Session — 2026-05-08 — Skill System Orientation

Purpose

Re-orient PRISM around its actual origin: the SaaS conversion of the algolia-search-audit parent skill and its subskill ecosystem.

Sources Read

• Parent skill: algolia-search-audit/SKILL.md
• Platform constitution: algolia-search-audit/AGENT-CONTEXT.md
• Intelligence skills: company, techstack, traffic, competitors, financial public/private, investor, hiring, social, news, partner, industry, queries
• Audit/delivery skills: browser, report, factcheck, eval, campaign ABX
• Key scripts: collection scripts, browser audit, renderer, validator, deterministic score/ROI/audit-data patchers
• PRISM repo orientation files and v2 module registry
• Crawler Factory vault structure as reference

Key Learning

PRISM should not be planned as a new greenfield research system. It should be planned as a productized execution environment for an already-rich audit methodology.

The current skill system has a strong operational contract:

• module wave ordering
• source-specific collection rules
• file and JSON output contracts
• evidence labels
• verification gates
• scoring rules
• report renderer assumptions
• factcheck gate before publish

Important Correction Captured

Scout and Discovery OS are future PRISM components, not the baseline definition of PRISM.

The baseline definition is:

PRISM = deterministic, persisted, authorized SaaS execution of the Algolia Search Audit skill pipeline.

Files Drafted

• README.md
• Wiki/README.md
• Wiki/Status.md
• Wiki/Architecture.md
• Wiki/Skill-System-Baseline.md
• Wiki/Module-Map.md
• Wiki/Product-Strategy.md
• Wiki/Open-Questions.md

Next Discussion

Before implementation planning, review:

1. Current report output screenshots and examples
2. Which skill outputs are still essential vs legacy
3. What v1 SaaS user journey should be with chat-first/copilot UX
4. Which current PRISM v2 modules are faithful enough to replace their source skills