ticket_id: "tkt_pmem_bootstrap_project_ingest"
title: "Project knowledge is bootstrapped into distilled project memories from repo context"
agent: "codex"
done: false
goal: "The configured project starts with a reusable set of distilled project memories built from repo context rather than raw file dumps."
Tasks
Run one project ingest pass using the source inputs: repo tree, architecture summary, important configs, current plans or tickets, and standing conventions.
Convert that ingest material into distilled project memories rather than storing raw file dumps.
Store only the memory types called out in the source: architecture invariants, important commands, service boundaries, persistent constraints, known failure modes, and active goals.
Acceptance criteria
An initial project memory set exists for the configured project namespace.
Stored project memories are distilled and reusable rather than raw repo content dumps.
The bootstrap covers the source categories for project knowledge.
Tests
Retrieve project memories after bootstrap and verify entries exist for the listed source categories.
Review the stored entries and verify they are summarized durable facts rather than copied file contents.
Notes
Source: "Bootstrap memory from the repo once" using repo tree, architecture summary, important configs, current plans/tickets, and standing conventions.
Constraints:
"Automatic memory from repo files directly is not native here."
"You need a bootstrap pass that summarizes the project into memory-worthy facts."
Evidence:
The source explicitly lists the desired distilled outputs: architecture invariants, important commands, service boundaries, persistent constraints, known failure modes, and active goals.
Dependencies:
TICKET-100-project-scoped-memory-writes.md
Unknowns:
Exact source materials to ingest for the current project were not provided.