Files
hermes-agent/hermes_cli/models.py

676 lines
23 KiB
Python
Raw Normal View History

2026-02-22 02:16:11 -08:00
"""
Canonical model catalogs and lightweight validation helpers.
2026-02-22 02:16:11 -08:00
Add, remove, or reorder entries here both `hermes setup` and
`hermes` provider-selection will pick up the change automatically.
"""
from __future__ import annotations
import json
import os
import urllib.request
import urllib.error
from difflib import get_close_matches
from typing import Any, Optional
2026-02-22 02:16:11 -08:00
# (model_id, display description shown in menus)
OPENROUTER_MODELS: list[tuple[str, str]] = [
("anthropic/claude-opus-4.6", "recommended"),
("anthropic/claude-sonnet-4.5", ""),
("openai/gpt-5.4-pro", ""),
("openai/gpt-5.4", ""),
("openai/gpt-5.3-codex", ""),
2026-02-22 02:16:11 -08:00
("google/gemini-3-pro-preview", ""),
("google/gemini-3-flash-preview", ""),
("qwen/qwen3.5-plus-02-15", ""),
("qwen/qwen3.5-35b-a3b", ""),
("stepfun/step-3.5-flash", ""),
("z-ai/glm-5", ""),
2026-02-22 02:16:11 -08:00
("moonshotai/kimi-k2.5", ""),
("minimax/minimax-m2.5", ""),
2026-02-22 02:16:11 -08:00
]
_PROVIDER_MODELS: dict[str, list[str]] = {
"nous": [
"claude-opus-4-6",
"claude-sonnet-4-6",
"gpt-5.4",
"gemini-3-flash",
"gemini-3.0-pro-preview",
"deepseek-v3.2",
],
"openai-codex": [
2026-03-13 21:06:06 -07:00
"gpt-5.3-codex",
"gpt-5.2-codex",
"gpt-5.1-codex-mini",
"gpt-5.1-codex-max",
],
"zai": [
"glm-5",
"glm-4.7",
"glm-4.5",
"glm-4.5-flash",
],
"kimi-coding": [
fix: improve Kimi model selection — auto-detect endpoint, add missing models (#1039) * fix: /reasoning command output ordering, display, and inline think extraction Three issues with the /reasoning command: 1. Output interleaving: The command echo used print() while feedback used _cprint(), causing them to render out-of-order under prompt_toolkit's patch_stdout. Changed echo to use _cprint() so all output renders through the same path in correct order. 2. Reasoning display not working: /reasoning show toggled a flag but reasoning never appeared for models that embed thinking in inline <think> blocks rather than structured API fields. Added fallback extraction in _build_assistant_message to capture <think> block content as reasoning when no structured reasoning fields (reasoning, reasoning_content, reasoning_details) are present. This feeds into both the reasoning callback (during tool loops) and the post-response reasoning box display. 3. Feedback clarity: Added checkmarks to confirm actions, persisted show/hide to config (was session-only before), and aligned the status display for readability. Tests: 7 new tests for inline think block extraction (41 total). * feat: add /reasoning command to gateway (Telegram/Discord/etc) The /reasoning command only existed in the CLI — messaging platforms had no way to view or change reasoning settings. This adds: 1. /reasoning command handler in the gateway: - No args: shows current effort level and display state - /reasoning <level>: sets reasoning effort (none/low/medium/high/xhigh) - /reasoning show|hide: toggles reasoning display in responses - All changes saved to config.yaml immediately 2. Reasoning display in gateway responses: - When show_reasoning is enabled, prepends a 'Reasoning' block with the model's last_reasoning content before the response - Collapses long reasoning (>15 lines) to keep messages readable - Uses last_reasoning from run_conversation result dict 3. Plumbing: - Added _show_reasoning attribute loaded from config at startup - Propagated last_reasoning through _run_agent return dict - Added /reasoning to help text and known_commands set - Uses getattr for _show_reasoning to handle test stubs * fix: improve Kimi model selection — auto-detect endpoint, add missing models Kimi Coding Plan setup: - New dedicated _model_flow_kimi() replaces the generic API-key flow for kimi-coding. Removes the confusing 'Base URL' prompt entirely — the endpoint is auto-detected from the API key prefix: sk-kimi-* → api.kimi.com/coding/v1 (Kimi Coding Plan) other → api.moonshot.ai/v1 (legacy Moonshot) - Shows appropriate models for each endpoint: Coding Plan: kimi-for-coding, kimi-k2.5, kimi-k2-thinking, kimi-k2-thinking-turbo Moonshot: full model catalog - Clears any stale KIMI_BASE_URL override so runtime auto-detection via _resolve_kimi_base_url() works correctly. Model catalog updates: - Added kimi-for-coding (primary Coding Plan model) and kimi-k2-thinking-turbo to models.py, main.py _PROVIDER_MODELS, and model_metadata.py context windows. - Updated User-Agent from KimiCLI/1.0 to KimiCLI/1.3 (Kimi's coding endpoint whitelists known coding agents via User-Agent sniffing).
2026-03-12 05:58:48 -07:00
"kimi-for-coding",
"kimi-k2.5",
"kimi-k2-thinking",
fix: improve Kimi model selection — auto-detect endpoint, add missing models (#1039) * fix: /reasoning command output ordering, display, and inline think extraction Three issues with the /reasoning command: 1. Output interleaving: The command echo used print() while feedback used _cprint(), causing them to render out-of-order under prompt_toolkit's patch_stdout. Changed echo to use _cprint() so all output renders through the same path in correct order. 2. Reasoning display not working: /reasoning show toggled a flag but reasoning never appeared for models that embed thinking in inline <think> blocks rather than structured API fields. Added fallback extraction in _build_assistant_message to capture <think> block content as reasoning when no structured reasoning fields (reasoning, reasoning_content, reasoning_details) are present. This feeds into both the reasoning callback (during tool loops) and the post-response reasoning box display. 3. Feedback clarity: Added checkmarks to confirm actions, persisted show/hide to config (was session-only before), and aligned the status display for readability. Tests: 7 new tests for inline think block extraction (41 total). * feat: add /reasoning command to gateway (Telegram/Discord/etc) The /reasoning command only existed in the CLI — messaging platforms had no way to view or change reasoning settings. This adds: 1. /reasoning command handler in the gateway: - No args: shows current effort level and display state - /reasoning <level>: sets reasoning effort (none/low/medium/high/xhigh) - /reasoning show|hide: toggles reasoning display in responses - All changes saved to config.yaml immediately 2. Reasoning display in gateway responses: - When show_reasoning is enabled, prepends a 'Reasoning' block with the model's last_reasoning content before the response - Collapses long reasoning (>15 lines) to keep messages readable - Uses last_reasoning from run_conversation result dict 3. Plumbing: - Added _show_reasoning attribute loaded from config at startup - Propagated last_reasoning through _run_agent return dict - Added /reasoning to help text and known_commands set - Uses getattr for _show_reasoning to handle test stubs * fix: improve Kimi model selection — auto-detect endpoint, add missing models Kimi Coding Plan setup: - New dedicated _model_flow_kimi() replaces the generic API-key flow for kimi-coding. Removes the confusing 'Base URL' prompt entirely — the endpoint is auto-detected from the API key prefix: sk-kimi-* → api.kimi.com/coding/v1 (Kimi Coding Plan) other → api.moonshot.ai/v1 (legacy Moonshot) - Shows appropriate models for each endpoint: Coding Plan: kimi-for-coding, kimi-k2.5, kimi-k2-thinking, kimi-k2-thinking-turbo Moonshot: full model catalog - Clears any stale KIMI_BASE_URL override so runtime auto-detection via _resolve_kimi_base_url() works correctly. Model catalog updates: - Added kimi-for-coding (primary Coding Plan model) and kimi-k2-thinking-turbo to models.py, main.py _PROVIDER_MODELS, and model_metadata.py context windows. - Updated User-Agent from KimiCLI/1.0 to KimiCLI/1.3 (Kimi's coding endpoint whitelists known coding agents via User-Agent sniffing).
