tracker-sync: reserve Linear Done for merged PRs; In Review after make-pr — fn-66#178
Merged
Merged
Conversation
…d In Review rung - status-sync.md flow→normalized map is now flowToNormalized(spec, prEvidence): terminal done/verified require a MERGED PR probe; local completion necessary, not sufficient (R1). Open PR → in-review rung (R2); closed-unmerged → non-terminal + NEEDS_HUMAN (R6). who-wins / deadlock-first ordering untouched. - linear-ladder.md + github.md + adapter-interface.md updated in lockstep: transport-blind terminal-write-requires-merge-evidence invariant (R8). - Worked fixtures S-G..S-J with EXACT expected states: no-PR all-done → stays In Progress (no advance); open-PR → In Review; merged-PR → Done; closed-unmerged → non-terminal + NEEDS_HUMAN (R7). - Regenerated Codex mirror. Task: fn-66-tracker-sync-reserve-linear-done-for.1 Claude-Session: https://claude.ai/code/session_01A295VQ42CDSgVTTPgkBY4x
…s/probe-error; S-G preserve rule Addresses impl-review NEEDS_WORK (3 findings): - in-review rung now has an explicit reconcile-loop branch (was only in the flow→normalized table; S-H setStatus(in-review) previously fell through to conflictTiebreak). Push fires only on a real open-PR signal. - S-G no-PR preserve rule promoted from fixture-only to policy: prEvidence=none projects in-review but the loop KEEPS an existing valid non-terminal state (no forced in-progress→in-review advance, no terminal). - prEvidence enum extended with `ambiguous` + `probe-error`; both (and closed-unmerged) route to a single non-terminal + NEEDS_HUMAN branch. Unknown branch_name / gh failure is probe-error, never `none`/`merged`. Terminal reachable ONLY from unambiguous MERGED. S-J extended to cover them. - Regenerated Codex mirror. Task: fn-66-tracker-sync-reserve-linear-done-for.1 Claude-Session: https://claude.ai/code/session_01A295VQ42CDSgVTTPgkBY4x
…minal / makePr→In Review / land.merged→Done-on-merge (fn-66.2) - completionReview (work phases.md 3g + retro-fire + SKILL table): comment-shaped verdict/R-ID coverage, NEVER terminal Done; ceremony seeds completionReview=comment (R4) - makePr: In Review status push rides the unconditional bridge-active PR-link path, not gated behind perEvent.makePr (R2) - land.merged: SOLE Done driver, active-by-default when bridge active, self-checks the GitHub MERGED probe before terminal write (R3/R10) - steps.md push/reconcile setStatus: merge-evidence gate via flowToNormalized(spec, prEvidence); manual reconcile MAY terminal-write Done iff MERGED (per-write invariant) - flowctl.py land.merged default documented (active-by-default, MERGED-gated); schema default stays off (fn-52.1 invariant); +test_land_tracker_event.py round-trip - docs/tracker-sync.md: invert spec-completion-review-owns-Done claim; perEvent table + unconditional-paths prose updated - Codex mirror regenerated; bundled .flow/bin/flowctl.py synced (dual-copy invariant) - full suite (1110) + ci_test.sh (67) green Task: fn-66-tracker-sync-reserve-linear-done-for.2 Claude-Session: https://claude.ai/code/session_01A295VQ42CDSgVTTPgkBY4x
… fix stale Done-owner prose (fn-66.2) - land.merged: TRACKER_TERMINAL_OK now selects the dispatch op explicitly (push on clean MERGED, comment-only + NEEDS_HUMAN otherwise) — land no longer trusts the caller's merge claim (review #2, R3) - land workflow.md: fix collapsed list item (RELEASED.4. → separate line) (review #5) - docs/tracker-sync.md recovery: work.completionReview recovered via comment, not push (review #3, R4) - github.md: closing-ref bypass prose → land.merged Done projection, not spec-completion-review (review #4) - docs/teams.md + flowctl.md: invert completionReview-owns-Done claims; land.merged active-by-default; make-pr In Review unconditional - Codex mirror regenerated (land/workflow.md, github.md) Task: fn-66-tracker-sync-reserve-linear-done-for.2 Claude-Session: https://claude.ai/code/session_01A295VQ42CDSgVTTPgkBY4x
