[aliromman]

The new choose_current_research_task.py CLI gives all 16 generic orientation cards the current selected-baseline command. The emitted command passes on the admitted PR4610 certificate; the old command reproduces KeyError: 'bit_lower'. This is a tooling repair. The complete kappa calculation and selected frontier are unchanged.

The implementation, regression controls and contract note show the exact boundary. Only the literal generic command is resolved. Each affected suggestion exposes its unchanged original entry as catalog_first_step and an explicit resolution record. The historical picker script, catalogue bytes, pure selection result, package original_card, task, method, scope key, scientific commands and acceptance gates are preserved. The new entry point delegates arguments to the frozen picker CLI and resolves its JSON output. Help and error status/text pass through. AGENTS.md and the standing procedure direct new work to this entry point; the root guide remains frozen because two packages use it at runtime. Missing or deleted current helpers fail; deliberate HEAD-matching sparse omissions are reported for materialization.

Base is 1ec1119972c9aefe42525172d07b332fcff81cb4 (selected record). This closes the emitted generic-command part of the scoped handoff. PR4906 documented 57 catalogue runtime consumers and repaired the standing procedure. PR5359 supplies a separate authenticated historical complex-ceiling replay; its scientific command is outside this change.

With the selected source closure materialized, run:

python3 -B scripts/choose_current_research_task.py --seed 00000000000000000000000000000000 --task portable-replay
python3 -B scripts/research_selected_baseline.py
python3 -B -m unittest discover -s tests -p test_research_current_assignment.py
python3 -B -m unittest discover -s tests -p test_research_coordination.py
python3 -B -m unittest discover -s tests -p test_research_selected_baseline.py
python3 -B -m unittest discover -s tests -p test_research_onboarding.py

Expected: current and original first steps are both visible; baseline status is PASS_RESEARCH_BASELINE_BINDINGS_ONLY; tests report OK. Local evidence: all 22 historical/current assignment tests pass normally and under optimized Python 3.12.13; nine coordination, 13 baseline and all 27 Python 3.12 onboarding tests also pass. The current tests include exact child help/error propagation and a byte-exact before/after comparison of historical CLI output. The actual emitted command was executed and returns selected kappa 15108879/5000000000. The replay planner on the committed diff selects zero research packages, and git diff --check passes.

The initial hosted run exposed an additional frozen consumer of the picker script and failed on that package's pre-existing selected-record drift. The final design preserves that script and both pinned root-guide consumers. It does not repin or modify the historical package.

This does not make every historical scientific card current. Scientific input migration, a current helper-inventory law, native replay, full supplier, Lean and Docker verification remain outside this repair. Source-binding success is not theorem acceptance. Full-history replay was not run in the shallow checkout.

Credit Mihir Srivastava for the migration handoffs and selected-baseline successor, Douglas Colkitt and the retained picker/source-checker contributors, and PR3725 for the frozen-receipt precedent. New CLI resolution and tests: Ali Romman with substantial OpenAI Codex assistance; Apache-2.0. No independent human review is claimed.