2026-03-12 05:58:48 -07:00
"kimi-k2-thinking-turbo",
"kimi-k2-turbo-preview",
"kimi-k2-0905-preview",
],
"minimax": [
"MiniMax-M2.5",
"MiniMax-M2.5-highspeed",
"MiniMax-M2.1",
],
"minimax-cn": [
"MiniMax-M2.5",
"MiniMax-M2.5-highspeed",
"MiniMax-M2.1",
],
feat: native Anthropic provider with Claude Code credential auto-discovery Add Anthropic as a first-class inference provider, bypassing OpenRouter for direct API access. Uses the native Anthropic SDK with a full format adapter (same pattern as the codex_responses api_mode). ## Auth (three methods, priority order) 1. ANTHROPIC_API_KEY env var (regular API key, sk-ant-api-*) 2. ANTHROPIC_TOKEN / CLAUDE_CODE_OAUTH_TOKEN env var (setup-token, sk-ant-oat-*) 3. Auto-discovery from ~/.claude/.credentials.json (Claude Code subscription) - Reads Claude Code's OAuth credentials - Checks token expiry with 60s buffer - Setup tokens use Bearer auth + anthropic-beta: oauth-2025-04-20 header - Regular API keys use standard x-api-key header ## Changes by file ### New files - agent/anthropic_adapter.py — Client builder, message/tool/response format conversion, Claude Code credential reader, token resolver. Handles system prompt extraction, tool_use/tool_result blocks, thinking/reasoning, orphaned tool_use cleanup, cache_control. - tests/test_anthropic_adapter.py — 36 tests covering all adapter logic ### Modified files - pyproject.toml — Add anthropic>=0.39.0 dependency - hermes_cli/auth.py — Add 'anthropic' to PROVIDER_REGISTRY with three env vars, plus 'claude'/'claude-code' aliases - hermes_cli/models.py — Add model catalog, labels, aliases, provider order - hermes_cli/main.py — Add 'anthropic' to --provider CLI choices - hermes_cli/runtime_provider.py — Add Anthropic branch returning api_mode='anthropic_messages' (before generic api_key fallthrough) - hermes_cli/setup.py — Add Anthropic setup wizard with Claude Code credential auto-discovery, model selection, OpenRouter tools prompt - agent/auxiliary_client.py — Add claude-haiku-4-5 as aux model - agent/model_metadata.py — Add bare Claude model context lengths - run_agent.py — Add anthropic_messages api_mode: * Client init (Anthropic SDK instead of OpenAI) * API call dispatch (_anthropic_client.messages.create) * Response validation (content blocks) * finish_reason mapping (stop_reason -> finish_reason) * Token usage (input_tokens/output_tokens) * Response normalization (normalize_anthropic_response) * Client interrupt/rebuild * Prompt caching auto-enabled for native Anthropic - tests/test_run_agent.py — Update test_anthropic_base_url_accepted to expect native routing, add test_prompt_caching_native_anthropic
2026-03-12 15:47:45 -07:00
"anthropic": [
"claude-opus-4-6",
"claude-sonnet-4-6",
"claude-opus-4-5-20251101",
"claude-sonnet-4-5-20250929",
feat: native Anthropic provider with Claude Code credential auto-discovery Add Anthropic as a first-class inference provider, bypassing OpenRouter for direct API access. Uses the native Anthropic SDK with a full format adapter (same pattern as the codex_responses api_mode). ## Auth (three methods, priority order) 1. ANTHROPIC_API_KEY env var (regular API key, sk-ant-api-*) 2. ANTHROPIC_TOKEN / CLAUDE_CODE_OAUTH_TOKEN env var (setup-token, sk-ant-oat-*) 3. Auto-discovery from ~/.claude/.credentials.json (Claude Code subscription) - Reads Claude Code's OAuth credentials - Checks token expiry with 60s buffer - Setup tokens use Bearer auth + anthropic-beta: oauth-2025-04-20 header - Regular API keys use standard x-api-key header ## Changes by file ### New files - agent/anthropic_adapter.py — Client builder, message/tool/response format conversion, Claude Code credential reader, token resolver. Handles system prompt extraction, tool_use/tool_result blocks, thinking/reasoning, orphaned tool_use cleanup, cache_control. - tests/test_anthropic_adapter.py — 36 tests covering all adapter logic ### Modified files - pyproject.toml — Add anthropic>=0.39.0 dependency - hermes_cli/auth.py — Add 'anthropic' to PROVIDER_REGISTRY with three env vars, plus 'claude'/'claude-code' aliases - hermes_cli/models.py — Add model catalog, labels, aliases, provider order - hermes_cli/main.py — Add 'anthropic' to --provider CLI choices - hermes_cli/runtime_provider.py — Add Anthropic branch returning api_mode='anthropic_messages' (before generic api_key fallthrough) - hermes_cli/setup.py — Add Anthropic setup wizard with Claude Code credential auto-discovery, model selection, OpenRouter tools prompt - agent/auxiliary_client.py — Add claude-haiku-4-5 as aux model - agent/model_metadata.py — Add bare Claude model context lengths - run_agent.py — Add anthropic_messages api_mode: * Client init (Anthropic SDK instead of OpenAI) * API call dispatch (_anthropic_client.messages.create) * Response validation (content blocks) * finish_reason mapping (stop_reason -> finish_reason) * Token usage (input_tokens/output_tokens) * Response normalization (normalize_anthropic_response) * Client interrupt/rebuild * Prompt caching auto-enabled for native Anthropic - tests/test_run_agent.py — Update test_anthropic_base_url_accepted to expect native routing, add test_prompt_caching_native_anthropic
2026-03-12 15:47:45 -07:00
"claude-opus-4-20250514",
"claude-sonnet-4-20250514",
feat: native Anthropic provider with Claude Code credential auto-discovery Add Anthropic as a first-class inference provider, bypassing OpenRouter for direct API access. Uses the native Anthropic SDK with a full format adapter (same pattern as the codex_responses api_mode). ## Auth (three methods, priority order) 1. ANTHROPIC_API_KEY env var (regular API key, sk-ant-api-*) 2. ANTHROPIC_TOKEN / CLAUDE_CODE_OAUTH_TOKEN env var (setup-token, sk-ant-oat-*) 3. Auto-discovery from ~/.claude/.credentials.json (Claude Code subscription) - Reads Claude Code's OAuth credentials - Checks token expiry with 60s buffer - Setup tokens use Bearer auth + anthropic-beta: oauth-2025-04-20 header - Regular API keys use standard x-api-key header ## Changes by file ### New files - agent/anthropic_adapter.py — Client builder, message/tool/response format conversion, Claude Code credential reader, token resolver. Handles system prompt extraction, tool_use/tool_result blocks, thinking/reasoning, orphaned tool_use cleanup, cache_control. - tests/test_anthropic_adapter.py — 36 tests covering all adapter logic ### Modified files - pyproject.toml — Add anthropic>=0.39.0 dependency - hermes_cli/auth.py — Add 'anthropic' to PROVIDER_REGISTRY with three env vars, plus 'claude'/'claude-code' aliases - hermes_cli/models.py — Add model catalog, labels, aliases, provider order - hermes_cli/main.py — Add 'anthropic' to --provider CLI choices - hermes_cli/runtime_provider.py — Add Anthropic branch returning api_mode='anthropic_messages' (before generic api_key fallthrough) - hermes_cli/setup.py — Add Anthropic setup wizard with Claude Code credential auto-discovery, model selection, OpenRouter tools prompt - agent/auxiliary_client.py — Add claude-haiku-4-5 as aux model - agent/model_metadata.py — Add bare Claude model context lengths - run_agent.py — Add anthropic_messages api_mode: * Client init (Anthropic SDK instead of OpenAI) * API call dispatch (_anthropic_client.messages.create) * Response validation (content blocks) * finish_reason mapping (stop_reason -> finish_reason) * Token usage (input_tokens/output_tokens) * Response normalization (normalize_anthropic_response) * Client interrupt/rebuild * Prompt caching auto-enabled for native Anthropic - tests/test_run_agent.py — Update test_anthropic_base_url_accepted to expect native routing, add test_prompt_caching_native_anthropic
