turns recurring agent failures into reviewable PRs (propose-only)

Self-healing: persistent agent failures become ready-to-merge PRs instead of manual debugging.

ran 19m ago

routes each AI call-site to haiku/sonnet/opus (data-only)

Cuts pulse's AI bill by routing each call-site to the cheapest model that's good enough. advisory · unused

ran 19m ago

assigns/removes tasks at each user's edge of ability and pushes that edge (data-only); scores itself 0–100

Keeps everyone working at the edge of their ability and pushes it — DMs each person their edge task + a question on Discord, makes sure every task is handled, and scores itself in one number. advisory · unused

ran 19m ago

Runs the warm daily Discord check-in with each team member — guiding their growth toward higher compensation at a comfortable pace (what they learned today, what they'll do better tomorrow toward their goal) — and feeds what it hears back to the MD. The ultimate goal of the check-in is to grow each team member toward HIGHER COMPENSATION at a comfortable, well-guided pace — 'comfortable' meaning they feel actively guided on their growth, never pressured. Each check-in should help the person learn, set a small concrete next step toward their goal, and feel supported. Engagement (how much they open up) is a means to that end, not the end itself.

metric reported