5 Commits
Author SHA1 Message Date
weichen ea2689641f Thread last-request and cache usage fields through the exec event bridge (N3 WP6)
ci / vet-and-test (pull_request) Successful in 51s
ci / docker-checkout-hostgateway (pull_request) Successful in 1s
Why: squad consumes exec.ProcessEvent, which lacked the split usage
fields, so the WP6 accounting could not reach squad runs.
Evidence: GOWORK=off go vet green; additive fields on ProcessEvent.
Impact: additive only.
Refs: Agent-Native#35 #37, N3 plan WP6
2026-10-07 17:10:53 +00:00
weichenandCursor 4a10818535 Drop the intent adapter and fold contract into exec
Why: native must compile without agent-intent, and the two execution packages were the same contract under two names.
Evidence: GOWORK=off go test -count=1 ./... passed, including schema.TestRequestAndResultRoundTrip and the native adapter tests after Options was split into embedded sub-structs.
Impact: adapters/intentgate is gone, go.mod no longer requires agent-intent, and the nested gitlink stays for the release task. Callers of Options composite literals name the sub-struct.
Refs: D1
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-10-03 13:31:48 +00:00
weichenandCursor 951c141f95 Let hosts register tools and observe token usage
Why: product runners need register_artifact and card tools inside the kernel loop, and the UI needs token usage on the same event stream as text and tool calls.
Evidence: go test ./... in agent-native; TestHostToolAndUsageEvent and TestHostToolNameConflictFailsClosed.
Impact: Options.HostTools fail closed on name clashes; ProcessEvent and runtime events gain a usage type with token counts. Existing callers that ignore unknown event types stay compatible.
Refs: deliverable-quality-system B1
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-10-01 08:57:21 +00:00
weichen dd2ac11b31 Extend exec.Request with capability fields and enforce budgets end to end
Why: the frozen exec port could not express per-request model, tool
allowlist, skills, budget, or identifiers, so squad roles and product
profiles had no way to steer one execution; contract v0.1.0 semantics
(budget/cancel) stayed unconsumed (parity-matrix #4/#10/#16).

Evidence: go vet ./... clean; go test ./... all ok, including new
loop budget caps (tool calls, prompt tokens, deadline), cancel
propagation into a blocking tool, registry Restrict fail-closed
policy tests, native adapter v2 tests (model override, allowlist,
deadline-bound context, execution-id log header, skills prompt line),
and settle classification marking budget_exceeded permanent.

Impact: exec.Request gains optional Model/ToolAllowlist/Skills/Budget/
SessionScope/ExecutionID/Env (zero value = legacy behavior); runtime
loop enforces a shared budget and emits budget_exceeded events;
adapters/native consumes all new fields. Consumers on the old port
keep compiling and behaving identically.

Refs: docs/parity-matrix.md rows #4/#10/#16 (Batch2 scope); deps/
development-guidelines standards/agent-execution.md (shared budget
across retry layers).
2026-10-01 05:26:19 +00:00
weichenandCursor fedf671a48 Import the current agent-native module as a fresh history.
ci / vet-and-test (push) Successful in 5s
ci / docker-checkout-hostgateway (push) Successful in 3s
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-29 12:29:04 +00:00