2026-03-12 15:47:45 -07:00
"claude-haiku-4-5-20251001",
],
"deepseek": [
"deepseek-chat",
"deepseek-reasoner",
],
"ai-gateway": [
"anthropic/claude-opus-4.6",
"anthropic/claude-sonnet-4.6",
"anthropic/claude-sonnet-4.5",
"anthropic/claude-haiku-4.5",
"openai/gpt-5",
"openai/gpt-4.1",
"openai/gpt-4.1-mini",
"google/gemini-3-pro-preview",
"google/gemini-3-flash",
"google/gemini-2.5-pro",
"google/gemini-2.5-flash",
"deepseek/deepseek-v3.2",
],
}
_PROVIDER_LABELS = {
"openrouter": "OpenRouter",
"openai-codex": "OpenAI Codex",
"nous": "Nous Portal",
"zai": "Z.AI / GLM",
"kimi-coding": "Kimi / Moonshot",
"minimax": "MiniMax",
"minimax-cn": "MiniMax (China)",
feat: native Anthropic provider with Claude Code credential auto-discovery Add Anthropic as a first-class inference provider, bypassing OpenRouter for direct API access. Uses the native Anthropic SDK with a full format adapter (same pattern as the codex_responses api_mode). ## Auth (three methods, priority order) 1. ANTHROPIC_API_KEY env var (regular API key, sk-ant-api-*) 2. ANTHROPIC_TOKEN / CLAUDE_CODE_OAUTH_TOKEN env var (setup-token, sk-ant-oat-*) 3. Auto-discovery from ~/.claude/.credentials.json (Claude Code subscription) - Reads Claude Code's OAuth credentials - Checks token expiry with 60s buffer - Setup tokens use Bearer auth + anthropic-beta: oauth-2025-04-20 header - Regular API keys use standard x-api-key header ## Changes by file ### New files - agent/anthropic_adapter.py — Client builder, message/tool/response format conversion, Claude Code credential reader, token resolver. Handles system prompt extraction, tool_use/tool_result blocks, thinking/reasoning, orphaned tool_use cleanup, cache_control. - tests/test_anthropic_adapter.py — 36 tests covering all adapter logic ### Modified files - pyproject.toml — Add anthropic>=0.39.0 dependency - hermes_cli/auth.py — Add 'anthropic' to PROVIDER_REGISTRY with three env vars, plus 'claude'/'claude-code' aliases - hermes_cli/models.py — Add model catalog, labels, aliases, provider order - hermes_cli/main.py — Add 'anthropic' to --provider CLI choices - hermes_cli/runtime_provider.py — Add Anthropic branch returning api_mode='anthropic_messages' (before generic api_key fallthrough) - hermes_cli/setup.py — Add Anthropic setup wizard with Claude Code credential auto-discovery, model selection, OpenRouter tools prompt - agent/auxiliary_client.py — Add claude-haiku-4-5 as aux model - agent/model_metadata.py — Add bare Claude model context lengths - run_agent.py — Add anthropic_messages api_mode: * Client init (Anthropic SDK instead of OpenAI) * API call dispatch (_anthropic_client.messages.create) * Response validation (content blocks) * finish_reason mapping (stop_reason -> finish_reason) * Token usage (input_tokens/output_tokens) * Response normalization (normalize_anthropic_response) * Client interrupt/rebuild * Prompt caching auto-enabled for native Anthropic - tests/test_run_agent.py — Update test_anthropic_base_url_accepted to expect native routing, add test_prompt_caching_native_anthropic
2026-03-12 15:47:45 -07:00
"anthropic": "Anthropic",
"deepseek": "DeepSeek",
"ai-gateway": "AI Gateway",
"custom": "Custom endpoint",
}
_PROVIDER_ALIASES = {
"glm": "zai",
"z-ai": "zai",
"z.ai": "zai",
"zhipu": "zai",
"kimi": "kimi-coding",
"moonshot": "kimi-coding",
"minimax-china": "minimax-cn",
"minimax_cn": "minimax-cn",
feat: native Anthropic provider with Claude Code credential auto-discovery Add Anthropic as a first-class inference provider, bypassing OpenRouter for direct API access. Uses the native Anthropic SDK with a full format adapter (same pattern as the codex_responses api_mode). ## Auth (three methods, priority order) 1. ANTHROPIC_API_KEY env var (regular API key, sk-ant-api-*) 2. ANTHROPIC_TOKEN / CLAUDE_CODE_OAUTH_TOKEN env var (setup-token, sk-ant-oat-*) 3. Auto-discovery from ~/.claude/.credentials.json (Claude Code subscription) - Reads Claude Code's OAuth credentials - Checks token expiry with 60s buffer - Setup tokens use Bearer auth + anthropic-beta: oauth-2025-04-20 header - Regular API keys use standard x-api-key header ## Changes by file ### New files - agent/anthropic_adapter.py — Client builder, message/tool/response format conversion, Claude Code credential reader, token resolver. Handles system prompt extraction, tool_use/tool_result blocks, thinking/reasoning, orphaned tool_use cleanup, cache_control. - tests/test_anthropic_adapter.py — 36 tests covering all adapter logic ### Modified files - pyproject.toml — Add anthropic>=0.39.0 dependency - hermes_cli/auth.py — Add 'anthropic' to PROVIDER_REGISTRY with three env vars, plus 'claude'/'claude-code' aliases - hermes_cli/models.py — Add model catalog, labels, aliases, provider order - hermes_cli/main.py — Add 'anthropic' to --provider CLI choices - hermes_cli/runtime_provider.py — Add Anthropic branch returning api_mode='anthropic_messages' (before generic api_key fallthrough) - hermes_cli/setup.py — Add Anthropic setup wizard with Claude Code credential auto-discovery, model selection, OpenRouter tools prompt - agent/auxiliary_client.py — Add claude-haiku-4-5 as aux model - agent/model_metadata.py — Add bare Claude model context lengths - run_agent.py — Add anthropic_messages api_mode: * Client init (Anthropic SDK instead of OpenAI) * API call dispatch (_anthropic_client.messages.create) * Response validation (content blocks) * finish_reason mapping (stop_reason -> finish_reason) * Token usage (input_tokens/output_tokens) * Response normalization (normalize_anthropic_response) * Client interrupt/rebuild * Prompt caching auto-enabled for native Anthropic - tests/test_run_agent.py — Update test_anthropic_base_url_accepted to expect native routing, add test_prompt_caching_native_anthropic
2026-03-12 15:47:45 -07:00
"claude": "anthropic",
"claude-code": "anthropic",
"deep-seek": "deepseek",
"aigateway": "ai-gateway",
"vercel": "ai-gateway",
"vercel-ai-gateway": "ai-gateway",
}
2026-02-22 02:16:11 -08:00
def model_ids() -> list[str]:
"""Return just the OpenRouter model-id strings."""
