project-context

Creates and refreshes a durable project context file with digest, glossary, and Know/Don't-Know ledger.

Updated Jul 10, 2026
One-click install
npx skills add https://github.com/subvisual/the-a-team --skill project-context-subvisual
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: project-context
Source: https://github.com/subvisual/the-a-team/tree/main/.claude/skills/project-context
Command: npx skills add https://github.com/subvisual/the-a-team --skill project-context-subvisual

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Teams lose track of project facts scattered across meetings, documents, and conversations, causing terminology drift and inconsistent downstream artifacts. This Skill consolidates all durable project knowledge into a single canonical docs/product/context.md file that every other skill and agent reads. ## Core Features & Use Cases - Context Seeding and Refresh: Creates or updates docs/product/context.md containing the overview, evidence digest, source index, glossary, design and technical context, and a Know/Don't-Know ledger. - Evidence Ingestion: Digests raw input batches from docs/product/input/ with line-level citations, staging external documents and connector pulls (Notion, Granola, Slack) verbatim before citing them. - Glossary and Conflict Tracking: Stabilizes domain terminology with settled/forming/TBD statuses and records disagreements between sources as explicit [conflict] ledger entries. - Use Case: After a client call adds new requirements, run this Skill to stage the call notes, digest them with citations, update the glossary, and flag blocking unknowns before writing a PRD. ## Quick Start Ask the agent to seed or refresh the project context for the target repository using the project-context skill, ingesting any new evidence batches in docs/product/input/.

Frequently Asked Questions about project-context

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a project context file for a new product?▼

Invoke the Skill with a target repository or project folder and whatever materials you have, such as docs, links, or conversation notes. It seeds docs/product/context.md with an overview, glossary, and ledger, keeping unknowns visible as TBD rather than guessing.

How to ingest meeting notes and documents into project context?▼

Drop raw evidence into docs/product/input/ as dated batches, or point the Skill at external files which it stages verbatim with a SOURCE.md first. Each batch is digested with line-level citations and indexed in the Sources table with coverage records.

Does this Skill work without a git repository?▼

Yes, it resolves the target repo-first but falls back to a plain project folder when no repository exists. The write path stays docs/product/context.md either way; git commits happen only where a repo exists, one commit per run.

When should I not use project-context?▼

Do not use it for repo scaffolding, CI setup, writing discovery plans, PRDs, or epics, or for minting Jobs to be Done. It records finished JTBDs by id and marks missing ones TBD; the jobs-to-be-done skill owns creating them.

What happens if no human is available to review the context?▼

The Skill escalates instead of guessing: it writes only blocking questions under an Awaiting answers section in context.md and halts. The human answers inline and re-invokes, and the section is removed as answers are ingested.