Repository navigation
Conversation
github-actions
Bot
added
the
documentation
11 tasks
This branch has not been deployed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
Goal mode is absent from current main after #880, while the requested experience is an explicit objective that persists until completion or a bounded stop. The historical implementation and #192 do not establish how the current OpenCode, Kimi Code, Claude Code, Codex and DeepSeek Harness implementations compare.
Value
Provide a source-pinned basis for choosing OpenPI's Goal design, with clear separation between observed implementation, recommendations and unexecuted evaluation protocols.
Approach
Add a research record and index entry. Compare the five products, distinguish OpenCode core from third-party Ralph and current Kimi Code from the archived Python CLI, and identify Pi 1.2.0's native settlement hooks. Recommend a single-Session Goal with explicit user activation, one model status tool, native continuation and bounded lifecycle authority. Include static design ablation, source receipts and a proposed evaluation protocol. This is a draft recommendation, not an adopted Decision.
Related: #192, #43, #879.
Validation
Using Bun 1.3.14, Node 24.20.0 and frozen-lockfile dependencies:
bun run check: passed.VITEST_MAX_WORKERS=1 bun run test: failed in the Node phase; 2,415 passed, eight skipped, six failed out of 2,429. The six failures are in the unchanged file-mutation renderer, native child renderer and custom footer tests. The same six failures reproduce across the same three files in a pristine archive of main31b2d045with the same dependencies (five passed, six failed). Their underlying cause is not diagnosed by this docs PR.VITEST_MAX_WORKERS=1 bun run test:ui: independently passed all 1,640 tests across 107 files, since the failed Node phase prevented the full runner from reaching UI.git diff --checkpassed.The first attempt under the host's default Node 20 could not start strip-types tests; the supported Node 24 runs above are the reported evidence. No test expectations were changed or skipped by this PR.
No competitive runtime, live model comparison, benchmark or installed-runtime smoke was performed.
Impact
Documentation only. User-visible behavior, model-facing tools/context, runtime/lifecycle and persisted config/data are unchanged. The research records the Pi compatibility boundary and unresolved evaluation questions; it does not restore the deleted Goal implementation.