chore: remove stale docs, gitignore, and Makefile entries#329
Merged
Conversation
- CONTRIBUTING.md: drop 'make run-remote' (no such Makefile target) - .gitignore: remove orphaned entries (local/remote_agent, sandbox-router, tasklog, axepp, google-cloud-sdk, test-sandbox-config.yaml) - Makefile: add missing deps, clean-logs to .PHONY
joycel-github
marked this pull request as ready for review
July 17, 2026 03:17
rakyll
approved these changes
Jul 17, 2026
feifeigood
added a commit
to feifeigood/ax
that referenced
this pull request
Jul 20, 2026
feifeigood
added a commit
to feifeigood/ax
that referenced
this pull request
Jul 20, 2026
…oogle#329 Upstream google#329 ('remove stale docs, gitignore, Makefile') accidentally checked in a 30MB linux/amd64 e2e ELF binary. Keep it out of the AgentFleet fork. Removal-neutral: re-added automatically on the next upstream merge if reverted.
feifeigood
added a commit
to feifeigood/ax
that referenced
this pull request
Jul 20, 2026
merge: sync upstream ax through 5d0e245 (google#329 google#331 google#327)
rakyll
pushed a commit
that referenced
this pull request
Jul 21, 2026
## Summary
Introduces `agentExecutor`, a lightweight executor for nested sub-agent
calls. Fixes the infinite loop when a sub-agent emits a confirmation,
and eliminates a wasted Gemini API call on every sub-agent delegation.
## Changes
### `internal/controller/executor/agent_executor.go` (new)
`agentExecutor` implements `agent.Executor` for sub-agent invocations:
- Does NOT load history from event log (caller passes messages)
- Does NOT short-circuit on `WaitsForConfirmation` or `COMPLETED` cache
- Writes execution-level logs for observability
- Returns `STATE_PENDING` when sub-agent emits a confirmation
### `internal/controller/executor/executor.go`
Child executor in `defaultExecutor.exec` changed from `defaultExecutor`
to `agentExecutor`. Also renames receiver `tm` → `de`/`ae` for clarity.
### `internal/gemini/gemini_planner.go`
After `e.Exec` returns in `loop`:
- Appends sub-agent outputs to `start.Messages` immediately (eliminates
wasted Gemini call — was 3, now 2)
- Respects `STATE_PENDING` (stops the infinite loop, surfaces
confirmation to user)
- Handles nil Gemini content gracefully (preserves `FinishReasonStop`
check, reports other reasons as errors)
## Confirmation resume routing
On Turn 2 (user sends approval), `tryResuming` finds the planner
PENDING and resumes it normally. The planner calls Gemini, which
re-delegates to the sub-agent. The sub-agent sees the approval in
`start.Messages` and completes. **This costs 1 extra Gemini call on
the resume turn.**
Direct routing (skipping Gemini on resume) is deferred to a follow-up.
Options explored include controller-side routing, ToolCallContent
pairing, and `pending_agent_id` on execution events — see branches
`agent-executor-confirmation`, `agent-executor-toolcall`, and
`agent-executor-two-layer` for prototypes.
## Before/After
| Scenario | Before | After |
|---|---|---|
| Plain delegation | 3 Gemini calls | 2 |
| Delegation + confirmation | ∞ (hangs) | Works (1 extra Gemini call on
resume) |
| Bash tool | No change | No change |
## Testing
6 new tests in `agent_executor_test.go`. E2e tested with `ax serve` +
`ax exec` across no-confirmation, confirmation (3x stress), and bash
tool scenarios.
## Sample logs (real `ax serve` + `ax exec`, gemini-3-flash-preview)
### Scenario 1: No confirmation
Commands:
```bash
go run ./examples/remote_agent # Terminal 1: lowercase agent on :50051
ax serve --config ax.yaml # Terminal 2
ax exec --server localhost:8494 --input "Please lowercase HELLO WORLD" # Terminal 3
```
Output: `please lowercase hello world`
<details>
<summary>Raw conversation log (2 events)</summary>
```json
{
"conversationId": "raw-no-confirm",
"seq": 1,
"execId": "6abb8d1c-2d0e-4aa2-9550-6c18c51b91a4",
"messages": [
{
"role": "user",
"content": {
"text": {
"text": "Please lowercase HELLO WORLD"
}
}
}
],
"state": "STATE_PENDING"
}
{
"conversationId": "raw-no-confirm",
"seq": 2,
"execId": "6abb8d1c-2d0e-4aa2-9550-6c18c51b91a4",
"messages": [
{
"role": "assistant",
"content": {
"text": {
"text": "please lowercase hello world"
}
}
}
],
"state": "STATE_PENDING"
}
```
</details>
<details>
<summary>Raw execution log — lowercase (3 events)</summary>
```json
{
"execId": "6abb8d1c-2d0e-4aa2-9550-6c18c51b91a4-lowercase",
"agentId": "lowercase",
"inputs": [
{
"role": "user",
"content": {
"text": {
"text": "Please lowercase HELLO WORLD"
}
}
}
],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:47.532433859Z"
}
{
"execId": "6abb8d1c-2d0e-4aa2-9550-6c18c51b91a4-lowercase",
"agentId": "lowercase",
"outputs": [
{
"role": "assistant",
"content": {
"text": {
"text": "please lowercase hello world"
}
}
}
],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:47.540714804Z"
}
{
"execId": "6abb8d1c-2d0e-4aa2-9550-6c18c51b91a4-lowercase",
"agentId": "lowercase",
"state": "STATE_COMPLETED",
"timestamp": "2026-05-10T23:20:47.541394315Z"
}
```
</details>
### Scenario 2: With confirmation
Commands (uses a confirm-enabled remote agent variant):
```bash
go run ./examples/remote_agent # modified to ask for confirmation
ax serve --config ax.yaml
ax exec --server localhost:8494 --input "Please lowercase HELLO WORLD" # Turn 1
# User approves the confirmation prompt # Turn 2
```
Turn 1: agent asks confirmation, planner stops with `STATE_PENDING`.
