Teach it once.
Every agent remembers.
Decisions, rules, facts and how-tos link up into a memory graph around each project. Before every reply, BrickYard finds the notes that matter, by meaning and not just keywords, and hands them to whichever agent you're using.
- Explore it as a graph, or as Markdown in Obsidian
- See exactly which notes each answer used
- Bring in CLAUDE.md, AGENTS.md and Cursor rules
Chores that do
themselves.
Describe a job in a sentence, like "weekdays at 9, review new pull requests". BrickYard runs it on time with the agent you pick, and only bothers you when it needs a decision.
- Templates: PR review, dependency watch, changelog, digests
- Follows your permission rules while you're away
- A nightly routine keeps your memory tidy
Agents that work
as a team.
Give a big job to a team: one agent writes, another tests, a third reviews. You get one answer back, and nothing lands until your tests pass.
- Mix agents: Claude writes, Codex tests, Gemini reviews
- Writers get their own branch; reviewers only read
- Set a spend cap for the whole job
Everything they make,
kept in Studio.
Ask for a deck, a dashboard or a tiny app. It opens beside the chat, and Studio keeps every version, so nothing gets lost in a scroll of messages.
- Docs, sheets, charts, slides, diagrams, apps and prototypes
- Pin a comment on the canvas to ask for a change
- Three variants side by side, or one from each agent
A browser your
agents can drive.
Every chat gets its own browser. Agents open pages, click, type and read them to test your app or handle web chores, and you watch it live.
- Passwords go through a secure form the agent never sees
- Console, network and page scripts for real debugging
- Sites you sign in to stay signed in