fix(prime): recover completed tool cycles after reconnecting

[rynfar]

A native Prime tool call and its matching result can reach a complete reconnect snapshot before their live events. Pylon rejected this valid completed cycle and failed the turn.

Recover ordered, uniquely matched completed cycles for the same delivered prompt, while retaining exact transcript overlap, connection/proof fences, and the no-queued-input requirement. Reject unmatched or reused calls, wrong names, duplicate results, and extra user messages. Resolve delayed assistant starts and deltas against their exact completion before publishing them, within the existing event bounds. New prompt IDs stream normally. Suppress exact recovered completions so recovery cannot duplicate output or transcript progress.

Validation: the focused regression failed before the fix; all 483 adapter/runtime/event tests pass, scoped lint passes, and the server typecheck passes. The bounded adversarial checkpoint passed after verifying complete delayed assistant event sequences. Fresh real web/mobile acceptance is underway under #114. The original acceptance snapshot was not retained; this regression proves the specific valid schedule, without claiming the failed snapshot's exact shape was recovered.

Model: GPT-6 Codex. Harness: Pylon/Codex API.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
pylon-marketing	Ready	Preview	Sep 12, 2026 10:13am UTC

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	14.0 KiB	13.9 KiB	−124 B (−0.9%)	15.1 KiB	✅
Codex	Thread snapshot wire	7.2 KiB	7.2 KiB	−4 B (−0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.8 KiB	6.7 KiB	−120 B (−1.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	58.8 KiB	58.0 KiB	−866 B (−1.4%)	66.4 KiB	✅
Codex	Live turn messages	9	8	−1 (−11.1%)	21	✅
Claude	Total thread wire	14.1 KiB	14.1 KiB	+24 B (+0.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	7.2 KiB	7.2 KiB	+8 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.9 KiB	6.9 KiB	+16 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	59.7 KiB	59.7 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 5a09b2f · PR result: 4f32510 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 115.6 KiB
• Claude decoded thread snapshot: 116.3 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.