gitea/agents
8
0
Fork 0
Code Issues 2 Pull Requests 1 Actions Packages Projects Releases Wiki Activity

feat(agents): automatic Ollama Cloud → xAI fallback #121

Merged
hermes merged 2 commits from fix/agents-xai-fallback-ollama-exhausted into main 2026-07-30 15:28:45 +02:00
Conversation 0 Commits 2 Files Changed 5
+39 -11
hermes commented 2026-07-30 15:25:47 +02:00
Collaborator
Copy Link
Copy Source

Why

Ollama Cloud hits 100% usage and agents die. Better than hard-switching: try primary, fall back automatically.

How

agents.json keeps Ollama Cloud as primary and adds:

"fallback": "xai-oc/grok-4.5"

run-agent.sh on quota/credit/429/capacity-class errors:

  1. log failover
  2. set MODEL=fallback once
  3. continue remaining retry attempts on xAI

install-opencode.sh registers both primary and fallback models in provider maps so opencode can resolve either.

Matrix

Agent Primary Fallback
pm ollama-cloud/minimax-m3 xai-oc/grok-4.5
junior ollama-cloud/kimi-k2.7-code xai-oc/grok-4.3
senior ollama-cloud/glm-5.2 xai-oc/grok-4.5
qa ollama-cloud/minimax-m3 xai-oc/grok-4.5
intern ollama/ornith:35b xai-oc/grok-4.3
lead/ops xai-oc/grok-4.5 (none)

Needs existing XAI_API_KEY Actions secret.

After merge

Re-@pm stuck issues (e.g. homelab #221).

## Why Ollama Cloud hits 100% usage and agents die. Better than hard-switching: **try primary, fall back automatically**. ## How `agents.json` keeps Ollama Cloud as primary and adds: ```json "fallback": "xai-oc/grok-4.5" ``` `run-agent.sh` on quota/credit/429/capacity-class errors: 1. log failover 2. set MODEL=fallback once 3. continue remaining retry attempts on xAI `install-opencode.sh` registers both primary and fallback models in provider maps so opencode can resolve either. ## Matrix | Agent | Primary | Fallback | |-------|---------|----------| | pm | ollama-cloud/minimax-m3 | xai-oc/grok-4.5 | | junior | ollama-cloud/kimi-k2.7-code | xai-oc/grok-4.3 | | senior | ollama-cloud/glm-5.2 | xai-oc/grok-4.5 | | qa | ollama-cloud/minimax-m3 | xai-oc/grok-4.5 | | intern | ollama/ornith:35b | xai-oc/grok-4.3 | | lead/ops | xai-oc/grok-4.5 | (none) | Needs existing `XAI_API_KEY` Actions secret. ## After merge Re-@pm stuck issues (e.g. homelab #221).
hermes added 1 commit 2026-07-30 15:25:47 +02:00
fix(agents): route pm/junior/senior/qa to xAI while Ollama Cloud is exhausted
ci / lint (pull_request) Successful in 13s
Details
bbdb200f32 
Ollama Cloud usage is at 100% so ollama-cloud/* agents fail. Move
pm, junior, senior, qa to xai-oc (grok-4.5 / grok-4.3). leave intern on
local ollama/ornith:35b; lead/ops already on xAI.
hermes changed title from Route agents to xAI while Ollama Cloud quota is exhausted to feat(agents): automatic Ollama Cloud → xAI fallback 2026-07-30 15:28:31 +02:00
hermes added 1 commit 2026-07-30 15:28:31 +02:00
feat(agents): automatic model fallback when Ollama Cloud is exhausted
ci / lint (pull_request) Successful in 13s
Details
0ddde87b40 
Keep ollama-cloud/* as primary. Each agent may declare fallback in
agents.json (xai-oc/grok-*). On quota/auth/provider failure, run-agent
switches once to the fallback and continues retries.

- agents.json: fallback fields for pm/junior/senior/qa/intern
- route.sh: emit fallback to GITHUB_OUTPUT
- agent.yml: pass FALLBACK env into run-agent
- install-opencode.sh: register primary+fallback models in provider maps
- run-agent.sh: failover on quota/credit/429/capacity-class errors
hermes merged commit 58db2996ad into main 2026-07-30 15:28:45 +02:00
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
Clear labels
autopilot
run agents without asking
No labels
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
ffaerber hermes intern junior lead ops (Ops Agent) pm qa (QA Agent) senior
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: gitea/agents#121