Turn 2: user approves, planner re-delegates to sub-agent (1 extra Gemini
call), sub-agent completes.
Agent log:
```
[agent] asking confirmation for: "Please lowercase HELLO WORLD"
[agent] user approved, lowercasing: "Please lowercase HELLO WORLD"
```
<details>
<summary>Raw conversation log (7 events)</summary>
```json
{
"conversationId": "raw-with-confirm",
"seq": 1,
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"messages": [{"role": "user", "content": {"text": {"text": "Please lowercase HELLO WORLD"}}}],
"state": "STATE_PENDING"
}
{
"conversationId": "raw-with-confirm",
"seq": 2,
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"messages": [{"role": "assistant", "content": {"confirmation": {"id": "dc65756b-d450-49e7-bf08-93b4b492363e", "question": "Can I lowercase \"Please lowercase HELLO WORLD\" for you?"}}}],
"state": "STATE_PENDING"
}
{
"conversationId": "raw-with-confirm",
"seq": 3,
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"state": "STATE_PENDING"
}
{
"conversationId": "raw-with-confirm",
"seq": 4,
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"messages": [{"role": "user", "content": {"confirmation": {"id": "dc65756b-d450-49e7-bf08-93b4b492363e", "approval": {"approved": true}}}}],
"state": "STATE_PENDING"
}
{
"conversationId": "raw-with-confirm",
"seq": 5,
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"messages": [{"role": "assistant", "content": {"text": {"text": "please lowercase hello world"}}}],
"state": "STATE_PENDING"
}
{
"conversationId": "raw-with-confirm",
"seq": 6,
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"messages": [{"role": "model", "content": {"text": {"text": "please lowercase hello world"}}}],
"state": "STATE_PENDING"
}
{
"conversationId": "raw-with-confirm",
"seq": 7,
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"state": "STATE_COMPLETED"
}
```
</details>
<details>
<summary>Raw execution log — __planner (5 events)</summary>
```json
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"agentId": "__planner",
"inputs": [{"role": "user", "content": {"text": {"text": "Please lowercase HELLO WORLD"}}}],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:54.629227521Z"
}
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"agentId": "__planner",
"outputs": [{"role": "assistant", "content": {"confirmation": {"id": "dc65756b-d450-49e7-bf08-93b4b492363e", "question": "Can I lowercase \"Please lowercase HELLO WORLD\" for you?"}}}],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:55.966125441Z"
}
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"agentId": "__planner",
"inputs": [
{"role": "user", "content": {"text": {"text": "Please lowercase HELLO WORLD"}}},
{"role": "assistant", "content": {"confirmation": {"id": "dc65756b-d450-49e7-bf08-93b4b492363e", "question": "Can I lowercase \"Please lowercase HELLO WORLD\" for you?"}}},
{"role": "user", "content": {"confirmation": {"id": "dc65756b-d450-49e7-bf08-93b4b492363e", "approval": {"approved": true}}}}
],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:55.994507633Z"
}
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"agentId": "__planner",
"outputs": [
{"role": "assistant", "content": {"text": {"text": "please lowercase hello world"}}},
{"role": "model", "content": {"text": {"text": "please lowercase hello world"}}}
],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:57.672836311Z"
}
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d",
"agentId": "__planner",
"state": "STATE_COMPLETED",
"timestamp": "2026-05-10T23:20:57.673539482Z"
}
```
</details>
<details>
<summary>Raw execution log — lowercase (5 events)</summary>
```json
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d-lowercase",
"agentId": "lowercase",
"inputs": [{"role": "user", "content": {"text": {"text": "Please lowercase HELLO WORLD"}}}],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:55.956365514Z"
}
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d-lowercase",
"agentId": "lowercase",
"outputs": [{"role": "assistant", "content": {"confirmation": {"id": "dc65756b-d450-49e7-bf08-93b4b492363e", "question": "Can I lowercase \"Please lowercase HELLO WORLD\" for you?"}}}],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:55.965184294Z"
}
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d-lowercase",
"agentId": "lowercase",
"inputs": [
{"role": "user", "content": {"text": {"text": "Please lowercase HELLO WORLD"}}},
{"role": "assistant", "content": {"confirmation": {"id": "dc65756b-d450-49e7-bf08-93b4b492363e", "question": "Can I lowercase \"Please lowercase HELLO WORLD\" for you?"}}},
{"role": "user", "content": {"text": {"text": "Please lowercase HELLO WORLD"}}},
