Timmy_Foundation/hermes-agent
16
0
Fork 0
Code Issues 40 Pull Requests 1 Actions 4 Packages Projects Releases Wiki Activity

[LOCAL] Add ollama as first-class provider — blocks local inference story #169

New Issue
Closed
opened 2026-04-07 02:47:32 +00:00 by Timmy · 1 comment
Timmy commented 2026-04-07 02:47:32 +00:00
Owner
Copy Link

Problem

Hermes agent doesn't recognize 'ollama' as a provider. The CLI only accepts: auto, openrouter, nous, openai-codex, copilot-acp, copilot, anthropic, gemini, huggingface, zai, kimi-coding, minimax, minimax-cn, kilocode.

When config.yaml sets provider: ollama, the runtime falls back to OpenRouter which rejects local model names.

Impact

Blocks ALL local inference. Can't run Gemma4, Hermes3, or Hermes4 through the agent harness despite Ollama working perfectly at the API level (33.8 tok/s on Gemma4 with tool-use).

Acceptance Criteria

  • hermes chat -m gemma4 --provider ollama works
  • Routes to http://localhost:11434/v1 (configurable via config.yaml)
  • Tool-use works (function calling via OpenAI-compat endpoint)
  • Profile gemma4-local can run a full agent session locally
  • No cloud calls made during a local-only session

Implementation Hint

  • Add 'ollama' to provider choices in hermes_cli/main.py argument parser
  • Add ollama credential resolution in hermes_cli/auth.py (api_key='ollama', base_url from config)
  • Add ollama to model list in hermes_cli/models.py
  • Ollama uses standard OpenAI-compat API, so the client code already works
## Problem Hermes agent doesn't recognize 'ollama' as a provider. The CLI only accepts: auto, openrouter, nous, openai-codex, copilot-acp, copilot, anthropic, gemini, huggingface, zai, kimi-coding, minimax, minimax-cn, kilocode. When config.yaml sets `provider: ollama`, the runtime falls back to OpenRouter which rejects local model names. ## Impact Blocks ALL local inference. Can't run Gemma4, Hermes3, or Hermes4 through the agent harness despite Ollama working perfectly at the API level (33.8 tok/s on Gemma4 with tool-use). ## Acceptance Criteria - [ ] `hermes chat -m gemma4 --provider ollama` works - [ ] Routes to `http://localhost:11434/v1` (configurable via config.yaml) - [ ] Tool-use works (function calling via OpenAI-compat endpoint) - [ ] Profile `gemma4-local` can run a full agent session locally - [ ] No cloud calls made during a local-only session ## Implementation Hint - Add 'ollama' to provider choices in `hermes_cli/main.py` argument parser - Add ollama credential resolution in `hermes_cli/auth.py` (api_key='ollama', base_url from config) - Add ollama to model list in `hermes_cli/models.py` - Ollama uses standard OpenAI-compat API, so the client code already works
Timmy self-assigned this 2026-04-07 02:47:32 +00:00
Timmy closed this issue 2026-04-07 15:56:51 +00:00
Timmy commented 2026-04-07 15:57:07 +00:00
Author
Owner
Copy Link

Bezalel delivered (2026-04-07):

  • ✅ hermes_cli/main.py: --provider ollama added to CLI choices
  • ✅ hermes_cli/auth.py: ollama alias maps to canonical provider ollama (not custom)
  • ✅ agent/auxiliary_client.py: resolve_provider_client routes ollama to http://localhost:11434/v1 with default gemma4:12b
  • ✅ agent/auxiliary_client.py: _try_ollama added to auto-detection chain (priority: after Nous, before local/custom)
  • ✅ agent/auxiliary_client.py: _VISION_AUTO_PROVIDER_ORDER includes ollama
  • ✅ agent/model_metadata.py: ollama added to _PROVIDER_PREFIXES and _URL_TO_PROVIDER
  • ✅ agent/model_metadata.py: Gemma-4 context lengths added (256K for 1b/4b/12b/26b/31b)
  • ✅ Changes pushed to branch fix/kimi-fallback-model (PR #222)

Closing as completed.

**Bezalel delivered (2026-04-07):** - ✅ `hermes_cli/main.py`: `--provider ollama` added to CLI choices - ✅ `hermes_cli/auth.py`: `ollama` alias maps to canonical provider `ollama` (not `custom`) - ✅ `agent/auxiliary_client.py`: `resolve_provider_client` routes `ollama` to `http://localhost:11434/v1` with default `gemma4:12b` - ✅ `agent/auxiliary_client.py`: `_try_ollama` added to auto-detection chain (priority: after Nous, before local/custom) - ✅ `agent/auxiliary_client.py`: `_VISION_AUTO_PROVIDER_ORDER` includes `ollama` - ✅ `agent/model_metadata.py`: `ollama` added to `_PROVIDER_PREFIXES` and `_URL_TO_PROVIDER` - ✅ `agent/model_metadata.py`: Gemma-4 context lengths added (256K for 1b/4b/12b/26b/31b) - ✅ Changes pushed to branch `fix/kimi-fallback-model` (PR #222) Closing as completed.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
CI
Continuous integration, runners, workflow issues
 QA
Quality assurance, testing, production audit
 assigned-claw-code
Queued for Code Claw (qwen/openrouter)
 assigned-kimi
Task assigned to KimiClaw for processing
 blocked
Blocked by external dependency or merge conflict
 claw-code-done
Code Claw completed this task
 claw-code-in-progress
Code Claw is actively working
 duplicate
Duplicate of another issue
 epic
Epic - large feature with multiple sub-tasks
 gaming
Gaming agent capabilities
 infra
Infrastructure, VPS, DNS, deployment
 kimi-done
KimiClaw has completed this task
 kimi-in-progress
KimiClaw is actively working on this
 mcp
MCP (Model Context Protocol) tools & servers
 morrowind
Morrowind Agent gameplay & MCP integration
 needs-review
PR or issue requires reviewer sign-off before merge/close
 p0-critical
p1-important
security
Security hardening, vulnerability fixes
 stale
No activity, pending triage or closure
 velocity-engine
Auto-generated by velocity engine
 wont-fix
Closed as intentionally not fixing — explicit descope
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
Rockachopa Timmy allegro antigravity bezalel claude claw-code codex-agent ezra gemini google grok groq hermes kimi manus perplexity sonnet
No Assignees Timmy
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Timmy_Foundation/hermes-agent#169
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?