mode-selection
/mode-selection
Wybiera zatwierdzony przez operatora tryb mode (project albo bounty) przed uruchomieniem, bo zgoda na pracę nie zatwierdza jeszcze procesu, bez konstruowania pól i routingu recenzentów.
- Autor
- mtarcure ↗
- Repozytorium
- claude-vibe-squad ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 20.09.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 20.09.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Wybiera zatwierdzony przez operatora tryb mode (project albo bounty) przed uruchomieniem, bo zgoda na pracę nie zatwierdza jeszcze procesu, bez konstruowania pól i routingu recenzentów.
Oryginalny opis (EN)
Use when choosing the operator-approved `mode` (`project` versus `bounty`) before launch—approval of the work does not approve that workflow. Not for constructing fields or reviewer routing.
Jak działa
- Name the mode and get approval — it is its own consent
- Both dispatch paths preserve an explicit choice
- Verify the mode that LANDED, never the one you intended
- The mode changes the contract, not just the wall
Podgląd SKILL.md
---
name: mode-selection
audience: chrono
description: Use when choosing the operator-approved `mode` (`project` versus `bounty`) before launch—approval of the work does not approve that workflow. Not for constructing fields or reviewer routing.
---
# Mode Selection
House procedure for choosing a dispatch mode before authoring the packet. The mode is a separate
operator decision from the work itself, and the wrong mode silently changes the wall, the
verification contract, and the gates a lane runs under. Verify every claim here against the current
scripts — do not trust this doc over the code.
## Name the mode and get approval — it is its own consent
- `chrono/CLAUDE.md` Dispatch step 1 requires opening the chosen file under `shared/modes/`, stating
which mode the work will run under, and waiting for the operator to agree. Hard Rule 1 forbids a
mode starting without explicit consent.
- **Approving the work is not approving the mode.** A convenience wrapper's silent default sent
most of a campaign's lanes under a mode nobody chose — without telling the operator. Say the mode
word out loud; never infer it from "yes, go".
## Both dispatch paths preserve an explicit choice
- Generated packets use `scripts/send-task.sh ... --mode project|bounty`. The wrapper rejects an
omitted or invalid value, writes the exact choice into frontmatter, and never supplies a default.
- Prepared packets carry `mode: project|bounty` in frontmatter and dispatch with
`bin/send-task.sh <packet-file>`, which carries the packet's own `mode` into the compiled contract
(`MODE_VALUE` -> `"mode"`).
- The wrapper survives because it still assembles standard frontmatter, resolves lanes and review
metadata, and routes through the hardened dispatcher. Retire it only if those conveniences move
elsewhere; do not retain or reintroduce a mode default as its reason to exist.
## Verify the mode that LANDED, never the one you intended
- The dispatch context carries the landed mode as a real field. Read it from the **exact attempt you
dispatched**, never from a glob — attempt ids are UUIDs, so glob order is not chronological and
`head -1` can hand you a previous attempt's mode after a retry:
```bash
python3 -c 'import json,sys; a=json.load(open(sys.argv[1]))["authority"]; \
p=a.get("mode_profile"); m=(a.get("memory_context") or {}).get("mode"); \
print(p if p==m else f"MISMATCH profile={p} memory={m}")' <context_path>
```
`mode_profile` and `memory_context.mode` must agree; if they disagree, stop — do not pick one.
- A wrapper `--dry-run` echoes `Packet mode: <mode>` and derives that mode's contract. It proves the
generated packet's choice but not a landed attempt; a green prepared-packet dry-run likewise does
not replace the exact-attempt check.
- If the landed mode differs from what the operator approved, stop and say so before the lane works.
## The mode changes the contract, not just the wall
- Walls come from `timeout_budget_for_mode()` in `dispatch_context_builder.py`: a flat 2700s for
`project`, 3600s for `bounty`. Scope to the emitted number; do not infer it from a ratio.
- `verification_contract.py` sets a different contract per mode. `project` requires
`project_tests`/`recipient_contract` and mandatory recall+record. `bounty` swaps in
`scope_gate`/`no_self_inflicted`/`poc_reproduction`, an exact target allowlist, forbids
submission-attempted, and makes recall OPTIONAL (a cold lane cannot recall without inheriting other
runs' conclusions).
- `project` and `bounty` both carry a `plan_review_policy` with `anti_affinity: author_family` — an
independent-family plan check is part of the contract. That is the controller's plan review and is
separate from the deliverable's own `review_model`; a `project` packet with `review_model: none`
still launches.
Wersje
W bazie nie ma innych skilli o nazwie mode-selection.
Migawka bazy z 13.11.2026 · wersja paczki 6f835286e15b (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100bounty-qualifier
Kwalifikuje otwarte bounty kodowe dla agentów AI według wartości oczekiwanej: liczy EV z nagrody netto, szansy wygranej i kosztu wysiłku, pomijając repozytoria bez testów.