Skip to content
fitme·story
fitme·story
View as
Case studies
v7.8.1 · 2 min read · 2026-05-07

Roadmap Stress Test — Compressing 10 Weeks of Backlog Through the v7.8.1 Protocol

Single-session experiment measuring throughput + protocol overhead of the v7.8.1 framework against ~10 weeks of sequenced backlog work. The meta-feature itself opted out of Mode C isolation (Q3 override) because nesting isolated worktrees inside an isolated worktree is excessive. Sub-features auto-isolated normally. Result: 1 sub-feature fully shipped (case-study comparison table via fitme-story #58) + 1 pre-complete (push-notifications-v2) + 1 partial (UCC passkey auth went 95% through cutover Parts 1-6 same session) + 6 deferred. Total: 1 fully + 1 pre-complete + 1 partial / 9 in scope. The experiment SURFACED 10 v7.9 candidates (F1-F10) which became the docket ranking input for the 2026-05-21 v7.9 promotion decision.

10
v7.9 candidates (F1–F10) surfaced by the single-session stress test — the durable output that became the docket ranking input for the 2026-05-21 v7.9 promotion
The experiment measured throughput, not completion: ~10 weeks of estimated backlog (9 sub-features) compressed into one ~3-hour session → 1 fully shipped + 1 pre-complete + 1 partial. The protocol-overhead findings, not the ship count, were the point.
What do T1 / T2 / T3 mean?▾
T1Instrumented — from a ledger/commit
T2Declared — stated, not measured
T3Narrative — estimate from memory

What this proved

The v7.8.1 protocol (Mechanism C session attribution + isolated worktree from Phase 1 onward + Tier 2.2 logging on every phase transition + Mechanism A coverage telemetry verification) scales through nested sub-features without protocol overhead exceeding the per-step work itself. The experiment was the empirical input that justified the 2026-05-21 v7.9 promotion decision (3 gates advisory → enforced).

The 10 candidates this experiment surfaced

F1-F10 (v7.9 candidates docket from docs/superpowers/specs/2026-05-08-framework-v7-9-candidates.md):

IDDomainStatus
F1-F5Various framework gapsmostly closed via v7.8.x cycle
F11-F13Cross-repo state syncshipped via v7.8.3 (2026-05-11)
F14Per-gate dispatch test enforcementSHIPPED 2026-05-22→05-23 via PR #451
F15Zero-coverage gate unit testsSHIPPED 2026-05-22→05-23 via PR #451 (joint)
F16pre-commit try-repo harnessv7.9.1 build window (~2026-06-04)
F17last_fired_at indexv7.9.1 build window
F18Mutation testingv8.0/v8.1 advisory→enforced

Full source

See docs/case-studies/roadmap-stress-test-2026-05-07-case-study.md for the full live append-only experiment journal including §99 resolution log + per-sub-feature timing data captured in .claude/features/roadmap-stress-test-2026-05-07/data-collection.json.

Why slot 25.5

This MDX showcase is being added retroactively on 2026-05-23 to close drift D-RECON-12 from docs/audits/internal/2026-05-23-comprehensive-pr-sync-audit.md. The case study itself shipped 2026-05-07 (v7.8.1 era) — slot 25 (framework-v7-8-1-branch-isolation) already exists for the predecessor v7.8.1 feature; this stress-test experiment ran the same day v7.8.1 shipped (parallel evening session) so it slots between v7.8.1 (slot 25) and v7.8.2 (slot 26 ucc-passkey-auth which the stress test continued).

Honest disclosures
  • •This is a single-session compressed run; 10 weeks of backlog cannot finish in one session. The experiment measures THROUGHPUT and protocol overhead, not completion.
  • •The meta-feature itself opted out of v7.8.1 isolation Mode C (Q3 override) — nesting isolated worktrees inside an isolated worktree is excessive. Sub-features auto-isolated normally.
  • •All measurements timestamped at moment of capture; no retroactive measurements.