2026-02-22 02:16:11 -08:00
return [mid for mid, _ in OPENROUTER_MODELS]
def menu_labels() -> list[str]:
"""Return display labels like 'anthropic/claude-opus-4.6 (recommended)'."""
labels = []
for mid, desc in OPENROUTER_MODELS:
labels.append(f"{mid} ({desc})" if desc else mid)
return labels
# All provider IDs and aliases that are valid for the provider:model syntax.
_KNOWN_PROVIDER_NAMES: set[str] = (
set(_PROVIDER_LABELS.keys())
| set(_PROVIDER_ALIASES.keys())
| {"openrouter", "custom"}
)
def list_available_providers() -> list[dict[str, str]]:
"""Return info about all providers the user could use with ``provider:model``.
Each dict has ``id``, ``label``, and ``aliases``.
Checks which providers have valid credentials configured.
"""
# Canonical providers in display order
_PROVIDER_ORDER = [
"openrouter", "nous", "openai-codex",
"zai", "kimi-coding", "minimax", "minimax-cn", "anthropic",
"ai-gateway", "deepseek",
]
# Build reverse alias map
aliases_for: dict[str, list[str]] = {}
for alias, canonical in _PROVIDER_ALIASES.items():
aliases_for.setdefault(canonical, []).append(alias)
result = []
for pid in _PROVIDER_ORDER:
label = _PROVIDER_LABELS.get(pid, pid)
alias_list = aliases_for.get(pid, [])
# Check if this provider has credentials available
has_creds = False
try:
from hermes_cli.runtime_provider import resolve_runtime_provider
runtime = resolve_runtime_provider(requested=pid)
has_creds = bool(runtime.get("api_key"))
except Exception:
pass
result.append({
"id": pid,
"label": label,
"aliases": alias_list,
"authenticated": has_creds,
})
return result
def parse_model_input(raw: str, current_provider: str) -> tuple[str, str]:
"""Parse ``/model`` input into ``(provider, model)``.
Supports ``provider:model`` syntax to switch providers at runtime::
openrouter:anthropic/claude-sonnet-4.5 ("openrouter", "anthropic/claude-sonnet-4.5")
nous:hermes-3 ("nous", "hermes-3")
anthropic/claude-sonnet-4.5 (current_provider, "anthropic/claude-sonnet-4.5")
gpt-5.4 (current_provider, "gpt-5.4")
The colon is only treated as a provider delimiter if the left side is a
recognized provider name or alias. This avoids misinterpreting model names
that happen to contain colons (e.g. ``anthropic/claude-3.5-sonnet:beta``).
Returns ``(provider, model)`` where *provider* is either the explicit
provider from the input or *current_provider* if none was specified.
"""
stripped = raw.strip()
colon = stripped.find(":")
if colon > 0:
provider_part = stripped[:colon].strip().lower()
model_part = stripped[colon + 1:].strip()
if provider_part and model_part and provider_part in _KNOWN_PROVIDER_NAMES:
return (normalize_provider(provider_part), model_part)
return (current_provider, stripped)
def curated_models_for_provider(provider: Optional[str]) -> list[tuple[str, str]]:
"""Return ``(model_id, description)`` tuples for a provider's model list.
Tries to fetch the live model list from the provider's API first,
falling back to the static ``_PROVIDER_MODELS`` catalog if the API
is unreachable.
"""
normalized = normalize_provider(provider)
if normalized == "openrouter":
return list(OPENROUTER_MODELS)
# Try live API first (Codex, Nous, etc. all support /models)
live = provider_model_ids(normalized)
if live:
return [(m, "") for m in live]
# Fallback to static catalog
models = _PROVIDER_MODELS.get(normalized, [])
return [(m, "") for m in models]
def detect_provider_for_model(
model_name: str,
current_provider: str,
) -> Optional[tuple[str, str]]:
"""Auto-detect the best provider for a model name.
Returns ``(provider_id, model_name)`` the model name may be remapped
(e.g. bare ``deepseek-chat`` ``deepseek/deepseek-chat`` for OpenRouter).
Returns ``None`` when no confident match is found.
