fix(agent-control): tag control messages with dedicated role and resume paused team runs by w0wl0lxd · Pull Request #129 · w0wl0lxd/n00n

w0wl0lxd · 2026-07-27T02:18:54Z

Summary

Add control flag to Message, AgentInput, and QueuedMessage so agent-to-agent control messages are visually and structurally distinct from user prompts.
Introduce DisplayRole::Control with theme styling so control messages render separately in the UI.
Wire n00n.session.prompt control option; agent_control message and resume now send steer=true, control=true.
Extend SessionRequest::Status with last_user so agent_control resume can read the paused team run_id from the target session history and construct a continuation prompt.
Include run_id in team run_waves pause payload for consistency.

Test plan

cargo fmt --all
cargo check --all --tests
cargo clippy --all --tests -- -D warnings
cargo nextest run --workspace (3873 passed, 1 skipped)
cargo deny check shows pre-existing license/source/advisory warnings unrelated to this change.

…queued prompts

coderabbitai · 2026-07-27T02:19:01Z

Warning

Review limit reached

@w0wl0lxd, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 18 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 3fdf42e8-8807-44a2-ad7b-050560a164a2

📥 Commits

Reviewing files that changed from the base of the PR and between 7a715ea and f0d43d3.

📒 Files selected for processing (17)

n00n-agent/src/agent/run.rs
n00n-lua/src/api/session.rs
n00n-lua/src/api/util/command.rs
n00n-lua/tests/plugin_host.rs
n00n-providers/src/providers/devin.rs
n00n-storage/src/sessions.rs
n00n-ui/src/agent/mod.rs
n00n-ui/src/agent/shared_queue.rs
n00n-ui/src/app/mod.rs
n00n-ui/src/app/queue.rs
n00n-ui/src/app/session.rs
n00n-ui/src/app/tests.rs
n00n-ui/src/components/input.rs
n00n-ui/src/event_loop.rs
plugins/agent_control/init.lua
plugins/team/init.lua
site/docs/content/lua-api/_index.md

📝 Walkthrough

Walkthrough

The PR adds control-message metadata across agent inputs, queues, storage, events, and UI rendering. Lua session prompts can steer control messages, while paused team status exposes resume metadata and team execution preserves mode and run state.

Changes

Agent control message flow

Layer / File(s)	Summary
Control metadata and compatibility `n00n-agent/...`, `n00n-providers/...`, `n00n-storage/...`	Adds backward-compatible `control` fields to agent inputs, messages, stored queue entries, and queue-consumed events.
Queue delivery and display `n00n-ui/src/agent/`, `n00n-ui/src/app/`, `n00n-ui/src/chat.rs`, `n00n-ui/src/components/*`	Routes steering messages with control metadata and renders them with dedicated control roles, surfaces, styles, and theme support.

Paused team resume

Layer / File(s)	Summary
Session prompt wiring `n00n-ui/src/event_loop.rs`, `n00n-lua/src/api/session.rs`, `site/docs/content/lua-api/_index.md`	Adds `steer` and `control` prompt options and exposes validated `paused_team` status metadata.
Resume orchestration `plugins/agent_control/init.lua`, `plugins/team/init.lua`, `n00n-lua/tests/plugin_host.rs`	Builds structured resume prompts and preserves paused team mode, run IDs, checkpoints, and resume-step guidance.

Estimated code review effort: 4 (Complex) | ~45 minutes

Possibly related PRs

w0wl0lxd/n00n#21: Both modify agent-control message handling and session prompt steering.
w0wl0lxd/n00n#34: Both modify queue-consumed event payload propagation through the UI.

