Cite as: Real Problem AI problem “Why can't Claude Code tell you what a feature actually cost across the sessions it took to build?”. Opportunity score 7.5 out of 10 (severity 6, AI feasibility 9, market signal 7, competition gap 8). Category AI / Agents. Trend LLM. Source signal: anthropics/claude-code GitHub issue #34803, 16 Mar 2026, closed as not planned.. Canonical URL: https://www.realproblem.ai/archive/why-cant-claude-code-tell-you-what-a-feature-actually-cost.
Why can't Claude Code tell you what a feature actually cost across the sessions it took to build?
The /cost command only shows the current session; nothing persists or aggregates tokens, cost, or session count per project, so teams can't report or budget AI usage against real work.
Who has it: Small dev teams and solo builders using Claude Code as their primary coding tool.
Evidence
The issue describes having no way to see cumulative Claude Code usage across sessions for a project, because the cost command covers only the current session.
Our summary of the public post linked below, not a quote. Nobody submitted it to Real Problem AI.
anthropics/claude-code GitHub issue #34803, 16 Mar 2026, closed as not planned.Why it is archived
Trimmed to 100-cap (lowest opportunity_score)
Scoring breakdown
Existing players
- Claude Code /cost · Session-scoped only, resets on restart, no persistence.
- Anthropic Console analytics · Org-level dashboards, not per-project or per-feature.
- Manual token estimation · Users ask Claude itself to estimate, unreliable and not logged.
What they are missing
A lightweight local usage ledger that persists per-project tokens, cost, session count, and duration, exportable for PR descriptions, sprint reports, and CI dashboards.
Stack hint
#ATB2 · Canonical URL: https://www.realproblem.ai/archive/why-cant-claude-code-tell-you-what-a-feature-actually-cost