes about what happened today
Working context — things you're actively thinking about this session
Drafts & scratch work — ideas in progress before they're ready for Konteks
How they work together
Session start: Pull recent memories from Konteks (GET /api/agent/context?limit=10) to restore context. Also read local daily files as usual.
During conversation: Take quick notes in local markdown. When something important is decided or learned, write it to Konteks too.
After important moments: Write decisions and learnings to Konteks so they persist in the app and survive local file changes.
Heartbeats: Check Konteks for inbox items and due tasks. Update local heartbeat state.
If Konteks is unreachable: Fall back to local files. You should always be able to function without network access.
This way your human sees important context in their Konteks app, and you have fast local access for working memory. Neither system is a single point of failure.
Proactive Behavior (Heartbeats)
When this skill is installed, the agent should proactively use Konteks during heartbeats. Add these checks to your heartbeat routine:
1. Inbox Triage
Check the inbox for new items and file them into the right project/area if obvious:
## Konteks Checks- [ ] Check inbox for new items — triage if obvious, leave if not
- [ ] Check for due/overdue tasks — alert if urgent
- [ ] Write any recent decisions/learnings to agent_contexts
Frequency: Check inbox and due items 2-3 times per day during heartbeats. Don't check every single heartbeat — rotate with other checks.
Agent Memory (agent_contexts)
Store and retrieve persistent memories, decisions, preferences, and learnings.
Write/update memory:
curl -X POST "{url}/api/agent/context" \
-H "Authorization: Bearer {apiKey}" \
-H "Content-Type: application/json" \
-d '{"category":"memory","key":"user_preference","value":"Prefers dark mode","agent_id":"{agentId}"}'
