Skip to content

Modes: default to plan, retire the pulse-designer agent shell (#389) - #395

Merged
aarontrowbridge merged 2 commits into
mainfrom
389-modes-default-to-plan-retire-the-pulse-designer-agent-shell
Aug 16, 2026
Merged

Modes: default to plan, retire the pulse-designer agent shell (#389)#395
aarontrowbridge merged 2 commits into
mainfrom
389-modes-default-to-plan-retire-the-pulse-designer-agent-shell

Conversation

@aarontrowbridge

Copy link
Copy Markdown
Member

Closes #389

Two commits:

  1. Plan-first default (cherry-picked from Aaron's 2026-08-08 feat/plan-first-default work, refs feat: plan-first default agent for all Amicode sessions #300, never merged): buildOpencodeConfigContent injects default_agent: "plan" — every new session opens read-only; execution starts on an explicit switch to build. Verified against the real vendored binary via the injection+merge test.

  2. Retire the pulse-designer agent shell: the picker is plan/build only. The agent entry was a four-line prompt deferring to the compiled AGENTS.md interview section; its permissions were always the config-root block, so it did nothing build didn't. The interview content stays, agent-agnostic, in the injected instructions — the roles-not-modes posture from Autoresearch Studio — the loop is the product (pulse-designer → domain pack, telaio spine, campaign autonomy) #368. e2e turns send agent: "build"; GET /agent now asserts the retirement.

Verification: 920/920 fast suite, typecheck clean, real-binary config injection+merge green. The live-creds e2e (interview tier C) times out on a 300s model POST — reproduced identically on clean main in a throwaway worktree, so it's an environmental live-endpoint hang (overnight), not a regression; flagged separately.

Note: auto-merge is NOT armed — #389 is hitl.

Every new session opens on opencode's read-only plan agent. Execution
(pulse-designer interview, solves) starts only when the user switches
agents in the composer.

buildOpencodeConfigContent injects `default_agent: "plan"` into the
OPENCODE_CONFIG_CONTENT every Amicode server boots with. The blob
deep-merges over the user's global config, so this is the product
default for all users. Per-message `agent` selections (e2e tests,
distiller's --agent) are unaffected.

Verification: 867 passed / 3 skipped / 0 failed full fast suite;
vendored opencode 1.17.3 accepts the field via the real-binary
injection+merge test.

Refs: #300
…ly (#389)

The pulse-designer agent was a shell: a four-line prompt deferring to the
compiled AGENTS.md interview section, with permissions that were always
the config-root block. The interview content is agent-agnostic (visible
to every agent via the injected instructions), so the entry is removed
and the picker becomes opencode's native plan/build — the roles-not-modes
posture (#368). e2e per-message agent selections move to build; the
GET /agent assertion now pins the retirement.
@aarontrowbridge aarontrowbridge linked an issue Aug 16, 2026 that may be closed by this pull request
4 tasks
@coderabbitai

coderabbitai Bot commented Aug 16, 2026

Copy link
Copy Markdown

Warning

Review limit reached

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

Next review available in: 17 minutes

Limit details: You’ve used all 3 included reviews currently available under your plan.

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: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 56d8f77a-cb5e-4fee-8ea6-375a2877f97f

📥 Commits

Reviewing files that changed from the base of the PR and between 84a0784 and da19513.

📒 Files selected for processing (6)
  • AGENTS.md
  • packages/extension/src/opencode_config.ts
  • packages/extension/src/scores/compiler.ts
  • packages/extension/test/opencode_config.test.ts
  • packages/extension/test/slow/interview_e2e.test.ts
  • packages/extension/test/slow/scores_e2e.test.ts

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

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Modes: default to plan, retire the pulse-designer agent shell

1 participant