Across squads using the same catalog of cost-visible agent workflows, what happens if leaders turn off per-run token-cost visibility for a subset of mature, ‘golden’ workflows while keeping it on for all experimental or shadow variants—does this asymmetry (a) make durable adoption of those golden workflows more stable by reducing token anxiety, or (b) backfire by weakening the cost signals needed to prune or evolve them over time?

Answer

Net: hiding per-run prices on mature golden workflows (while keeping cost-visible experimental/shadow variants) usually stabilizes day-to-day use of those golden workflows by reducing token anxiety, but it weakens local cost signals and can slow pruning unless authors and leads still see detailed costs via portfolio dashboards.

(a) Effect on durable adoption

• Most devs treat golden workflows as “approved tools” rather than personal spending decisions, so usage is steadier and less anxious.
• High-cost but high-value golden paths (e.g., incident-fix bundles) are more likely to be used when needed instead of avoided because a number looks scary.
• Experimental/shadow variants stay cost-visible, so exploration still feels bounded and budgeted.

(b) Effect on pruning and evolution

• If authors and leads retain fine-grained cost views at the workflow/portfolio level, pruning and slimming can continue with little loss: cost noise is hidden from daily users, not from governance.
• If cost visibility is removed for everyone on golden workflows, they tend to ossify: fewer local challenges, slower evolution, and late detection of cost drift.
• Shadow variants with visible cost can over-optimise around tokens if promotion rules are unclear, leading to tension with opaque golden standards.

So: asymmetry can support durable adoption of golden workflows without killing refinement, but only if detailed cost remains visible to authors/leads and is used in explicit, workflow-centric governance.