open/ktg-plugin-marketplace
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity Actions

docs(ultraplan-local): seal Path C closed with Q3 NEGATIVE finding [skip-docs]

Browse source 
D1 of post-v3.4.1 stabilisering. Path C (cache-warm sentinel + identical-tool
parallel) is closed 2026-05-05 per Q3 experiment NEGATIVE result:
median cache_creation_input_tokens = 163,903 across 3 fork-children at
186K parent context (CC v2.1.128, Sonnet 4.6).

Master-plan thresholds: <= 1,500 POSITIVE / >= 3,500 NEGATIVE — NEGATIVE
solidly. CLAUDE_CODE_FORK_SUBAGENT does not preserve cache prefix across
identical-tool children at our context size.

Path C migration is deferred indefinitely. Reassessment is appropriate
when CC v2.2.xxx ships fork-cache-relevant features. Harness
(scripts/q3-cache-prefix-experiment.mjs) and analyser
(lib/stats/cache-analyzer.mjs) remain available for re-run.

Brief: .claude/projects/2026-05-04-spor-c-q3-cache-prefix-experiment/brief.md
Result: q3-experiment-results.local.md (gitignored)
This commit is contained in:
Kjell Tore Guttormsen 2026-05-05 10:36:36 +02:00
commit f339437e6d
1 changed files with 3 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

6
plugins/ultraplan-local/CLAUDE.md
View file

@ -157,15 +157,15 @@ All four pipeline commands accept `--gates {open|closed|adaptive}`:
Under the hood: `lib/util/autonomy-gate.mjs` runs the state machine `idle → approved → executing → merge-pending → main-merged`. `lib/stats/event-emit.mjs` records each transition to `${CLAUDE_PLUGIN_DATA}/ultra*-stats.jsonl`. The main-merge gate is the final autonomy boundary before HEAD lands on `main`.
### Path A/B/C decision (v3.4.0)
### Path A/B/C decision (v3.4.0; Path C closed 2026-05-05)
Three architectural options were considered for the speedup work:
- **Path A — cache-first** (drop `--allowedTools` per child to recover cross-phase cache sharing): REJECTED. Inverts the security model; plugin hooks don't fire reliably in `claude -p` (research/06 GH #36071).
- **Path B — sequential `--no-ff` parallel waves with manifest-driven failure recovery**: CHOSEN. Ships in v3.4.0. Phase 2.6 of `/ultraexecute-local` runs the wave executor with hardenings for plugin-in-monorepo + gitignored-state topology.
- **Path C — hybrid (cache-warm sentinel + identical-tool parallel)**: DEFERRED to v3.5.0 contingent on cache-telemetry data harvested by `lib/stats/event-emit.mjs`. Path C requires unverified Q3 (CLAUDE_CODE_FORK_SUBAGENT cache-prefix preservation at 150-250K context).
- **Path C — hybrid (cache-warm sentinel + identical-tool parallel)**: **CLOSED 2026-05-05.** Q3 experiment measured median `cache_creation_input_tokens` = 163,903 across 3 fork-children at 186K parent context (CC v2.1.128, Sonnet 4.6). Master-plan thresholds: ≤ 1,500 POSITIVE / ≥ 3,500 NEGATIVE. Result is solidly NEGATIVE — `CLAUDE_CODE_FORK_SUBAGENT` does not preserve cache prefix across identical-tool children at our context size. Path C migration is deferred indefinitely; reassessment is appropriate when CC v2.2.xxx ships fork-cache-relevant features. Harness: `scripts/q3-cache-prefix-experiment.mjs`. Companion analyser: `lib/stats/cache-analyzer.mjs`.
Path C migration would require: (1) re-architecting tool-list to be identical across all wave children, (2) cache-telemetry analysis confirming Q3 holds, (3) prompt-level deny re-enablement to compensate for tool scoping rollback.
A revived Path C (post-v2.2.xxx) would require: (1) re-architecting tool-list to be identical across all wave children, (2) cache-telemetry analysis confirming the new fork-cache behaviour holds, (3) prompt-level deny re-enablement to compensate for tool scoping rollback.
## Architecture