Sequence Diagram(s)

sequenceDiagram
  participant AgentControl
  participant SessionEventLoop
  participant MessageQueue
  participant TeamPlugin
  AgentControl->>SessionEventLoop: request status
  SessionEventLoop->>AgentControl: return paused_team metadata
  AgentControl->>SessionEventLoop: submit resume prompt with steer=true and control=true
  SessionEventLoop->>MessageQueue: queue control steering message
  MessageQueue->>TeamPlugin: resume paused run with preserved mode and run_id

Poem

A bunny sends a control cue,
Through queues where gentle signals flew.
A paused team wakes with mode in hand,
Resuming steps as they were planned.
“Hop onward!” twitches every ear.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: control-message tagging and paused team-run resume support.
Description check	✅ Passed	The description is clearly aligned with the changeset and covers the new control flag, UI role, prompt wiring, and resume flow.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch fix/agent-control-steer

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

github-actions · 2026-07-27T02:26:12Z

Dependency Review

✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.

Scanned Files

None

codecov · 2026-07-27T02:39:43Z

Codecov Report

❌ Patch coverage is 77.94486% with 88 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
n00n-ui/src/event_loop.rs	75.72%	25 Missing ⚠️
n00n-ui/src/chat.rs	20.00%	20 Missing ⚠️
n00n-ui/src/components/messages/render.rs	0.00%	14 Missing ⚠️
n00n-ui/src/components/tool_display.rs	0.00%	9 Missing ⚠️
n00n-ui/src/app/queue.rs	86.95%	6 Missing ⚠️
n00n-ui/src/components/messages/mod.rs	0.00%	5 Missing ⚠️
n00n-ui/src/app/session.rs	93.54%	2 Missing ⚠️
n00n-acp/src/server.rs	50.00%	1 Missing ⚠️
n00n-agent/src/headless.rs	0.00%	1 Missing ⚠️
n00n-providers/src/providers/devin.rs	0.00%	1 Missing ⚠️
... and 4 more

📢 Thoughts on this report? Let us know!

github-actions

Criterion

Details

Benchmark suite	Current: `f0d43d3`	Previous: `9f141ea`	Ratio
`fib/jit_mlua_hook`	`6768349` ns/iter (`± 269460`)	`6812960` ns/iter (`± 90866`)	`0.99`
`fib/jit_watchdog`	`2221422` ns/iter (`± 6031`)	`2221884` ns/iter (`± 14410`)	`1.00`
`fib/jit_none`	`2220932` ns/iter (`± 7487`)	`2221511` ns/iter (`± 61848`)	`1.00`
`fib/interp_mlua_hook`	`7978906` ns/iter (`± 92610`)	`7974083` ns/iter (`± 51292`)	`1.00`
`fib/interp_watchdog`	`4337138` ns/iter (`± 32070`)	`4320307` ns/iter (`± 19095`)	`1.00`
`fib/interp_none`	`4282319` ns/iter (`± 14059`)	`4271546` ns/iter (`± 18837`)	`1.00`
`buffer_rw/jit_mlua_hook`	`583936` ns/iter (`± 1350`)	`588077` ns/iter (`± 2625`)	`0.99`
`buffer_rw/jit_watchdog`	`191666` ns/iter (`± 315`)	`192371` ns/iter (`± 550`)	`1.00`
`buffer_rw/jit_none`	`191574` ns/iter (`± 329`)	`192286` ns/iter (`± 391`)	`1.00`
`buffer_rw/interp_mlua_hook`	`1039697` ns/iter (`± 9418`)	`1619333` ns/iter (`± 22599`)	`0.64`
`buffer_rw/interp_watchdog`	`589026` ns/iter (`± 1461`)	`1050720` ns/iter (`± 5338`)	`0.56`
`buffer_rw/interp_none`	`588118` ns/iter (`± 1248`)	`589332` ns/iter (`± 2822`)	`1.00`
`splash_render_120x40`	`51928` ns/iter (`± 1645`)	`52724` ns/iter (`± 10231`)	`0.98`
`splash_render_200x60`	`195960` ns/iter (`± 1128`)	`194530` ns/iter (`± 7674`)	`1.01`

This comment was automatically generated by workflow using github-action-benchmark.

