Conversation
hachall
pushed a commit
that referenced
this pull request
Jan 23, 2026
* ci: add in e2e-test-docker * fix: remove sops * ci: remove credential unpacking * fix: only test clay api * ci: use http for hurl tests * ci: bugfixes to e2e tests
3 tasks
sejori
added a commit
that referenced
this pull request
Apr 29, 2026
Adds the production-shaped bridge between dwctl's existing tool
registry and onwards' StepExecutor trait, plus integration tests that
drive the full multi-step lifecycle against real HTTP via wiremock.
Architecture (dwctl/src/responses/step_executor.rs):
- DwctlStepExecutor wraps the existing HttpToolExecutor (reused, not
re-implemented — same analytics, headers, timeouts, JSON shape as
the single-step path).
- dispatch_tool_call reads tool_sources.kind:
'http' → HttpToolExecutor::execute → ToolDispatch::Executed(_)
'agent' → ToolDispatch::Recurse (loop recurses into sub-loop)
No tool-kind knowledge leaks into onwards.
- execute_model_call goes through a pluggable ModelCaller trait. The
test uses StaticModelCaller pointing at a wiremock; production
wiring through onwards' load balancer is the focus of COR-349.
Surfacing tool kind:
- ToolDefinition gains a `kind: String` field.
- resolve_tools_for_request query selects ts.kind so it propagates
through ResolvedToolSet to the executor without an extra round-trip.
- Existing HttpToolExecutor ignores the field — no behavior change to
the single-step path.
Integration tests (dwctl/src/test/multi_step_executor.rs):
- dwctl_step_executor_drives_real_tool_and_model_calls_against_wiremock:
full model→tool→model lifecycle. Two wiremock servers (model + tool),
real DwctlStepExecutor, real FusilladeResponseStore against the
fusillade schema. Asserts both wiremocks received the expected
number of POSTs and the chain persisted with correct kinds,
sequences, and prev_step_id chaining. The tool's wiremock body
ends up verbatim in step #2's response_payload.
- dwctl_step_executor_dispatches_subagent_tools_via_recurse: registers
a kind='agent' tool, asserts the loop recurses into a sub-loop
scoped under the spawning step rather than firing HTTP. Sub-loop
step's parent_step_id points at the top-level tool step exactly
as plan §C11 specifies.
Both tests pass. The bridge is now self-contained: every type lives
either in dwctl (the tool registry, executor, transition function) or
onwards (the loop, traits, dispatch), with no leakage in either
direction.
8 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adding in docker e2e which spins up the docker compose production version and runs a suite of hurl tests at them.