Status — PRISM

Last updated: 2026-05-08

Current Understanding

PRISM is the productization of the algolia-search-audit skill ecosystem. The current skill workflow already encodes a complete prospect research and search audit methodology, but it runs through local skills, file-based scratchpads, subprocess scripts, and manual/agent orchestration. PRISM should make that repeatable, durable, governed, and usable by Algolia teams.

Current Repo State

The repo has: - Python backend under prism_platform/ - FastAPI routers under prism_platform/api/routers/ - Temporal-oriented orchestration under prism_platform/orchestrator/ - v2 agentic module framework under prism_platform/v2/ - Next.js frontend under frontend/ - Tests under tests/ and tests/v2/

The worktree is dirty. Many old v1 prism_platform/modules/* files are deleted, while v2 modules and tests are present. Treat existing changes as user/session work unless explicitly told to clean them.

Current v2 Registered Modules

From prism_platform/v2/registry.py, the registered modules are:

PRISM v2 module	Status in registry
intel-company	Registered
intel-techstack	Registered
intel-traffic	Registered
intel-financial-public	Registered
intel-financial-private	Registered
intel-news	Registered
intel-hiring	Registered

Not all original skill-system modules are represented yet.

What The Current Skill System Already Produces

The baseline skill system produces: - Research scratchpads in research/ - JSON sidecars for machine-readable downstream use - Browser screenshots - Browser findings - Scoring matrix - audit-data.json - SPA report - AE report - battle card - prospect leave-behind - PDF - AE playbook - strategic signal brief - ABX campaign package - factcheck report, correction manifest, gate verdict, and skill feedback

Important Correction

Scout and Discovery OS are not the origin point of PRISM. They are future components to incorporate after the SaaS baseline is understood.

The origin point is the multi-skill Algolia Search Audit workflow.

Known Gaps

• PRISM needs a complete mapping from all current skills to SaaS modules.
• The current repo has some v2 modules but not the full parent skill pipeline.
• The frontend direction should become chat-first/copilot-first rather than traditional navigation-first.
• Current deliverable outputs should be studied from screenshots/examples before UX planning.
• Authentication and authorization are product requirements, but exact roles and rollout policy are not yet captured.
• Data model needs to preserve source provenance, evidence tiers, claim registry, and factcheck gates.