Repository navigation
Conversation
GitHub cancels a pending run when a newer run joins its concurrency group, even with cancel-in-progress off. Master pushes shared one group through github.ref, so a quick series of merges cancelled most of their checks. Pull requests still group by number and cancel their running check. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
main.rs, transcript.rs and src/tui/ read nightly fields, so protocol-fidelity now covers them. idle-cost now covers main.rs and transcript.rs and names the per-event work a CLI stream can repeat. The instructions list the modules and tests that exist now. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
research/next-steps.md was listed as the agreed plan, but it is the earlier proposal and its CLI syntax is stale. The context now names README.md as the command contract, docs/parity.md as the plan pinned to the nightly, with its statuses as that baseline, docs/parity-inventory.md for the desktop's source references and docs/sprint.md for the process and the merged PRs. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
wait exits 7 with the outcome needs-attention when T3 streams a pending approval, and 1 with the outcome failed when the run fails. send --wait exits 6 with THREAD_WAIT_TIMEOUT when the turn outlasts --timeout, and the error names the message it sent. Each case checks the plain output and that --json prints one object. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The record lists each finding from Astra and Claude Fable 5.1 with Sol's outcome: what #23 fixes, what #20 had fixed, the ruleset change Sol made, the CLI control-character gap left for its own PR, and the evidence for the parts that needed no change. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
#22 merged at cb51f8c while this branch was open, and both added to the README's tests paragraph and the sprint table. The merge keeps #22's text and row, adds this branch's sentences after it and fills in row 15's merge commit, so the record counts 15 merged sprint PRs. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
protocol-fidelity now includes src/models.rs, which reads the providers, models and option descriptors in the server config. The rule and the instructions no longer say every stream resumes with afterSequence. The shell and thread subscriptions do, and subscribeServerConfig resubscribes with an empty payload and starts from a fresh snapshot, as client.rs does since #22. The audit table in docs/sprint.md records both changes. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
krishhgg
marked this pull request as ready for review
October 10, 2026 05:37
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.
Each master commit gets its own CI run, Greptile checks more of the wire-reading and CLI code, and three integration tests cover attention, failed-run and timeout outcomes. This completes the warranted CI/config changes from Astra and Fable's first-ten audit.
Sol passed formatting, strict Clippy and all 296 tests at
d21868c: 274 library, 2 main, 12 CLI and 8 fake-RPC tests. The finalb3d3796changes only Greptile prose/scope and the audit record; its source, tests and workflow are identical. The diff check and Greptile JSON/context validation pass. Final CI passed onb3d3796. Greptile reviewed that same commit, gave it 5/5 and found no actionable issue.Decision log
What I noticed
Astra and Claude Fable 5.1, both at xhigh, independently audited sprint PRs #8 through #17. They found cancelled master CI runs, gaps in Greptile's protocol/cost coverage and missing CLI wait outcomes. The context file still called an earlier proposal the agreed plan.
How I confirmed it
Sol checked all ten cancelled master runs from #8 through #21; none had a job. The CLI tests run the actual binary with temporary homes and fake HTTP, WebSocket and auth endpoints. Plain and JSON cases verify exits 7, 1 and 6, partial replies, exactly one JSON result, and the timeout's sent message ID. Sol also verified master's ruleset now requires macOS and Greptile Review.
Options
Group PR runs by number and master runs by commit SHA. Keeping one master group still replaces pending runs even when cancellation is off. Expand the existing protocol and idle-cost rules, and use README for current commands, the parity documents for the pinned baseline and source references, and sprint.md for progress. Reuse the existing fake-server harness. A queue redesign or new testing framework was unnecessary.
What review changed
Sol reconciled both audits. #20 had already fixed repeated handoff descriptions, so this PR leaves that code alone. Sol added
models.rsto protocol coverage and corrected reconnect guidance: shell/thread streams resume withafterSequence; config subscriptions use an empty request and a fresh snapshot. Claude kept #22's changes when bringing master into this branch. The audit record tracks 15 merged sprint PRs and the next audit at 20. Greptile's final review found no actionable issue.Known limits
This changes no product code or TUI layout. The raw-control-character gap in plain CLI output needs a separate maintenance PR. Attention coverage exercises a command approval, not a pending question. The timeout test takes about two seconds across its two runs. Failed-run JSON retains the existing
ok: trueandoutcome: "failed"contract. Maintenance PRs do not count toward the next audit.
Confidence Score: 5/5
This PR appears safe to merge; no actionable issue was found.Summary
This PR updates CI grouping, review guidance, and CLI wait tests without changing product code.
Acknowledgments supplied by
krishhgg:ok: truewithoutcome: "failed".Reviews (1) · Last reviewed commit: "greptile: cover models.rs and state each..." · Reviewed by Greptile