feat: add foreman provider type for endpoint overrides

Accept "foreman" in both resolveModel (GADFLY_BASE_URL) and endpointProvider
(GADFLY_ENDPOINT_*) switches, mapping to majordomo's ollama.Foreman() preset
(handles foreman's non-streaming/long-poll quirks). Unlike the HTTPS-only
LLM_* foreman:// DSN, the base URL is verbatim, so a plaintext http:// foreman
queue works. Tests + README provider table + endpoint-aliases example updated.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

README.md

@@ -65,6 +65,7 @@ majordomo failover chain / alias) is used verbatim.
 |----------|-------------------|---------|--------|
 | **Ollama Cloud** (default) | `ollama-cloud` | `OLLAMA_API_KEY` / `OLLAMA_CLOUD_API_KEY` | ✅ in active use |
 | **Local Ollama** | `ollama` | none (`OLLAMA_HOST` or `GADFLY_BASE_URL` for a remote daemon) | ✅ tested |
+| **[foreman](https://gitea.stevedudenhoeffer.com/steve/foreman)** (native-Ollama queue daemon) | `foreman` + `GADFLY_BASE_URL`, or a `GADFLY_ENDPOINT_*` / `LLM_*` `foreman://` entry | optional bearer (via the endpoint/DSN) | ✅ native-Ollama path |
 | **OpenAI-compatible** (incl. local Ollama's `/v1`) | `openai` + `GADFLY_BASE_URL` | `OPENAI_API_KEY` (any non-empty for Ollama) | ✅ tested against Ollama |
 | **OpenAI** | `openai` | `OPENAI_API_KEY` | ⚠️ wired, **untested** |
 | **Anthropic** | `anthropic` | `ANTHROPIC_API_KEY` | ⚠️ wired, **untested** |
@@ -87,7 +88,8 @@ name with env vars and then reference `name/model` in `GADFLY_MODEL`/`GADFLY_MOD
 # http-capable (Gadfly-native) — base URL used verbatim, so plaintext LAN works:
 GADFLY_ENDPOINT_BIGBOX="ollama|http://192.168.1.50:11434"
 GADFLY_ENDPOINT_GPU="openai|http://gpu.lan:8000/v1|sk-local"
-GADFLY_MODELS="bigbox/qwen2.5-coder:7b,gpu/llama3.1"
+GADFLY_ENDPOINT_M1="foreman|http://foreman-m1:8080|tok"   # native-Ollama queue daemon
+GADFLY_MODELS="bigbox/qwen2.5-coder:7b,gpu/llama3.1,m1/qwen3:14b"
 