…eremony + dispatch grammar (fn-66.2) - steps.md:56 duplicate discovery ceremony seed: completionReview reconcile → comment; line-60 'one exception' → 'two exceptions' (make-pr link/In Review + land.merged) (review #1, R4/R10) - make-pr dispatch grammar: operation token back to canonical 'operation: push <spec-id>, event: makePr' (PR URL / In Review as evidence prose, not in the op token); same for the §5.7 retro-fire path (review #2, R2) - work/phases.md completionReview dispatch examples: drop 'verdict + R-ID coverage' from the op token → canonical 'operation: comment <spec-id>, event: work.completionReview' - Codex mirror regenerated Task: fn-66-tracker-sync-reserve-linear-done-for.2 Claude-Session: https://claude.ai/code/session_01A295VQ42CDSgVTTPgkBY4x
…ional paths (fn-66.2) - flowctl.py:1024 header comment no longer claims a stray enabled=true 'does nothing until a specific event is opted in' — calls out the two fn-66 unconditional bridge-active paths (make-pr link/In Review + land.merged Done-on-merge) (review #1, R10) - bundled .flow/bin/flowctl.py synced (dual-copy invariant) Task: fn-66-tracker-sync-reserve-linear-done-for.2 Claude-Session: https://claude.ai/code/session_01A295VQ42CDSgVTTPgkBY4x
…t NO_WORK (fn-66.3) - All-done + open PR: explicit greppable PILOT_VERDICT=DEFERRED_TO_LAND (stage=land), not a silent skip→NO_WORK collapse - All-done + no PR (FLOW-15 case): always classifies make-pr, never NO_WORK - Closed-unmerged / missing-branch / merged-but-open-spec stay NEEDS_HUMAN - Register DEFERRED_TO_LAND in SKILL.md verdict grammar + land stage; update classification table + crash-class note - Regenerate Codex mirror Task: fn-66-tracker-sync-reserve-linear-done-for.3
…r (fn-66.3) - ralph.md Codex /goal verdict grammar + Claude /goal recipe now name DEFERRED_TO_LAND and route it to land, not a pilot re-run Task: fn-66-tracker-sync-reserve-linear-done-for.3
…ion bump 2.1.2 - audit confirms fn-66.1/.2/.3 already inverted tracker-sync.md/teams.md "completion-review owns Done" → "land.merged owns Done" + In Review documented - decision record Consequences: In Review actively projected at make-pr; Done gated on GitHub MERGED probe (projection-not-coordination unchanged) - CHANGELOG 2.1.2 (Fixed: merge-evidence gate + In Review; Changed: pilot DEFERRED_TO_LAND) - version bump 2.1.1 → 2.1.2 across all manifests; Codex mirror regenerated Task: fn-66-tracker-sync-reserve-linear-done-for.4 Claude-Session: https://claude.ai/code/session_01A295VQ42CDSgVTTPgkBY4x
…use stale pre-merge MERGED_PR_NUM (fn-66 R3/R10, completion-review)
Owner
Author
|
@codex review |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d17861278e
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".
…open-PR→In Review and merged→Done precede local task/completion rows (fn-66, PR #178 review) Two P1s: (A) all-done OPEN spec + open PR mapped in-progress not in-review (make-pr push no-op); (B) ungated/unknown-completion merged spec stayed in-review so land.merged never wrote Done. Reordered the table prEvidence-first; added fixtures S-K/S-L. Codex mirror regenerated.
Owner
Author
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d6d9d593bc
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".
