provider
/provider
Sprawdza lub wybiera połączenie wykonawcze Clade, rozdzielając środowisko agenta, dostawcę wnioskowania, protokół i model, m.in. dla Claude, Codex, Anthropic, OpenAI i bramek własnych.
- Autor
- shenxingy ↗
- Repozytorium
- clade ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 24.10.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 5
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 24.10.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Sprawdza lub wybiera połączenie wykonawcze Clade, rozdzielając środowisko agenta, dostawcę wnioskowania, protokół i model, m.in. dla Claude, Codex, Anthropic, OpenAI i bramek własnych.
Oryginalny opis (EN)
Inspect or select a Clade execution connection while keeping agent runtime, inference provider, wire protocol, and model distinct. Use for Claude/Codex runtime selection, Anthropic/OpenAI/MiniMax/Kimi/custom gateways, model routing, or connection troubleshooting.
Jak działa
- Canonical Clade workflow
- No argument / inspect
- Select or change
- Clade orchestrator configuration
- Completion
- Codex surface adapter
- Additional skill reference
- Delivery completion
Podgląd SKILL.md
---
name: provider
description: "Inspect or select a Clade execution connection while keeping agent runtime, inference provider, wire protocol, and model distinct. Use for Claude/Codex runtime selection, Anthropic/OpenAI/MiniMax/Kimi/custom gateways, model routing, or connection troubleshooting."
---
# Clade for Codex
This package composes the provider-neutral Clade core contract with the native
Codex surface adapter. Run the workflow directly in Codex; do not launch
another agent CLI or route it through Clade MCP.
Package provenance:
- core contract: `clade.execution/v1`
- surface adapter: `codex/v1`
- explicit invocation: `$clade:provider`
- generated from: `configs/skills/<name>`
## Canonical Clade workflow
<command-metadata>
name: provider
contract: clade.execution/v1
completion-status: DONE | DONE_WITH_CONCERNS | BLOCKED | NEEDS_CONTEXT
</command-metadata>
Treat these as independent dimensions:
1. surface;
2. agent runtime (`claude`, `codex`, or another installed adapter);
3. secret-free connection identity;
4. inference provider;
5. wire protocol;
6. opaque provider-scoped model ID;
7. capability profile and task policy.
Never infer provider/capabilities from a model prefix. Never silently fall
back to Claude, a default account, or another billing identity.
## No argument / inspect
Show the resolved execution identity and provenance. List available connection
identities only from trusted user/runtime configuration. Do not read or print
credentials, raw authorization headers, or secret-bearing files.
When the orchestrator provider registry is available, report its catalog
state (`fresh`, `stale`, `unavailable`, or `declared`), observation time,
adapter/profile provenance, health, and capability provenance. A live catalog
is a time-bounded observation, not a permanent model allowlist.
## Select or change
1. Parse the requested runtime, connection, model, effort, and task profile.
2. Read trusted repository policy and the current surface adapter.
3. Validate that the connection belongs to the selected runtime.
4. Resolve required/preferred/optional/forbidden capabilities.
5. Refresh an expired discovery-managed catalog before selection.
6. Preview requested vs. resolved values, catalog provenance, and explicit
degradations.
7. Change only the requested user-scoped selection. Repository files may name
a connection/profile but must not supply credentials or silently mutate
user-level provider configuration.
8. Re-read the effective configuration and report whether a restart/new
session is required.
Unknown required capabilities fail before spending tokens or changing Git.
Unknown preferred capabilities may continue only with a visible degradation.
If discovery fails or expires, continue only when the resolved opaque model ID
is explicitly pinned on that connection. Mark the execution as a stale catalog
degradation; never substitute another model, account, or provider.
## Clade orchestrator configuration
Use canonical fields:
```json
{
"agent_runtime": "claude",
"runtime_connections": {"claude": "minimax-work"},
"connections": {
"minimax-work": {
"agent_runtime": "claude",
"inference_provider": "minimax",
"wire_protocol": "anthropic-compatible",
"endpoint_identity": "minimax-user-config",
"models": {"strong": "MiniMax-M2.5"},
"pinned_models": ["MiniMax-M2.5"],
"discovery": {
"adapter": "minimax",
"store": "claude-providers",
"profile": "minimax-work",
"ttl_seconds": 300
},
"capabilities": {}
}
}
}
```
This object is metadata only. Keep endpoint URLs, API keys, tokens, shell
exports, and machine-specific credential paths in the native user store.