 # pure spec alias (a model, or a failover chain):
 GADFLY_ALIAS_FAST="bigbox/qwen2.5-coder:7b,ollama-cloud/gpt-oss:120b-cloud"
@@ -95,9 +97,9 @@ GADFLY_MODEL="fast"
 ```
 
 `<NAME>` is lowercased to form the registry name (`GADFLY_ENDPOINT_BIGBOX` → `bigbox`). This
-is the same idea as majordomo's built-in **`LLM_*` env DSNs** (`LLM_BIGBOX=ollama://tok@host`),
+is the same idea as majordomo's built-in **`LLM_*` env DSNs** (`LLM_BIGBOX=ollama://tok@host`,
-which Gadfly also honors — but those are **HTTPS-only**, so for plaintext local Ollama use
+`LLM_M1=foreman://tok@host`), which Gadfly also honors — but those are **HTTPS-only**, so for a
-`GADFLY_ENDPOINT_*` instead.
+plaintext local Ollama or `http://` foreman use `GADFLY_ENDPOINT_*` instead.
 
 > **Gitea Actions note:** repo `vars`/`secrets` aren't auto-exposed as env — add each alias to
 > the stub workflow's `env:` block, e.g. `GADFLY_ENDPOINT_BIGBOX: ${{ vars.GADFLY_ENDPOINT_BIGBOX }}`.

cmd/gadfly/model.go

@@ -79,6 +79,11 @@ func resolveModel() (llm.Model, error) {
 			opts = append(opts, ollama.WithToken(apiKey))
 		}
 		return ollama.New(opts...).Model(model)
+	case "foreman":
+		// foreman (gitea.stevedudenhoeffer.com/steve/foreman) is a native-Ollama
+		// queue daemon; the preset also smooths its non-streaming/long-poll
+		// quirks. Base URL is used verbatim, so a plaintext http:// foreman works.
+		return ollama.Foreman(baseURL, apiKey).Model(model)
 	case "anthropic":
 		opts := []anthropic.Option{anthropic.WithBaseURL(baseURL)}
 		if apiKey != "" {
@@ -92,7 +97,7 @@ func resolveModel() (llm.Model, error) {
 		}
 		return google.New(opts...).Model(model)
 	default:
-		return nil, fmt.Errorf("GADFLY_BASE_URL is set but GADFLY_PROVIDER %q has no endpoint-override support (use openai/ollama/anthropic/google, or unset GADFLY_BASE_URL to resolve via majordomo)", provider)
+		return nil, fmt.Errorf("GADFLY_BASE_URL is set but GADFLY_PROVIDER %q has no endpoint-override support (use openai/ollama/foreman/anthropic/google, or unset GADFLY_BASE_URL to resolve via majordomo)", provider)
 	}
 }
 
@@ -131,9 +136,12 @@ func buildSpec(provider, model string) string {
 //	GADFLY_ENDPOINT_<NAME> = "<provider>|<base-url>[|<key>]"
 //	    Registers a provider at an explicit endpoint. Unlike majordomo's LLM_*
 //	    DSNs (which are HTTPS-only), the base URL is used verbatim, so a
-//	    plaintext local Ollama works:
+//	    plaintext local Ollama (or foreman queue) works:
 //	        GADFLY_ENDPOINT_BIGBOX="ollama|http://192.168.1.50:11434"
 //	        GADFLY_MODEL=bigbox/qwen2.5-coder:7b
+//	    provider is one of ollama/foreman/openai/anthropic/google; "foreman"
+//	    targets a foreman daemon (native Ollama on the wire):
+//	        GADFLY_ENDPOINT_M1="foreman|http://foreman-m1:8080|tok"
 //
 //	GADFLY_ALIAS_<NAME> = "<majordomo spec>"
 //	    Registers a plain alias that expands inline (a model, or a failover
@@ -190,6 +198,11 @@ func endpointProvider(name, raw string) (llm.Provider, error) {
 			opts = append(opts, ollama.WithToken(key))
 		}
 		return ollama.New(opts...), nil
+	case "foreman":
+		// foreman is native-Ollama on the wire; the preset additionally handles
+		// its non-streaming degradation. Unlike the HTTPS-only LLM_* foreman://
+		// DSN, the base URL here is verbatim, so a plaintext http:// foreman works.
+		return ollama.Foreman(baseURL, key, ollama.WithName(name)), nil
 	case "openai", "openai-compatible":
 		opts := []openai.Option{openai.WithName(name), openai.WithBaseURL(baseURL)}
 		if key != "" {
@@ -209,6 +222,6 @@ func endpointProvider(name, raw string) (llm.Provider, error) {
 		}
 		return google.New(opts...), nil
 	default:
-		return nil, fmt.Errorf("unknown provider %q (use ollama/openai/anthropic/google)", provider)
+		return nil, fmt.Errorf("unknown provider %q (use ollama/foreman/openai/anthropic/google)", provider)
 	}
 }

cmd/gadfly/model_test.go

@@ -17,6 +17,21 @@ func TestEndpointProvider(t *testing.T) {
 			t.Fatalf("unexpected error: %v", err)
 		}
 	})
+	t.Run("foreman queue registers under its name", func(t *testing.T) {
+		p, err := endpointProvider("m1", "foreman|http://foreman-m1:8080|tok")
+		if err != nil {
+			t.Fatalf("unexpected error: %v", err)
+		}
+		// WithName(name) must win over the Foreman preset's default "foreman".
+		if p.Name() != "m1" {
+			t.Errorf("Name() = %q, want %q", p.Name(), "m1")
+		}
+	})
+	t.Run("foreman without token", func(t *testing.T) {
+		if _, err := endpointProvider("m5", "foreman|http://foreman-m5:8080"); err != nil {
+			t.Fatalf("unexpected error: %v", err)
+		}
+	})
 	for _, bad := range []string{"", "ollama", "noprovider-no-pipe", "mystery|http://x"} {
 		t.Run("rejects "+bad, func(t *testing.T) {
 			if _, err := endpointProvider("n", bad); err == nil {

examples/endpoint-aliases.yml

@@ -5,10 +5,15 @@
 # can then reference as "<name>/<model>" (NAME lowercases: BIGBOX -> bigbox).
 # The base URL is used verbatim, so plaintext http LAN endpoints work.
 #
+# provider is ollama / foreman / openai / anthropic / google. "foreman" targets a
+# foreman queue daemon (https://gitea.stevedudenhoeffer.com/steve/foreman) — native
+# Ollama on the wire, so just give it the daemon's URL (and optional bearer token).
+#
 # Gitea note: vars/secrets aren't auto-exposed as env, so map each alias here.
-# Suggested repo vars:
+# Suggested repo vars (and a secret when the value carries a token):
 #   GADFLY_ENDPOINT_BIGBOX = ollama|http://192.168.1.50:11434
 #   GADFLY_ENDPOINT_GPU    = openai|http://gpu.lan:8000/v1
+#   GADFLY_ENDPOINT_M1     = foreman|http://foreman-m1:8080|<token>   (use a secret)
 
 name: Adversarial Review (Gadfly)
 
@@ -49,8 +54,9 @@ jobs:
           # --- named endpoints (mapped from repo vars) ---
           GADFLY_ENDPOINT_BIGBOX: ${{ vars.GADFLY_ENDPOINT_BIGBOX }}   # "ollama|http://192.168.1.50:11434"
           GADFLY_ENDPOINT_GPU: ${{ vars.GADFLY_ENDPOINT_GPU }}         # "openai|http://gpu.lan:8000/v1"
+          GADFLY_ENDPOINT_M1: ${{ secrets.GADFLY_ENDPOINT_M1 }}        # "foreman|http://foreman-m1:8080|<token>"
           # one reviewer (one comment) per model, across the aliased endpoints:
-          GADFLY_MODELS: "bigbox/qwen2.5-coder:7b,gpu/llama3.1"
+          GADFLY_MODELS: "bigbox/qwen2.5-coder:7b,gpu/llama3.1,m1/qwen3:14b"
           # --- event context (leave as-is) ---
           EVENT_NAME: ${{ github.event_name }}
           PR: ${{ github.event.pull_request.number || github.event.issue.number || github.event.inputs.pr_number }}