Priority:
1. Direct provider with credentials (highest)
2. Direct provider without credentials remap to OpenRouter slug
3. OpenRouter catalog match
"""
name = (model_name or "").strip()
if not name:
return None
name_lower = name.lower()
# Aggregators list other providers' models — never auto-switch TO them
_AGGREGATORS = {"nous", "openrouter"}
# If the model belongs to the current provider's catalog, don't suggest switching
current_models = _PROVIDER_MODELS.get(current_provider, [])
if any(name_lower == m.lower() for m in current_models):
return None
# --- Step 1: check static provider catalogs for a direct match ---
direct_match: Optional[str] = None
for pid, models in _PROVIDER_MODELS.items():
if pid == current_provider or pid in _AGGREGATORS:
continue
if any(name_lower == m.lower() for m in models):
direct_match = pid
break
if direct_match:
# Check if we have credentials for this provider
has_creds = False
try:
from hermes_cli.auth import PROVIDER_REGISTRY
pconfig = PROVIDER_REGISTRY.get(direct_match)
if pconfig:
import os
for env_var in pconfig.api_key_env_vars:
if os.getenv(env_var, "").strip():
has_creds = True
break
except Exception:
pass
if has_creds:
return (direct_match, name)
# No direct creds — try to find this model on OpenRouter instead
or_slug = _find_openrouter_slug(name)
if or_slug:
return ("openrouter", or_slug)
# Still return the direct provider — credential resolution will
# give a clear error rather than silently using the wrong provider
return (direct_match, name)
# --- Step 2: check OpenRouter catalog ---
# First try exact match (handles provider/model format)
or_slug = _find_openrouter_slug(name)
if or_slug:
if current_provider != "openrouter":
return ("openrouter", or_slug)
# Already on openrouter, just return the resolved slug
if or_slug != name:
return ("openrouter", or_slug)
return None # already on openrouter with matching name
return None
def _find_openrouter_slug(model_name: str) -> Optional[str]:
"""Find the full OpenRouter model slug for a bare or partial model name.
Handles:
- Exact match: ``anthropic/claude-opus-4.6`` as-is
- Bare name: ``deepseek-chat`` ``deepseek/deepseek-chat``
- Bare name: ``claude-opus-4.6`` ``anthropic/claude-opus-4.6``
"""
name_lower = model_name.strip().lower()
if not name_lower:
return None
# Exact match (already has provider/ prefix)
for mid, _ in OPENROUTER_MODELS:
if name_lower == mid.lower():
return mid
# Try matching just the model part (after the /)
for mid, _ in OPENROUTER_MODELS:
if "/" in mid:
_, model_part = mid.split("/", 1)
if name_lower == model_part.lower():
return mid
return None
def normalize_provider(provider: Optional[str]) -> str:
"""Normalize provider aliases to Hermes' canonical provider ids.
Note: ``"auto"`` passes through unchanged use
``hermes_cli.auth.resolve_provider()`` to resolve it to a concrete
provider based on credentials and environment.
"""
normalized = (provider or "openrouter").strip().lower()
return _PROVIDER_ALIASES.get(normalized, normalized)
def provider_label(provider: Optional[str]) -> str:
"""Return a human-friendly label for a provider id or alias."""
original = (provider or "openrouter").strip()
normalized = original.lower()
if normalized == "auto":
return "Auto"
normalized = normalize_provider(normalized)
return _PROVIDER_LABELS.get(normalized, original or "OpenRouter")
def provider_model_ids(provider: Optional[str]) -> list[str]:
"""Return the best known model catalog for a provider.
Tries live API endpoints for providers that support them (Codex, Nous),
falling back to static lists.
"""
normalized = normalize_provider(provider)
if normalized == "openrouter":
return model_ids()
if normalized == "openai-codex":
from hermes_cli.codex_models import get_codex_model_ids
return get_codex_model_ids()
if normalized == "nous":
# Try live Nous Portal /models endpoint
try:
from hermes_cli.auth import fetch_nous_models, resolve_nous_runtime_credentials
creds = resolve_nous_runtime_credentials()
if creds:
live = fetch_nous_models(creds.get("api_key", ""), creds.get("base_url", ""))
if live:
return live
except Exception:
pass
if normalized == "anthropic":
live = _fetch_anthropic_models()
if live:
return live
if normalized == "ai-gateway":
live = _fetch_ai_gateway_models()
if live:
return live
return list(_PROVIDER_MODELS.get(normalized, []))
def _fetch_anthropic_models(timeout: float = 5.0) -> Optional[list[str]]:
"""Fetch available models from the Anthropic /v1/models endpoint.
Uses resolve_anthropic_token() to find credentials (env vars or
Claude Code auto-discovery). Returns sorted model IDs or None.
"""
try:
from agent.anthropic_adapter import resolve_anthropic_token, _is_oauth_token
except ImportError:
return None
token = resolve_anthropic_token()
if not token:
return None
headers: dict[str, str] = {"anthropic-version": "2023-06-01"}
if _is_oauth_token(token):
headers["Authorization"] = f"Bearer {token}"
fix: Anthropic OAuth — beta header, token refresh, config contamination, reauthentication (#1132) Fixes Anthropic OAuth/subscription authentication end-to-end: Auth failures (401 errors): - Add missing 'claude-code-20250219' beta header for OAuth tokens. Both clawdbot and OpenCode include this alongside 'oauth-2025-04-20' — without it, Anthropic's API rejects OAuth tokens with 401 authentication errors. - Fix _fetch_anthropic_models() to use canonical beta headers from _COMMON_BETAS + _OAUTH_ONLY_BETAS instead of hardcoding. Token refresh: - Add _refresh_oauth_token() — when Claude Code credentials from ~/.claude/.credentials.json are expired but have a refresh token, automatically POST to console.anthropic.com/v1/oauth/token to get a new access token. Uses the same client_id as Claude Code / OpenCode. - Add _write_claude_code_credentials() — writes refreshed tokens back to ~/.claude/.credentials.json, preserving other fields. - resolve_anthropic_token() now auto-refreshes expired tokens before returning None. Config contamination: - Anthropic's _model_flow_anthropic() no longer saves base_url to config. Since resolve_runtime_provider() always hardcodes Anthropic's URL, the stale base_url was contaminating other providers when users switched without re-running 'hermes model' (e.g., Codex hitting api.anthropic.com). - _update_config_for_provider() now pops base_url when passed empty string. - Same fix in setup.py. Flow/UX (hermes model command): - CLAUDE_CODE_OAUTH_TOKEN env var now checked in credential detection - Reauthentication option when existing credentials found - run_oauth_setup_token() runs 'claude setup-token' as interactive subprocess, then auto-detects saved credentials - Clean has_creds/needs_auth flow in both main.py and setup.py Tests (14 new): - Beta header assertions for claude-code-20250219 - Token refresh: successful refresh with credential writeback, failed refresh returns None, no refresh token returns None - Credential writeback: new file creation, preserving existing fields - Auto-refresh integration in resolve_anthropic_token() - CLAUDE_CODE_OAUTH_TOKEN fallback, credential file auto-discovery - run_oauth_setup_token() (5 scenarios)
2026-03-12 20:45:50 -07:00
from agent.anthropic_adapter import _COMMON_BETAS, _OAUTH_ONLY_BETAS
headers["anthropic-beta"] = ",".join(_COMMON_BETAS + _OAUTH_ONLY_BETAS)
else:
headers["x-api-key"] = token
req = urllib.request.Request(
"https://api.anthropic.com/v1/models",
headers=headers,
)
try:
with urllib.request.urlopen(req, timeout=timeout) as resp:
data = json.loads(resp.read().decode())
models = [m["id"] for m in data.get("data", []) if m.get("id")]
# Sort: latest/largest first (opus > sonnet > haiku, higher version first)
return sorted(models, key=lambda m: (
"opus" not in m, # opus first
"sonnet" not in m, # then sonnet
"haiku" not in m, # then haiku
m, # alphabetical within tier
))
except Exception as e:
import logging
logging.getLogger(__name__).debug("Failed to fetch Anthropic models: %s", e)
return None
def probe_api_models(
api_key: Optional[str],
base_url: Optional[str],
timeout: float = 5.0,
) -> dict[str, Any]:
"""Probe an OpenAI-compatible ``/models`` endpoint with light URL heuristics."""
