For end users

knowledge-digest

Read everything in knowledge/raw/ and produce a single insights.md other skills consume.

What it does

You drop your own context into pectusai/knowledge/raw/: AnswerThePublic exports, BigQuery results, screenshots, audience research PDFs, anything useful. The skill reads all of it and writes a single knowledge/insights.md that downstream skills (weekly-analysis, plan-sitemap) treat as a stable input.

Your other skills don’t read the raw files; they read this digest. That keeps each skill’s input cheap and gives you one editable file to inspect or hand-correct.

Inputs

• workspace_id

Outputs

• insights_md — written to knowledge/insights.md (or knowledge/<workspace_code>/insights.md for workspace-scoped digests)

When to run

Whenever you add new files to knowledge/raw/. The output is cached as a stable input across other skills, so successive runs in a 5-minute window read it from cache instead of re-paying.

Where it lives

pectusai/skills/knowledge-digest/SKILL.md