Agent Harness Bundle
AI coding infrastructure
Long coding sessions lose context, and parallel agents can overwrite each other's work. I built a shared toolkit that gives coding agents common rules, tasks, and file access.
The problem
A single coding agent can lose the important rules as the session grows. Multiple agents create a second problem: two panes can edit the same file, repeat work, or accept a change that fails the project checks.
What I built
I designed the shared-brain contract, the front-end adapters, the install/update lifecycle, and the operating field guide. The bundle assembles existing tools, skills, platform capabilities, and ecosystem projects into one coherent environment.
Engineering decisions
Durable rules stay separate from model-specific settings. Task graphs make dependencies visible, reservations remain advisory until a commit guard checks them, and the Windows/WSL bridge treats credentials, hot data, and service health as operational concerns.
Why it matters
The useful idea is structural: move uncertainty earlier, make work reviewable, and leave an update path that does not depend on one front end or one machine.
Attribution boundary
The repository records direct contributions, partner contributions, marketplace skills, and external projects. The portfolio presents Kamal's integration and synthesis without claiming authorship of every tool inside the bundle.
See how the environment stays coherent across tools and machines.
This is an integration project, not a claim that one person wrote every included tool. Explore the shared context, planning loop, coordination boundaries, Windows/WSL bridge, and provenance rules.
A coherent environment, not a claim of ownership
Explore the layers Kamal assembled and the boundaries that keep the project honest.
Integration boundary
The bundle assembles existing tools, skills, platform capabilities, and ecosystem projects. Kamal's contribution is the integration contract, adapters, lifecycle, and operating field guide.
External tools are credited where their source is known.
Partner and ecosystem contributions remain visible.
The project does not claim authorship of every included skill.
direct contribution: architecture + adapters + lifecycle
integrated contribution: external tools and skills
public evidence: repository history + provenance notesOne canonical brain reduces drift between front ends.
Direct work, partner contributions, and external components are not collapsed into one claim.
Install, update, drift checks, and smoke tests are part of the operating model.
The bundle is open source. Its strongest story is the integration and operating discipline, not a claim that Kamal wrote every tool it contains.