Fix Google self-heal: drop impossible messages.google.com:OSID requirement, persist rotated cookies#148
Merged
Merged
Conversation
MaxGhenis
added a commit
that referenced
this pull request
Jul 22, 2026
Adversarial review of the OSID self-heal fix flagged the one real operational risk: googleCredentialRepairer.RepairCredentials has no rate limit, and the supervisor's 5-same-fingerprint breaker resets on any successful activity (resetFailureHistory on reaching Online). So if Google ever revokes the exported cookie session on a short (~minutes) cadence, connect -> brief activity -> 401 -> repair -> reconnect could churn indefinitely and risk account-level over-connect throttling. Add a cooldown that paces automatic repairs to >= a minimum interval (default 90s, override via OPENMESSAGE_REPAIR_MIN_INTERVAL). The wait runs on the parent context BEFORE the 20s refresh-timeout context, so a too-soon repair is delayed rather than hard-parked (parking would strand the session: Blocked is terminal with no auto-retry). Transparent to legitimate expiry, which is measured at >= ~14 min apart. A pacedCount (exposed via PacedRepairCount) increments on each delayed repair and doubles as the "fast-revocation is actually happening" signal for the durability watch. ctx cancellation during the wait returns without flagging repair or refreshing (it's a shutdown, not a failure). Scope: cmd/google_supervisor.go only; internal/bridge untouched. Tests cover first-repair-immediate, pace-too-soon (>= interval + count), after-interval-immediate, and cancel-no-flag-no-refresh with an injected clock; counterfactual confirms the pacing test fails without the gate. Follow-up to b27bdd6 on the held PR #148 branch. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ement, persist rotated cookies
The reconnect watchdog's cookie refresh hard-required a
messages.google.com:OSID Chrome cookie. OSID is a per-subdomain Google
service cookie that only exists if the user has opened Messages-for-web
in that Chrome profile — on profiles that never did, every credential
repair failed with "missing required cookies: messages.google.com:OSID"
and the app looped in needs_repair forever; re-pairs bought minutes.
Proven live (2026-07-20, no re-pair): with only fresh .google.com
account cookies (SID/HSID/SSID/APISID/SAPISID + SAPISIDHASH), both
/web/config and the RegisterRefresh RPC return 200 and mint fresh 24h
tachyon tokens; a refreshed snapshot stays valid ~14 minutes while
Chrome concurrently advances the account session, then dies with
SESSION_COOKIE_INVALID — so durable health is the watchdog heal cycle,
which this unblocks.
- googlecookies: require only the five .google.com account cookies;
carry messages.google.com OSID when present (preferred, never
required); deterministic host tie-break via ORDER BY host_key, name
- scripts/refresh-google-session-cookies-{macos,linux}.py: mirror the
same required/preferred split and ordering
- client/events: persist rotated cookies to session.json (sha256
change detection, 5-min throttle, injectable clock) so restarts
resume from fresh values instead of pair-time snapshots; take
AuthData.CookiesLock around session marshals
- runbook: correct the self-healing section (requirements, ~14-min
heal-cycle steady state, revive-don't-re-pair guidance)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adversarial review of the OSID self-heal fix flagged the one real operational risk: googleCredentialRepairer.RepairCredentials has no rate limit, and the supervisor's 5-same-fingerprint breaker resets on any successful activity (resetFailureHistory on reaching Online). So if Google ever revokes the exported cookie session on a short (~minutes) cadence, connect -> brief activity -> 401 -> repair -> reconnect could churn indefinitely and risk account-level over-connect throttling. Add a cooldown that paces automatic repairs to >= a minimum interval (default 90s, override via OPENMESSAGE_REPAIR_MIN_INTERVAL). The wait runs on the parent context BEFORE the 20s refresh-timeout context, so a too-soon repair is delayed rather than hard-parked (parking would strand the session: Blocked is terminal with no auto-retry). Transparent to legitimate expiry, which is measured at >= ~14 min apart. A pacedCount (exposed via PacedRepairCount) increments on each delayed repair and doubles as the "fast-revocation is actually happening" signal for the durability watch. ctx cancellation during the wait returns without flagging repair or refreshing (it's a shutdown, not a failure). Scope: cmd/google_supervisor.go only; internal/bridge untouched. Tests cover first-repair-immediate, pace-too-soon (>= interval + count), after-interval-immediate, and cancel-no-flag-no-refresh with an injected clock; counterfactual confirms the pacing test fails without the gate. Follow-up to b27bdd6 on the held PR #148 branch. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