normalized = (base_url or "").strip().rstrip("/")
if not normalized:
return {
"models": None,
"probed_url": None,
"resolved_base_url": "",
"suggested_base_url": None,
"used_fallback": False,
}
if normalized.endswith("/v1"):
alternate_base = normalized[:-3].rstrip("/")
else:
alternate_base = normalized + "/v1"
candidates: list[tuple[str, bool]] = [(normalized, False)]
if alternate_base and alternate_base != normalized:
candidates.append((alternate_base, True))
tried: list[str] = []
headers: dict[str, str] = {}
if api_key:
headers["Authorization"] = f"Bearer {api_key}"
for candidate_base, is_fallback in candidates:
url = candidate_base.rstrip("/") + "/models"
tried.append(url)
req = urllib.request.Request(url, headers=headers)
try:
with urllib.request.urlopen(req, timeout=timeout) as resp:
data = json.loads(resp.read().decode())
return {
"models": [m.get("id", "") for m in data.get("data", [])],
"probed_url": url,
"resolved_base_url": candidate_base.rstrip("/"),
"suggested_base_url": alternate_base if alternate_base != candidate_base else normalized,
"used_fallback": is_fallback,
}
except Exception:
continue
return {
"models": None,
"probed_url": tried[-1] if tried else normalized.rstrip("/") + "/models",
"resolved_base_url": normalized,
"suggested_base_url": alternate_base if alternate_base != normalized else None,
"used_fallback": False,
}
def _fetch_ai_gateway_models(timeout: float = 5.0) -> Optional[list[str]]:
"""Fetch available language models with tool-use from AI Gateway."""
api_key = os.getenv("AI_GATEWAY_API_KEY", "").strip()
if not api_key:
return None
base_url = os.getenv("AI_GATEWAY_BASE_URL", "").strip()
if not base_url:
from hermes_constants import AI_GATEWAY_BASE_URL
base_url = AI_GATEWAY_BASE_URL
url = base_url.rstrip("/") + "/models"
headers: dict[str, str] = {"Authorization": f"Bearer {api_key}"}
req = urllib.request.Request(url, headers=headers)
try:
with urllib.request.urlopen(req, timeout=timeout) as resp:
data = json.loads(resp.read().decode())
return [
m["id"]
for m in data.get("data", [])
if m.get("id")
and m.get("type") == "language"
and "tool-use" in (m.get("tags") or [])
]
except Exception:
return None
def fetch_api_models(
api_key: Optional[str],
base_url: Optional[str],
timeout: float = 5.0,
) -> Optional[list[str]]:
"""Fetch the list of available model IDs from the provider's ``/models`` endpoint.
Returns a list of model ID strings, or ``None`` if the endpoint could not
be reached (network error, timeout, auth failure, etc.).
"""
return probe_api_models(api_key, base_url, timeout=timeout).get("models")
def validate_requested_model(
model_name: str,
provider: Optional[str],
*,
api_key: Optional[str] = None,
base_url: Optional[str] = None,
) -> dict[str, Any]:
"""
Validate a ``/model`` value for the active provider.
Performs format checks first, then probes the live API to confirm
the model actually exists.
Returns a dict with:
- accepted: whether the CLI should switch to the requested model now
- persist: whether it is safe to save to config
- recognized: whether it matched a known provider catalog
- message: optional warning / guidance for the user
"""
requested = (model_name or "").strip()
normalized = normalize_provider(provider)
if normalized == "openrouter" and base_url and "openrouter.ai" not in base_url:
normalized = "custom"
if not requested:
return {
"accepted": False,
"persist": False,
"recognized": False,
"message": "Model name cannot be empty.",
}
if any(ch.isspace() for ch in requested):
return {
"accepted": False,
"persist": False,
"recognized": False,
"message": "Model names cannot contain spaces.",
}
if normalized == "custom":
probe = probe_api_models(api_key, base_url)
api_models = probe.get("models")
if api_models is not None:
if requested in set(api_models):
return {
"accepted": True,
"persist": True,
"recognized": True,
"message": None,
}
suggestions = get_close_matches(requested, api_models, n=3, cutoff=0.5)
suggestion_text = ""
if suggestions:
suggestion_text = "\n Similar models: " + ", ".join(f"`{s}`" for s in suggestions)
message = (
f"Note: `{requested}` was not found in this custom endpoint's model listing "
f"({probe.get('probed_url')}). It may still work if the server supports hidden or aliased models."
f"{suggestion_text}"
)
if probe.get("used_fallback"):
message += (
f"\n Endpoint verification succeeded after trying `{probe.get('resolved_base_url')}`. "
f"Consider saving that as your base URL."
)
return {
"accepted": True,
"persist": True,
"recognized": False,
"message": message,
}
message = (
f"Note: could not reach this custom endpoint's model listing at `{probe.get('probed_url')}`. "
f"Hermes will still save `{requested}`, but the endpoint should expose `/models` for verification."
)
if probe.get("suggested_base_url"):
message += f"\n If this server expects `/v1`, try base URL: `{probe.get('suggested_base_url')}`"
return {
"accepted": True,
"persist": True,
"recognized": False,
"message": message,
}
# Probe the live API to check if the model actually exists
api_models = fetch_api_models(api_key, base_url)
if api_models is not None:
if requested in set(api_models):
# API confirmed the model exists
return {
"accepted": True,
"persist": True,
"recognized": True,
"message": None,
}
else:
fix: stop rejecting unlisted models, accept with warning instead * fix: use session_key instead of chat_id for adapter interrupt lookups monitor_for_interrupt() in _run_agent was using source.chat_id to query the adapter's has_pending_interrupt() and get_pending_message() methods. But the adapter stores interrupt events under build_session_key(source), which produces a different string (e.g. 'agent:main:telegram:dm' vs '123456'). This key mismatch meant the interrupt was never detected through the adapter path, which is the only active interrupt path for all adapter-based platforms (Telegram, Discord, Slack, etc.). The gateway-level interrupt path (in dispatch_message) is unreachable because the adapter intercepts the 2nd message in handle_message() before it reaches dispatch_message(). Result: sending a new message while subagents were running had no effect — the interrupt was silently lost. Fix: replace all source.chat_id references in the interrupt-related code within _run_agent() with the session_key parameter, which matches the adapter's storage keys. Also adds regression tests verifying session_key vs chat_id consistency. * debug: add file-based logging to CLI interrupt path Temporary instrumentation to diagnose why message-based interrupts don't seem to work during subagent execution. Logs to ~/.hermes/interrupt_debug.log (immune to redirect_stdout). Two log points: 1. When Enter handler puts message into _interrupt_queue 2. When chat() reads it and calls agent.interrupt() This will reveal whether the message reaches the queue and whether the interrupt is actually fired. * fix: accept unlisted models with warning instead of rejecting validate_requested_model() previously hard-rejected any model not found in the provider's API listing. This was too aggressive — users on higher plan tiers (e.g. Z.AI Pro/Max) may have access to models not shown in the public listing (like glm-5 on coding endpoints). Changes: - validate_requested_model: accept unlisted models with a warning note instead of blocking. The model is saved to config and used immediately. - Z.AI setup: always offer glm-5 in the model list regardless of whether a coding endpoint was detected. Pro/Max plans support it. - Z.AI setup detection message: softened from 'GLM-5 is not available' to 'GLM-5 may still be available depending on your plan tier'
