docs(agents): discuss upstream features that duplicate Pylon's own

[rynfar]

The upstream review skill told agents to preserve Pylon behavior, with no guidance for the case where upstream independently ships something Pylon already built. That default silently resolves a question the maintainer wants to answer, and the source tends to get filed as "already covered" without anyone comparing the two implementations.

Records the standing preference in both places an agent reads: a boundary bullet in SKILL.md and a scope row in the decision framework. Neither version wins by default — surface the overlap, compare on merits, and treat adopting upstream's implementation and deleting Pylon's as a legitimate outcome. Preservation stays the default for adaptations and conflicts.

Instructions only; no product code, source disposition, review cursor or Git ancestry changes. Table reflow in the framework is the formatter realigning columns for the new row.

Model: Claude Opus 5 (1M context), harness: Pylon / Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[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	13.8 KiB	13.8 KiB	+17 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	7.1 KiB	7.1 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.7 KiB	6.7 KiB	+15 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.4 KiB	57.4 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	10	10	0 (0.0%)	21	✅
Claude	Total thread wire	13.8 KiB	13.8 KiB	−32 B (−0.2%)	15.1 KiB	✅
Claude	Thread snapshot wire	7.1 KiB	7.1 KiB	−6 B (−0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.7 KiB	6.7 KiB	−26 B (−0.4%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	58.1 KiB	58.1 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	8	8	0 (0.0%)	21	✅

Baseline: 76db4b5 · PR result: ba46e71 · 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: 111.2 KiB
• Claude decoded thread snapshot: 111.9 KiB

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