MaxGhenis
force-pushed
the
claude/beautiful-hertz-fb2a1c
branch
from
July 22, 2026 00:20
0d79c3e to
c548c8c
Compare
MaxGhenis
added a commit
that referenced
this pull request
Jul 22, 2026
…-save retry, paced-repair observability (#151) Non-blocking polish from PR #148's cross-family review (sol). #148 made maybePersistRotatedCookies rewrite session.json every ~5 minutes, which turned two latent nits into real exposure: - internal/client/session.go: SaveSession now writes a same-dir 0600 temp file, fsyncs, and renames over session.json instead of os.WriteFile in place, so a crash mid-save can no longer truncate the only copy of the paired credentials. Mirrors googlecookies.UpdateSessionCookies, plus fsync. Regression test proven red against the old implementation. - internal/client/events.go: a failed rotated-cookie save retries after interval/10 (~30s) instead of waiting the full ~5-min throttle, still bounded so a persistent failure can't write per event; lastCookieHash stays untouched on failure so the retry still sees the pending rotation. Test proven red against the old advance-before-save semantics. - Paced-repair observability: delayed repairs now log a warn (wait/since_last_repair/paced_total) and publish google.repairs_paced in /api/status and MCP get_status (omitted when 0; only the transport-owning daemon builds a repairer). Runbook documents the check and the healthy-is-zero expectation. All touched packages pass with -race; full CI green (Go Test, Go Race, Web E2E, macOS App Build, Site Build). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Owner
Author
|
Correction to this PR's description (posting rather than editing, so the record stays intact). A verified-Fable calibration audit found two mechanism claims here that don't hold up:
Neither correction affects the fix itself: the root cause (an impossible Separately, credit where due: the earlier analysis in #108 had already independently identified all three components that shipped here — the OSID-absence lead, the cookie-persistence gap (with file:line cites that were more accurate than this PR's), and the missing rate limit on the repair path. The disagreement was only over whether client-side durability was achievable at all. Runbook follow-up in #153. |
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.
The bug (live install, reproduced 4×)
A fresh Google re-pair connects, holds ~3-4 minutes, then flips to
auth_expired/needs_repairwith "Google Messages session cookie expired; refreshing and reconnecting…" and loops there forever. Re-pairing buys minutes, never a fix.Root cause (code-verified, then proven live)
internal/googlecookies.LoadChromeCookies) hard-required amessages.google.com:OSIDChrome cookie. OSID is a per-subdomain Google service cookie — it exists only if the user has opened Messages-for-web in that Chrome profile. On profiles that never did (zeromessages.google.comrows), every repair failed withmissing required cookies: messages.google.com:OSID→FlagGoogleNeedsRepair→ backoff → retry → forever. Same impossible requirement in both refresh scripts.UpdateCookiesFromResponse) but the app persisted them only onAuthTokenRefreshed(~daily), so each generation/restart resumed from pair-time cookies.Live evidence (no re-pair, probe against a copy of the real session)
.google.comaccount cookies (SID/HSID/SSID/APISID/SAPISID + SAPISIDHASH),/web/configand theRegisterRefreshRPC return 200 and mint fresh 24h tachyon tokens — the "dead" session revived instantly.SESSION_COOKIE_INVALIDwhile the config GET still returned 200 — Google invalidates replayed browser-cookie snapshots once Chrome advances the account session. Fresh-pair sessions died in ~3-4 min.Changes
internal/googlecookies: require only the five.google.comaccount cookies; carrymessages.google.comOSID when present (preferred, never required); deterministic host tie-break (ORDER BY host_key, name); truthful docs.scripts/refresh-google-session-cookies-{macos,linux}.py: mirrored.internal/client/events.go: throttled persistence of rotated cookies (sha256 change detection, 5-min window, injectable clock for tests);CookiesLockheld around session marshals (pre-existing race).docs/agent-runbook.md: self-healing section rewritten to the proven mechanics (revive, don't re-pair).Verification
go test -race ./internal/googlecookies/ ./internal/client/ ./internal/bridgeadapters/google/ ./internal/app/✅;go test ./cmd/✅;go vet ./...✅;go build ./...✅; both scriptspy_compile✅requiredCookieschange makesTestLoadChromeCookiesDoesNotRequireMessagesOSIDfail; restoring passes./api/statuswatch: in progress, will report on this PR.Coordination note: deliberately scoped away from
internal/bridge/and the supervisor (in-flight refactor); PR #145 untouched.🤖 Generated with Claude Code