2026-03-12 16:02:35 -07:00
# API responded but model is not listed. Accept anyway —
# the user may have access to models not shown in the public
# listing (e.g. Z.AI Pro/Max plans can use glm-5 on coding
# endpoints even though it's not in /models). Warn but allow.
suggestions = get_close_matches(requested, api_models, n=3, cutoff=0.5)
suggestion_text = ""
if suggestions:
fix: stop rejecting unlisted models, accept with warning instead * fix: use session_key instead of chat_id for adapter interrupt lookups monitor_for_interrupt() in _run_agent was using source.chat_id to query the adapter's has_pending_interrupt() and get_pending_message() methods. But the adapter stores interrupt events under build_session_key(source), which produces a different string (e.g. 'agent:main:telegram:dm' vs '123456'). This key mismatch meant the interrupt was never detected through the adapter path, which is the only active interrupt path for all adapter-based platforms (Telegram, Discord, Slack, etc.). The gateway-level interrupt path (in dispatch_message) is unreachable because the adapter intercepts the 2nd message in handle_message() before it reaches dispatch_message(). Result: sending a new message while subagents were running had no effect — the interrupt was silently lost. Fix: replace all source.chat_id references in the interrupt-related code within _run_agent() with the session_key parameter, which matches the adapter's storage keys. Also adds regression tests verifying session_key vs chat_id consistency. * debug: add file-based logging to CLI interrupt path Temporary instrumentation to diagnose why message-based interrupts don't seem to work during subagent execution. Logs to ~/.hermes/interrupt_debug.log (immune to redirect_stdout). Two log points: 1. When Enter handler puts message into _interrupt_queue 2. When chat() reads it and calls agent.interrupt() This will reveal whether the message reaches the queue and whether the interrupt is actually fired. * fix: accept unlisted models with warning instead of rejecting validate_requested_model() previously hard-rejected any model not found in the provider's API listing. This was too aggressive — users on higher plan tiers (e.g. Z.AI Pro/Max) may have access to models not shown in the public listing (like glm-5 on coding endpoints). Changes: - validate_requested_model: accept unlisted models with a warning note instead of blocking. The model is saved to config and used immediately. - Z.AI setup: always offer glm-5 in the model list regardless of whether a coding endpoint was detected. Pro/Max plans support it. - Z.AI setup detection message: softened from 'GLM-5 is not available' to 'GLM-5 may still be available depending on your plan tier'
2026-03-12 16:02:35 -07:00
suggestion_text = "\n Similar models: " + ", ".join(f"`{s}`" for s in suggestions)
return {
fix: stop rejecting unlisted models, accept with warning instead * fix: use session_key instead of chat_id for adapter interrupt lookups monitor_for_interrupt() in _run_agent was using source.chat_id to query the adapter's has_pending_interrupt() and get_pending_message() methods. But the adapter stores interrupt events under build_session_key(source), which produces a different string (e.g. 'agent:main:telegram:dm' vs '123456'). This key mismatch meant the interrupt was never detected through the adapter path, which is the only active interrupt path for all adapter-based platforms (Telegram, Discord, Slack, etc.). The gateway-level interrupt path (in dispatch_message) is unreachable because the adapter intercepts the 2nd message in handle_message() before it reaches dispatch_message(). Result: sending a new message while subagents were running had no effect — the interrupt was silently lost. Fix: replace all source.chat_id references in the interrupt-related code within _run_agent() with the session_key parameter, which matches the adapter's storage keys. Also adds regression tests verifying session_key vs chat_id consistency. * debug: add file-based logging to CLI interrupt path Temporary instrumentation to diagnose why message-based interrupts don't seem to work during subagent execution. Logs to ~/.hermes/interrupt_debug.log (immune to redirect_stdout). Two log points: 1. When Enter handler puts message into _interrupt_queue 2. When chat() reads it and calls agent.interrupt() This will reveal whether the message reaches the queue and whether the interrupt is actually fired. * fix: accept unlisted models with warning instead of rejecting validate_requested_model() previously hard-rejected any model not found in the provider's API listing. This was too aggressive — users on higher plan tiers (e.g. Z.AI Pro/Max) may have access to models not shown in the public listing (like glm-5 on coding endpoints). Changes: - validate_requested_model: accept unlisted models with a warning note instead of blocking. The model is saved to config and used immediately. - Z.AI setup: always offer glm-5 in the model list regardless of whether a coding endpoint was detected. Pro/Max plans support it. - Z.AI setup detection message: softened from 'GLM-5 is not available' to 'GLM-5 may still be available depending on your plan tier'
2026-03-12 16:02:35 -07:00
"accepted": True,
"persist": True,
"recognized": False,
"message": (
fix: stop rejecting unlisted models, accept with warning instead * fix: use session_key instead of chat_id for adapter interrupt lookups monitor_for_interrupt() in _run_agent was using source.chat_id to query the adapter's has_pending_interrupt() and get_pending_message() methods. But the adapter stores interrupt events under build_session_key(source), which produces a different string (e.g. 'agent:main:telegram:dm' vs '123456'). This key mismatch meant the interrupt was never detected through the adapter path, which is the only active interrupt path for all adapter-based platforms (Telegram, Discord, Slack, etc.). The gateway-level interrupt path (in dispatch_message) is unreachable because the adapter intercepts the 2nd message in handle_message() before it reaches dispatch_message(). Result: sending a new message while subagents were running had no effect — the interrupt was silently lost. Fix: replace all source.chat_id references in the interrupt-related code within _run_agent() with the session_key parameter, which matches the adapter's storage keys. Also adds regression tests verifying session_key vs chat_id consistency. * debug: add file-based logging to CLI interrupt path Temporary instrumentation to diagnose why message-based interrupts don't seem to work during subagent execution. Logs to ~/.hermes/interrupt_debug.log (immune to redirect_stdout). Two log points: 1. When Enter handler puts message into _interrupt_queue 2. When chat() reads it and calls agent.interrupt() This will reveal whether the message reaches the queue and whether the interrupt is actually fired. * fix: accept unlisted models with warning instead of rejecting validate_requested_model() previously hard-rejected any model not found in the provider's API listing. This was too aggressive — users on higher plan tiers (e.g. Z.AI Pro/Max) may have access to models not shown in the public listing (like glm-5 on coding endpoints). Changes: - validate_requested_model: accept unlisted models with a warning note instead of blocking. The model is saved to config and used immediately. - Z.AI setup: always offer glm-5 in the model list regardless of whether a coding endpoint was detected. Pro/Max plans support it. - Z.AI setup detection message: softened from 'GLM-5 is not available' to 'GLM-5 may still be available depending on your plan tier'
2026-03-12 16:02:35 -07:00
f"Note: `{requested}` was not found in this provider's model listing. "
f"It may still work if your plan supports it."
