Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
20 commits
Select commit Hold shift + click to select a range
50de4c3
docs(execplans): add execplan for schema fidelity and execution routi…
leynos Mar 23, 2026
e72bedb
test(hosted-remote-tools): add tests for schema fidelity and executio…
leynos Mar 23, 2026
4a009b2
test(worker_remote_tool_proxy): fix format string reference in execut…
leynos Mar 23, 2026
15e0ecb
test(api): add round-trip serialization tests for remote tool types
leynos Mar 23, 2026
5bf1998
test(worker/api): add serialization fidelity and round-trip tests for…
leynos Mar 24, 2026
d9af592
test(worker/api): refactor and modularize worker API tests
leynos Mar 25, 2026
2589f3c
test(orchestrator/api): add remote tool fidelity tests for schema and…
leynos Mar 26, 2026
3d3682b
test(worker_remote_tool_proxy): add ProxyTestServer fixture to simpli…
leynos Mar 26, 2026
ddf7687
refactor(tests): centralize complex tool definition in shared test_su…
leynos Mar 26, 2026
82a2837
test(worker_remote_tool_proxy): move tests to a dedicated tests module
leynos Mar 27, 2026
902b930
test(remote_tools): add transport parity and hosted fidelity tests
leynos Mar 27, 2026
a70c3a9
test(channels/webhook_server): enhance test fixtures and coverage for…
leynos Mar 27, 2026
f084073
test(worker): add failure-mode mock server for remote-tool API tests
leynos Mar 27, 2026
2f46087
Implement Drop for RemoteToolFailureServer to abort background task
leynos Mar 27, 2026
007399f
Update execplan status and fix JSON parsing in test
leynos Mar 27, 2026
9d168e5
Return Result from started_webhook_server fixture and consuming tests
leynos Mar 27, 2026
75b2efd
Extract hosted_catalog_harness fixture for worker container tests
leynos Mar 27, 2026
a675be7
Expand BDD acronym in execplan constraints
leynos Mar 27, 2026
64af506
Fix markdownlint issues and duration field naming in execplan
leynos Mar 27, 2026
f2a3fe2
Align execplan with transport contract field names and update validat…
leynos Mar 27, 2026
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions docs/contents.md
Original file line number Diff line number Diff line change
Expand Up @@ -119,6 +119,9 @@
- [Merge remote Model Context Protocol (MCP) tool definitions into the worker reasoning context](execplans/1-1-3-merge-mcp-defs-into-worker-reasoning-context.md)
plans roadmap item `1.1.3` for the worker-side merged hosted-tool
reasoning surface.
- [Add hosted-mode tests for schema fidelity and execution routing](execplans/1-1-4-tests-for-schema-fidelity-and-execution-routing.md)
plans roadmap item `1.1.4` for the test matrix that locks down MCP tool
contract fidelity and execution routing guarantees.

## RFCs

Expand Down

Large diffs are not rendered by default.

16 changes: 9 additions & 7 deletions docs/rfcs/0001-expose-mcp-tool-definitions.md
Original file line number Diff line number Diff line change
Expand Up @@ -5,13 +5,15 @@
- **RFC number:** 0001
- **Status:** Proposed
- **Created:** 2026-03-11
- **Implementation status:** Roadmap items `1.1.1`, `1.1.2`, and `1.1.3` are
implemented in this branch through the shared `src/worker/api/` transport
types, the worker catalog-fetch startup path, the orchestrator generic
remote-tool execution endpoint, the canonical `ToolRegistry`-owned
hosted-visible filter for active MCP tools, and the explicit worker-side
merged reasoning surface used both at context build and later refresh.
Roadmap item `1.1.4` still owns the broader schema-parity and routing matrix.
- **Implementation status:** Roadmap items `1.1.1`, `1.1.2`, `1.1.3`, and
`1.1.4` are complete. The implementation includes the shared `src/worker/api/`
transport types, the worker catalog-fetch startup path, the orchestrator
generic remote-tool execution endpoint, the canonical `ToolRegistry`-owned
hosted-visible filter for active MCP tools, the explicit worker-side merged
reasoning surface used both at context build and later refresh, and the
comprehensive test matrix for schema fidelity, execution routing, and
worker-orchestrator contract parity. All roadmap items in section `1.1` are
now complete.

## Summary

Expand Down
7 changes: 1 addition & 6 deletions docs/roadmap.md
Original file line number Diff line number Diff line change
Expand Up @@ -87,7 +87,7 @@ worker-orchestrator contract hardening belongs inside 1.1.1.
- Success: hosted model requests include the real tool descriptions and JSON
Schemas, and worker-local tools plus orchestrator-owned tools appear as one
unified tool surface.
- [ ] 1.1.4. Add hosted-mode tests for schema fidelity and execution routing.
- [x] 1.1.4. Add hosted-mode tests for schema fidelity and execution routing.
Requires 1.1.3.
- See [RFC 0001 §Migration Plan](./rfcs/0001-expose-mcp-tool-definitions.md#migration-plan).
- Tracks the worker-orchestrator parity portion of
Expand Down Expand Up @@ -1339,7 +1339,6 @@ provenance, transformation policy, and per-channel capability negotiation before
place, unsupported media falls back predictably, and automated tests cover
preview, cache reuse, transcription, TTS, and transformed-media flows.


## 7. Raise assurance for safety and lifecycle invariants

Phase objective: add proof-oriented and generated verification where Axinite's
Expand Down Expand Up @@ -1405,7 +1404,6 @@ host-matching contract used by installer- and allowlist-adjacent code.
and make installer regressions fail on structured generated input instead of
one-off examples only.


### 7.3. Bounded checking for allowlist and host-matcher semantics

Objective: use Kani where the highest-value invariants are small, deterministic,
Expand Down Expand Up @@ -1447,7 +1445,6 @@ feeds 7.5 if the extracted matcher proves stable enough for a later full proof.
explores larger host and path combinations, and unwind bounds stay close to
the harnesses they justify.


### 7.4. Model-check the job lifecycle with Stateright

Objective: model the scheduler, worker, token, reaper, and retained-result
Expand Down Expand Up @@ -1486,7 +1483,6 @@ contract work exposes adjacent semantic cleanup patterns.
scheduled runs can extend depth and state count, and any safety-property
counterexample is surfaced as a dedicated formal-verification failure.


### 7.5. Add a narrow later-stage Verus proof path

Objective: keep a proof-only path available for the few invariants that remain
Expand All @@ -1509,7 +1505,6 @@ production code have already converged on one matcher contract.
suffix spoofing remains impossible in the proof model, and proof execution
stays isolated from the normal Cargo-driven test path.


### 7.1. Formal-verification infrastructure and workflow split

Objective: add the repository structure, tool runners, and CI split needed for
Expand Down
Loading
Loading