`discovery.store` and `discovery.profile` are secret-free references to that
store. Supported adapters are Anthropic, OpenAI, MiniMax, Moonshot,
OpenAI-compatible custom gateways, and explicitly declared native-static
profiles.
## Completion
- `DONE`: selection is resolved and verified.
- `DONE_WITH_CONCERNS`: selection works with named degradation/restart need.
- `BLOCKED`: required capability or trusted connection is unavailable.
- `NEEDS_CONTEXT`: the user must choose among materially different accounts,
costs, or authority boundaries.
## Codex surface adapter
# Codex connection adapter
- Invoke the installed plugin workflow explicitly as `$clade:provider`; bare
`$provider` is not the Clade plugin identity.
- Use user-scoped Codex `model_provider`/model configuration and its native
credential mechanism. Project config is trust-gated and must not donate
credentials.
- Prefer the Responses protocol when supported; treat Chat Completions or a
compatible gateway as an explicit protocol choice.
- Preserve unknown custom provider/model IDs. Validate capabilities through
declared metadata or probes, never prefixes.
- A discovery-managed connection references the trusted user profile as
`store: codex-config` plus its `model_providers` profile name. Model-list
probes may use that profile's endpoint and credential environment variable,
but neither value may enter repository settings, status, or cache output.
- Start a new session if the installed Codex version does not reload provider
configuration dynamically.
## Additional skill reference
# Provider
Manage secret-free connection identities and produce a
`clade.execution/v1` preview before changing execution selection. Read
`prompt.md` for the semantic workflow and only the matching file under
`surfaces/` for native configuration mechanics.
## Delivery completion
If this workflow changes files or external state:
- Inspect the real final state before responding, including `git status` for a
repository task.
- Never report `DONE` while task-owned changes are uncommitted. Use or continue
`$clade:delivery` and create a repository-compliant checkpoint or preserve
the work when committing is unavailable.
- When the user request or trusted repository policy makes publication,
deployment, or live verification part of the task, do not silently downgrade
the result to local-only work.
- If a required delivery transition lacks authority, credentials, a destination,
or reachable external state, report `BLOCKED` or `NEEDS_CONTEXT` rather than
appending a "not committed/pushed/deployed" caveat after `DONE`.
Pliki w paczce (5)
- SKILL.md 6 KB
- surfaces/claude-code.md 1 KB
- surfaces/codex.md 1 KB
- surfaces/generic.md 1 KB
- surfaces/mcp.md 1 KB
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · shenxingy/clade | 100 | MIT |
| evanca/flutter-ai-rules | 100 | MIT |
Migawka bazy z 28.10.2026 · wersja paczki 4d220d0f6821 (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
- 100clade-core-workflow-a
Przekierowuje do claude-model-inference po wzorce streamingu Messages API, vision i structured output, jako główny workflow Anthropic. - 100clade-model-inference
Obsługuje streaming odpowiedzi Claude, prompty systemowe, rozmowy wieloturowe i ustrukturyzowane wyjście przez Messages API. - 100sync-model-catalog
Odświeża katalogi modeli Agenta, listy dostawców OpenAI, Anthropic, Codex, Pi i OpenRouter oraz domyślne ustawienia przed wydaniem. - 95claude-api
Buduje aplikacje z Claude API lub Anthropic SDK, gdy kod importuje anthropic, @anthropic-ai/sdk lub claude_agent_sdk. - 100claude-api
Buduje aplikacje z Claude API lub SDK Anthropic, gdy kod importuje anthropic, @anthropic-ai/sdk lub claude_agent_sdk albo użytkownik prosi o Claude API lub Agent SDK. - 100claude-api
Buduje aplikacje z Claude API lub SDK Anthropic, gdy kod importuje anthropic, @anthropic-ai/sdk lub claude_agent_sdk albo użytkownik prosi o Claude API lub Agent SDK.