…not push (full re-render clobbers tracker edits) — fn-66, PR #178 review push(spec) renders the full spec body + writeIssue before status, so opening a PR could overwrite human tracker-side edits just to flip In Review. reconcile runs the 3-way body merge that preserves edits and sets In Review in the same op. Codex mirror regenerated.
Owner
Author
|
@codex review |
|
Codex Review: Didn't find any major issues. 👍 Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
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". |
gmickel
added a commit
that referenced
this pull request
Jun 18, 2026
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
1 participant
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.
What & why
tracker-sync pushed Linear
Doneon local Flow completion (all tasks done + completion-reviewSHIP) even with no merged PR — a SapienXT spec with no PR landed on the board asDoneand a human had to drag it back.Doneis a claim that the work shipped, so this was a correctness bug, not cosmetic.This makes the status policy lifecycle-accurate:
Doneis reserved for merge-confirmed PRs, an open PR maps toIn Review, completion-review never completes the issue, and pilot never declaresNO_WORKfor an all-done spec that hasn't shipped.R-ID coverage
flowToNormalized(spec, prEvidence); terminalDoneneedsMERGEDevidencestatus-sync.mdIn Reviewon the unconditional bridge-active PR-link pathmake-pr/workflow.mdland.merged→Doneonly on a fresh post-mergeMERGEDre-probe (sole driver)land/workflow.mdwork/phases.md, tracker-syncSKILL.md/steps.mdmake-pr; open PR →DEFERRED_TO_LAND; neverNO_WORKpilot/workflow.mdDone(NEEDS_HUMAN/non-terminal)status-sync.md,pilot/workflow.mdstatus-sync.mdgithub.md,adapter-interface.mdland.mergedactive-by-default when bridge active; per-write merge-evidence invariant incl. manual reconcileflowctl.py,test_land_tracker_event.pyCritical changes — where to look
references/status-sync.md— the policy core:flowToNormalized(spec, prEvidence)with aprEvidenceenum (merged|open|closed-unmerged|none|ambiguous|probe-error); terminal impossible withoutMERGED; In Review rung; fixtures S-G..S-J. Who-wins / deadlock ordering untouched (gate is upstream).flow-next-land/workflow.md—land.mergedis the sole Done driver, active-by-default when the bridge is active, and (completion-review fix) re-probesMERGEDfresh after the merge rather than reusing the stale pre-mergeMERGED_PR_NUM(which is empty on the normal OPEN→merge path — would've silently skipped the Done push).flow-next-work/phases.md+ tracker-syncSKILL.md/steps.md— completionReview dispatchreconcile→comment; setStatus merge-evidence guard.flow-next-make-pr/workflow.md— In Review on the unconditional PR-link path.flow-next-pilot/workflow.md+SKILL.md— newDEFERRED_TO_LANDverdict (open PR), no-PR all-done → make-pr, neverNO_WORK.flowctl.py(+ dogfood copy) —land.mergeddefault; newtest_land_tracker_event.py. Codex mirror regenerated; flow-next.dev updated (gmickel/flow-next.dev@5c53197, 2.1.2).Decision context
The merge-evidence gate is a per-write invariant on the outbound terminal write — no touchpoint or manual
/flow-next:tracker-syncreconcile writesDonewithout aMERGEDprobe (so a manual reconcile can still recoverDoneonce a merge exists). Three distinct evidence-backed states end to end: done locally (In Progress) → in review (open PR) → shipped (merged). Two bugs caught + memory'd during the work: the policy-map needing a matching reconcile branch; and "invert a policy claim → sweep ALL surfaces" (both ceremony copies, docs, CLI header).Verification
python3 -m unittest discover -s plugins/flow-next/tests, 1110);ci_test.sh67/0.land.mergedactivation, theflowToNormalized(spec, prEvidence)shape, per-write recovery).MERGED_PR_NUMDone-skip (R3/R10).🤖 Generated with Claude Code via
/flow-next:make-pr