…am runs - Add `control` flag to `Message`, `AgentInput`, and `QueuedMessage` so agent-control messages are tagged independently of user prompts. - Propagate the flag through `AgentEvent::QueueItemConsumed` and the UI display pipeline, adding `DisplayRole::Control` with its own theme style so control messages render distinctly from user messages. - Wire `n00n.session.prompt` `control` option and update `agent_control` `message`/`resume` to use both `steer` and `control`. - Surface `last_user` from `SessionRequest::Status` so `agent_control` resume can retrieve the paused `team` `run_id` from the target session history and build a continuation prompt. - Include `run_id` in `team` `run_waves` pause payload for consistency. Tests: cargo nextest run --workspace (3873 passed), cargo clippy --all --tests -- -D warnings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c59a5ad7d2

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Wire PluginHost ui_action_tx into an in-process daemon listener so CLI agent list unions TUI sessions while the UI is up. Document stacked follow-ups for worker absorb (#134) and steer/control (#129).

coderabbitai

Actionable comments posted: 5

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (3)

n00n-ui/src/app/queue.rs (1)
109-130: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Preserve the control tag when editing queued messages.

The caller converts this message to Submission, dropping control; re-queueing then rebuilds AgentInput with control: false. Keep the original flag in MessageQueue::editing and apply it in replace_editing.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@n00n-ui/src/app/queue.rs` around lines 109 - 130, Preserve the original
control flag across queued-message editing: extend MessageQueue::editing to
store input.control, capture it in take_focused_for_edit, and have
replace_editing reuse that stored value when rebuilding AgentInput instead of
forcing control to false.
n00n-ui/src/app/session.rs (1)
70-121: 🎯 Functional Correctness | 🟠 Major | 🏗️ Heavy lift

Persist queued delivery mode, not only control.

A control prompt queued while streaming uses Delivery::Steering, but restoration calls queue_restored_submission, which hard-codes Delivery::TurnEnd. After restart, a resume/control message waits for turn end instead of interrupting safely. Persist and restore Delivery alongside the AgentInput; extend the restart test to assert Delivery::Steering.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@n00n-ui/src/app/session.rs` around lines 70 - 121, Persist the queued
message’s Delivery value alongside the existing control and AgentInput fields in
stored_message, then restore it in restored_submission so resumed control
prompts retain Delivery::Steering instead of defaulting to Delivery::TurnEnd.
Update the restart test covering queued delivery to assert the restored message
uses Delivery::Steering.
plugins/team/init.lua (1)
749-757: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick win

Waves-mode pause payload is missing paused = true, breaking resume detection for waves runs.

Unlike run_autonomous's and run_single_pass's pause objects (which both include paused = true), the pause table built here from wave_result never sets paused. Since n00n-ui/src/event_loop.rs::paused_team_run requires payload["paused"] == true before surfacing status.paused_team, this means agent_control resume can never detect a paused waves-mode team run — it will always report "no paused team run found", silently defeating this PR's resume feature for that mode.
🐛 Proposed fix
           if wave_result.paused then
             pause = {
+              paused = true,
               run_id = run_id,
               mode = input.mode,
               failed_step = failed_step_index,
               failed_role = failed_role,
               error = wave_result.error,
             }
           end
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@plugins/team/init.lua` around lines 749 - 757, Update the pause payload
constructed when wave_result.paused in the waves-mode flow to include paused =
true, matching the pause objects created by run_autonomous and run_single_pass.
Preserve the existing run_id, mode, failure, and error fields so paused_team_run
can detect and resume waves runs.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@n00n-agent/src/agent/run.rs`:
- Around line 744-762: Update the initial message creation in Agent::run to
preserve input.control, using the appropriate control-message constructor or
classification instead of always calling Message::user_with_images. Keep the
existing user-message behavior when input.control is false, and add a direct-run
test verifying a control input produces a control-tagged first history message.

In `@n00n-lua/tests/plugin_host.rs`:
- Around line 4780-4828: Strengthen
agent_control_resume_preserves_paused_team_mode by destructuring the Prompt
request’s submission flags and asserting that resume is submitted with steer and
control enabled. Keep the existing prompt text assertion and response flow
unchanged.

