status.md
CI-OS Project Status
Last verified: 2026-07-14
Current Position
CI-OS is an Algolia-first Competitive Intelligence Operating System operated by Argus on Hermes. It is designed to connect three kinds of evidence:
- What competitors shipped.
- What the market is saying.
- What Algolia audiences are responding to.
The target output is an evidence-backed recommendation for Product, Product Marketing, Sales, Content, or executive teams.
Current product completion is approximately 35 percent. The engineering scaffold is approximately 65 percent complete. CI-OS is not launch-ready.
What Has Been Built
- Hermes extension package and Argus operating model.
- Multi-tenant competitor and source registry.
- Source collection and evidence ledger.
- Product event, conversation, demand, pattern, recommendation, and learning models.
- Competitor-specific briefs and monitored-competitor views.
- Product Muscle and demand work queues.
- Local administration for competitors, sources, product surfaces, demand, and runs.
- Dashboard renderer and selected browser-validation journeys.
- 1,251 passing local tests in the latest verified suite.
- Dedicated
ciosapplication user, secure Hermes queue handoff, and delegated cgroup containment.
Current Verified State
| Area | Status | Current evidence |
|---|---|---|
| Phase 0 baseline | Passed | Clean, published CI-OS recovery baseline. |
| Phase 1 Hermes execution | Passed | Two consecutive real Hermes runs exited 0 as cios, with no permission error, timeout, orphan work, or ownership drift. |
| Deployed package | Verified | Commit 1fa7ac5; immutable archive and rollback-bundle drill passed. |
| Competitor registry | Partial | 27 competitors represented. |
| Source coverage | Current | Latest Hermes run reports 43 active, 43 checked, and 0 failed sources. |
| Product reality | Partial | 12 historical product events; current product-surface extraction is not proven. |
| Market conversation | Present | 500 themes and 3 candidate patterns. |
| Audience demand | Blocked | No ready GA4 / Looker source; 0 demand signals. |
| Recommendations | Blocked | 0 current recommendations because required evidence is incomplete. |
| Learning | Unproven | No learning effect visible in the current run. |
| Production UI | In redesign | Accepted Product Muscle IA is not yet the live application. |
| Launch readiness | Not ready | Fresh run status is blocked_on_evidence; demand and later phase gates remain open. |
What Comes Next
- Make publication atomic, current-run bound, and safety-validated.
- Complete Scout-backed product and feature comparison after the publication gate.
- Connect GA4 / Looker audience-demand evidence after its predecessor gate.
- Prove one useful cross-plane Argus recommendation and one learning effect.
- Build the accepted Argus Read, Product Muscle Matrix, Conversation Heatmap, Demand Lens, Pattern Board, Actions, Registry, Evidence Lab, and Command/Admin workflows.
- Complete exhaustive frontend, backend, semantic, accessibility, security, and live-cron validation.
- Release a controlled Algolia pilot only after every gate passes.
Next Gate
Phase 2 is active. Every public status and launch result must describe one fresh, complete run; planted stale, mismatched, spoofed, partial-copy, self-attested, and path-leaking cases must fail before one complete run may publish.
Product Muscle, Audience Demand, Argus intelligence, and production UI work remain locked behind their predecessor gates.