f"{suggestion_text}"
),
}
fix: stop rejecting unlisted models, accept with warning instead * fix: use session_key instead of chat_id for adapter interrupt lookups monitor_for_interrupt() in _run_agent was using source.chat_id to query the adapter's has_pending_interrupt() and get_pending_message() methods. But the adapter stores interrupt events under build_session_key(source), which produces a different string (e.g. 'agent:main:telegram:dm' vs '123456'). This key mismatch meant the interrupt was never detected through the adapter path, which is the only active interrupt path for all adapter-based platforms (Telegram, Discord, Slack, etc.). The gateway-level interrupt path (in dispatch_message) is unreachable because the adapter intercepts the 2nd message in handle_message() before it reaches dispatch_message(). Result: sending a new message while subagents were running had no effect — the interrupt was silently lost. Fix: replace all source.chat_id references in the interrupt-related code within _run_agent() with the session_key parameter, which matches the adapter's storage keys. Also adds regression tests verifying session_key vs chat_id consistency. * debug: add file-based logging to CLI interrupt path Temporary instrumentation to diagnose why message-based interrupts don't seem to work during subagent execution. Logs to ~/.hermes/interrupt_debug.log (immune to redirect_stdout). Two log points: 1. When Enter handler puts message into _interrupt_queue 2. When chat() reads it and calls agent.interrupt() This will reveal whether the message reaches the queue and whether the interrupt is actually fired. * fix: accept unlisted models with warning instead of rejecting validate_requested_model() previously hard-rejected any model not found in the provider's API listing. This was too aggressive — users on higher plan tiers (e.g. Z.AI Pro/Max) may have access to models not shown in the public listing (like glm-5 on coding endpoints). Changes: - validate_requested_model: accept unlisted models with a warning note instead of blocking. The model is saved to config and used immediately. - Z.AI setup: always offer glm-5 in the model list regardless of whether a coding endpoint was detected. Pro/Max plans support it. - Z.AI setup detection message: softened from 'GLM-5 is not available' to 'GLM-5 may still be available depending on your plan tier'
2026-03-12 16:02:35 -07:00
# api_models is None — couldn't reach API. Accept and persist,
# but warn so typos don't silently break things.
provider_label = _PROVIDER_LABELS.get(normalized, normalized)
return {
"accepted": True,
fix: stop rejecting unlisted models, accept with warning instead * fix: use session_key instead of chat_id for adapter interrupt lookups monitor_for_interrupt() in _run_agent was using source.chat_id to query the adapter's has_pending_interrupt() and get_pending_message() methods. But the adapter stores interrupt events under build_session_key(source), which produces a different string (e.g. 'agent:main:telegram:dm' vs '123456'). This key mismatch meant the interrupt was never detected through the adapter path, which is the only active interrupt path for all adapter-based platforms (Telegram, Discord, Slack, etc.). The gateway-level interrupt path (in dispatch_message) is unreachable because the adapter intercepts the 2nd message in handle_message() before it reaches dispatch_message(). Result: sending a new message while subagents were running had no effect — the interrupt was silently lost. Fix: replace all source.chat_id references in the interrupt-related code within _run_agent() with the session_key parameter, which matches the adapter's storage keys. Also adds regression tests verifying session_key vs chat_id consistency. * debug: add file-based logging to CLI interrupt path Temporary instrumentation to diagnose why message-based interrupts don't seem to work during subagent execution. Logs to ~/.hermes/interrupt_debug.log (immune to redirect_stdout). Two log points: 1. When Enter handler puts message into _interrupt_queue 2. When chat() reads it and calls agent.interrupt() This will reveal whether the message reaches the queue and whether the interrupt is actually fired. * fix: accept unlisted models with warning instead of rejecting validate_requested_model() previously hard-rejected any model not found in the provider's API listing. This was too aggressive — users on higher plan tiers (e.g. Z.AI Pro/Max) may have access to models not shown in the public listing (like glm-5 on coding endpoints). Changes: - validate_requested_model: accept unlisted models with a warning note instead of blocking. The model is saved to config and used immediately. - Z.AI setup: always offer glm-5 in the model list regardless of whether a coding endpoint was detected. Pro/Max plans support it. - Z.AI setup detection message: softened from 'GLM-5 is not available' to 'GLM-5 may still be available depending on your plan tier'
2026-03-12 16:02:35 -07:00
"persist": True,
"recognized": False,
"message": (
fix: stop rejecting unlisted models, accept with warning instead * fix: use session_key instead of chat_id for adapter interrupt lookups monitor_for_interrupt() in _run_agent was using source.chat_id to query the adapter's has_pending_interrupt() and get_pending_message() methods. But the adapter stores interrupt events under build_session_key(source), which produces a different string (e.g. 'agent:main:telegram:dm' vs '123456'). This key mismatch meant the interrupt was never detected through the adapter path, which is the only active interrupt path for all adapter-based platforms (Telegram, Discord, Slack, etc.). The gateway-level interrupt path (in dispatch_message) is unreachable because the adapter intercepts the 2nd message in handle_message() before it reaches dispatch_message(). Result: sending a new message while subagents were running had no effect — the interrupt was silently lost. Fix: replace all source.chat_id references in the interrupt-related code within _run_agent() with the session_key parameter, which matches the adapter's storage keys. Also adds regression tests verifying session_key vs chat_id consistency. * debug: add file-based logging to CLI interrupt path Temporary instrumentation to diagnose why message-based interrupts don't seem to work during subagent execution. Logs to ~/.hermes/interrupt_debug.log (immune to redirect_stdout). Two log points: 1. When Enter handler puts message into _interrupt_queue 2. When chat() reads it and calls agent.interrupt() This will reveal whether the message reaches the queue and whether the interrupt is actually fired. * fix: accept unlisted models with warning instead of rejecting validate_requested_model() previously hard-rejected any model not found in the provider's API listing. This was too aggressive — users on higher plan tiers (e.g. Z.AI Pro/Max) may have access to models not shown in the public listing (like glm-5 on coding endpoints). Changes: - validate_requested_model: accept unlisted models with a warning note instead of blocking. The model is saved to config and used immediately. - Z.AI setup: always offer glm-5 in the model list regardless of whether a coding endpoint was detected. Pro/Max plans support it. - Z.AI setup detection message: softened from 'GLM-5 is not available' to 'GLM-5 may still be available depending on your plan tier'
2026-03-12 16:02:35 -07:00
f"Could not reach the {provider_label} API to validate `{requested}`. "
f"If the service isn't down, this model may not be valid."
),
}