Agent Directory & Field Profiles
Inspect active company-agent profiles, future specialist placeholders, source allowances, constraints, and static runtime snapshots grounded in read-only evidence.
Specialist Roster
Ubundi Chief of Staff
Orchestrator: intake, routing, source freshness, review, proof closure, and team-readable closeout.
Ubundi Research Agent
Research specialist: source-grounded briefs, evidence audits, repo/doc synthesis, vendor diligence, and uncertainty labels.
Ubundi Work Agent
Execution specialist: presentations, reports, docs, PRs, patches, scripts, work papers, action registers, and reviewed repo/doc changes.
QA / Review AgentPlaceholder
Validation specialist: checks artifacts, data, specs, or docs against rubrics and constraints. (Future Placeholder)
Comms AgentPlaceholder
Communications support: drafts internal or external updates after human approval. (Future Placeholder)
Facilities & Vibe AgentPlaceholder
Office automation: office music, vibe scheduling, ambient telemetry. (Future Placeholder)
Ubundi Chief of Staff
Core Role
The company-owned Chief of Staff for Team Ubundi. It keeps the CoS context ledger current, scopes incoming Slack asks, routes specialist work, reviews handbacks, and closes the original requester thread with concise next steps.
Ask this agent when
- You want to kick off a request that may need routing
- You need an active context check across approved company sources
- You want CoS to review a specialist output before sharing it
- You need a decision, open question, or source change captured in the ledger
Do not ask when
- You need deep research that belongs with Research Agent
- You want decks, reports, code, scripts, or work papers produced directly
- You need PR branches, repo patches, or file conversions executed
System Boundaries & Limitations
- Does not silently do specialist research or artifact production when a routed specialist is the right owner
- Does not merge PRs, deploy builds, or bypass human approval gates
- Does not use Company Agents Hub, context maps, manifests, or Slack canvas projection as active source truth
- Keeps long proof, drafts, and tables out of employee-facing Slack
Tools Accessible
Primary Sources
Coordination & Handoff Rule
Teammate request -> CoS scopes and routes in #company-agents -> specialist executes -> CoS reviews -> CoS closes the original thread.
Agent Repository
Static navigation link to the configured Gitea repository. Raw workspace files are not rendered in V1.
kwanda-agents/ubundi-chief-of--dseeiougxldk2cjs-02aed8c @ main
Operational Observation
Captured on: 2026-05-28 11:18:59 UTC by Codex read-only SSM audit
Snapshot Facts
Instance ID
i-02aed8c856c87518b
Running t4g.medium in us-east-1a; SSM Online on Ubuntu 24.04; setup-status=ready and assignment-status=assigned.
Workspace Commit
287540f workspace: normalize role-specific soul context
Workspace clean on main...origin/main at /home/ubuntu/.openclaw/workspace.
Primary Model
openrouter/deepseek/deepseek-v4-pro
Fallback: amazon-bedrock/global.anthropic.claude-sonnet-4-6.
Gateway
OK (130ms)
openclaw.service active; Slack configured; WhatsApp Cloud API configured.
Context Tasks
AM heavy, PM heavy, hourly light
Kwanda Ops showed CoS Context Heavy Refresh AM, Heavy Refresh PM, and Light Refresh Hourly as recurring.
Caveats & Warning Points
- Hub, context maps, per-agent manifests, Hub API submission, and Slack canvas projection are retired and must not be treated as current context truth.
- This website stores a curated static projection only; no runtime reads are initiated by the site.