{"role": "assistant", "content": {"confirmation": {"id": "dc65756b-d450-49e7-bf08-93b4b492363e", "question": "Can I lowercase \"Please lowercase HELLO WORLD\" for you?"}}},
{"role": "user", "content": {"confirmation": {"id": "dc65756b-d450-49e7-bf08-93b4b492363e", "approval": {"approved": true}}}}
],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:56.876585877Z"
}
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d-lowercase",
"agentId": "lowercase",
"outputs": [{"role": "assistant", "content": {"text": {"text": "please lowercase hello world"}}}],
"state": "STATE_PENDING",
"timestamp": "2026-05-10T23:20:56.886341400Z"
}
{
"execId": "d2271679-8812-4d13-967c-2401b94a761d-lowercase",
"agentId": "lowercase",
"state": "STATE_COMPLETED",
"timestamp": "2026-05-10T23:20:56.887095309Z"
}
```
</details>
rakyll
pushed a commit
that referenced
this pull request
Jul 21, 2026
google-antigravity 0.1.7's _connect_websocket retries both "localhost" and "127.0.0.1" when dialing localharness (local_connection.py:993), so the sidecar no longer needs to patch /etc/hosts for substrate/gVisor environments where localhost may not resolve. Ref #325 antigravity: guard conversation_id for safe save_dir use (#312) harness_server uses the caller-supplied conversation_id directly as a per-conversation storage directory name (save_dir = state_dir / conversation_id) before the Antigravity harness ever sees it, so a value like "../escape" could traverse outside state_dir. Reject an empty id or one containing a path separator or a "." / ".." component at the boundary in _run_turn, returning an INVALID_ARGUMENT end frame. This is intentionally only a path-safety guard, not the id-format contract. The Antigravity harness owns that: it validates the forwarded cascade_id (length and character set) per cl/948016352. Keeping the format rule in one place avoids the two layers drifting. chore: remove stale docs, gitignore, and Makefile entries (#329) - CONTRIBUTING.md: drop 'make run-remote' (no such Makefile target) - .gitignore: remove orphaned entries (local/remote_agent, sandbox-router, tasklog, axepp, google-cloud-sdk, test-sandbox-config.yaml) - Makefile: add missing deps, clean-logs to .PHONY antigravity: remove vertex env-var override (superseded by AGY 0.1.7) (#331) * antigravity: remove vertex env-var override (superseded by AGY 0.1.7) google-antigravity 0.1.7 reads GOOGLE_GENAI_USE_VERTEXAI / GOOGLE_GENAI_USE_ENTERPRISE + GOOGLE_CLOUD_{PROJECT,LOCATION} natively, so the sidecar no longer needs _vertex_kwargs_from_env to thread them into LocalAgentConfig. - Delete _vertex_kwargs_from_env + VertexKwargs. - _build_default_config: bare LocalAgentConfig(system_instructions=...). - _has_credentials: read project/location from env (SDK hydrates them onto VertexEndpoint, not LocalAgentConfig). - Update tests accordingly. Requires the 0.1.7 bump. Credential resolution is startup-only and does not touch the save_dir/state_dir resume path. Ref #325 * antigravity: drop stale _has_credentials docstring note about vertex env override Interactions: persist interactions harness with durable dir and resume from it (#327) interactions: make WorkDir authoritative for run_command (#335) The interactions harness executed run_command relative to the process's ambient working directory, which is not reliably the agent's workspace: with no Cwd argument the command ran wherever the process cwd happened to be, and a model-supplied Cwd (including "/") was honored blindly. As a result tool calls often ran from "/" instead of the configured workspace (#332). Make the workspace directory authoritative in the executor rather than relying on ambient process state: - Add AntigravityInteractionsConfig.WorkDir (defaults from AX_HARNESS_WORKDIR). - run_command now resolves its directory via resolveRunDir: no Cwd -> WorkDir; relative Cwd -> joined under WorkDir; absolute Cwd -> honored as-is. Empty WorkDir preserves the previous process-cwd behavior. - Add WorkspaceSystemInstruction to orient the agent about its working directory so it emits workspace-relative paths, and wire it in cmd/ax. Fixes #332 Remove the binary (#334) * Remove the binary * chore: add e2e directory to .gitignore * refactor: migrate binary file check to a dedicated workflow file * fix: restrict binary file check to added, copied, or modified files
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
2 participants
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.
Summary
Housekeeping for references to code/targets that no longer exist.
make run-remotesection — no such Makefile target.local_agent,remote_agent,sandbox-router,examples/sandbox_agent/cloudbuild.yaml,tasklog/*,/axepp,google-cloud-sdk/,google-cloud-cli-darwin-arm.tar.gz,test-sandbox-config.yaml).depsandclean-logsto.PHONY.No functional/build changes;
go build ./...passes.