In `@n00n-ui/src/event_loop.rs`:
- Around line 128-172: Update paused_team_run so malformed team tool-result JSON
is treated as no paused run instead of returned as an error: log the parse
failure with structured context and continue scanning or return Ok(None). Remove
the ? propagation from the SessionRequest::Status call that computes the
auxiliary paused-team field, while preserving the rest of the status response
fields and behavior.
- Around line 147-151: Replace the inline "team" comparison in the
is_team_result history scan with a named tool-name constant. Define the constant
immediately after imports, following the existing TASK_TOOL_NAME,
WRITE_TOOL_NAME, and WIRE_TOOL_NAME convention, then use it in the tool_uses
predicate.

In `@site/docs/content/lua-api/_index.md`:
- Around line 2680-2687: Update the return-value documentation for
n00n.session.status() to describe the nested paused_team JSON payload, including
its available fields such as run_id, rather than only marking paused_team as
optional. Keep the existing top-level return shape and condition unchanged.

---

Outside diff comments:
In `@n00n-ui/src/app/queue.rs`:
- Around line 109-130: Preserve the original control flag across queued-message
editing: extend MessageQueue::editing to store input.control, capture it in
take_focused_for_edit, and have replace_editing reuse that stored value when
rebuilding AgentInput instead of forcing control to false.

In `@n00n-ui/src/app/session.rs`:
- Around line 70-121: Persist the queued message’s Delivery value alongside the
existing control and AgentInput fields in stored_message, then restore it in
restored_submission so resumed control prompts retain Delivery::Steering instead
of defaulting to Delivery::TurnEnd. Update the restart test covering queued
delivery to assert the restored message uses Delivery::Steering.

In `@plugins/team/init.lua`:
- Around line 749-757: Update the pause payload constructed when
wave_result.paused in the waves-mode flow to include paused = true, matching the
pause objects created by run_autonomous and run_single_pass. Preserve the
existing run_id, mode, failure, and error fields so paused_team_run can detect
and resume waves runs.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: b585cd42-130c-4fcf-a66a-5ba7c9ea0530

📥 Commits

Reviewing files that changed from the base of the PR and between 6351b86 and 7a715ea.

📒 Files selected for processing (36)

changelog.d/agent-control-resume.fixed.md
n00n-acp/src/server.rs
n00n-acp/src/translate.rs
n00n-agent/src/agent/history.rs
n00n-agent/src/agent/run.rs
n00n-agent/src/headless.rs
n00n-agent/src/lib.rs
n00n-agent/src/mcp/mod.rs
n00n-agent/src/types.rs
n00n-lua/src/api/agent.rs
n00n-lua/src/api/session.rs
n00n-lua/src/api/util/command.rs
n00n-lua/tests/plugin_host.rs
n00n-providers/src/providers/anthropic/shared.rs
n00n-providers/src/providers/devin.rs
n00n-providers/src/types.rs
n00n-storage/src/sessions.rs
n00n-ui/src/agent/agent_loop.rs
n00n-ui/src/agent/shared_queue.rs
n00n-ui/src/app/mod.rs
n00n-ui/src/app/mode.rs
n00n-ui/src/app/queue.rs
n00n-ui/src/app/session.rs
n00n-ui/src/app/tests.rs
n00n-ui/src/chat.rs
n00n-ui/src/components/messages/mod.rs
n00n-ui/src/components/messages/render.rs
n00n-ui/src/components/messages/segment.rs
n00n-ui/src/components/mod.rs
n00n-ui/src/components/tool_display.rs
n00n-ui/src/event_loop.rs
n00n-ui/src/theme.rs
plugins/agent_control/init.lua
plugins/team/init.lua
site/docs/content/lua-api/_index.md
src/sdk_mode.rs

📜 Review details

⏰ Context from checks skipped due to timeout. (3)

GitHub Check: Analyze (python)
GitHub Check: Criterion
GitHub Check: Analyze (rust)

🧰 Additional context used

📓 Path-based instructions (6)

**/*