ambient-agent

Implement or repair one planned GitHub issue in a shared workspace, and author its pull request only when the deterministic coordinator assigns publication.

Exercise and judge one exact pull-request head, then submit one formal review without repairing or merging.

Produce the ordered implementation and behavioral verification plan that gates a new-issue coding run before any edits begin.

Verify that a code change actually does what it's supposed to by exercising it end-to-end and observing behavior — drive the affected flow, not just tests or typecheck. Run before committing nontrivial changes; bootstraps this repo's project verify skill if none exists yet. Don't invoke it on a diff that only touches tests, docs, or other code with no runtime surface to drive (a change to product source always has one) — there's nothing to observe.

Behavior-verify the ambient-agent runtime by driving the real WhatsApp thread → GitHub, on the live code-factory rig. Use to prove a change works end-to-end (Coder opens a PR, Speaker tools act, Scribe writes the graph) — not to run unit tests.

Extract the shared-graph ontology from a chat's batched inputs — people, threads, GitHub objects, topics, interest edges, and commitments — recording honestly, not certainly.

Capture bug and feature reports as conversations, file complete issues, and return issue and pull-request links to the managed chat.

Apply this agent's ratified identity and WhatsApp teammate behavior to every accepted managed-chat Window.