chore(opencode): route reasoning effort through model aliases #251

Merged

jwilger merged 1 commit from fix-agent-variant-effort into main

2026-05-18 11:44:46 -07:00

jwilger commented

2026-05-18 11:40:55 -07:00

Owner

Why:

Agent switching preserves reasoning effort when agents share the same model ID, so build and plan need distinct configured model identities.

What:

Add low/high gpt-5.5 OpenAI aliases with matching reasoning effort options.
Point build at the low alias and high-reasoning agents at the high alias.
Update the plan-advisor agent file to use the high alias.

Validation:

opencode models openai
opencode agent list

Why: - Agent switching preserves reasoning effort when agents share the same model ID, so build and plan need distinct configured model identities. What: - Add low/high `gpt-5.5` OpenAI aliases with matching reasoning effort options. - Point build at the low alias and high-reasoning agents at the high alias. - Update the plan-advisor agent file to use the high alias. Validation: - `opencode models openai` - `opencode agent list`

jwilger added 1 commit

2026-05-18 11:40:55 -07:00

chore(opencode): route reasoning effort through model aliases

CI / Clippy (pull_request) Successful in 40s

Details

CI / Request auto_review semantic review (pull_request) Successful in 1s

Details

CI / Build PR artifacts (no token) (pull_request) Successful in 2s

Details

auto_review auto_review: no findings

CI / Format check (pull_request) Successful in 5s

Details

CI / Dependency policy (pull_request) Successful in 12s

Details

CI / Test (pull_request) Successful in 54s

Details

CI / Build (pull_request) Successful in 33s

Details

adeecbfcae

Why:
- Agent switching preserves reasoning effort when agents share the same model ID, so build and plan need distinct configured model identities.

What:
- Add low/high gpt-5.5 OpenAI aliases with matching reasoning effort options.
- Point build at the low alias and high-reasoning agents at the high alias.

Validation:
- opencode models openai
- opencode agent list

auto-review approved these changes

2026-05-18 11:42:37 -07:00

auto-review left a comment

The PR introduces model aliases for OpenAI's gpt-5.5 to differentiate between low and high reasoning efforts, updating the configuration to use these aliases. This change appears safe to merge, as it aligns with the goal of preserving reasoning effort across agent switches.

Walkthrough

.opencode/agents/plan-advisor.md:
- Updated the model to use the openai/gpt-5.5-high alias, reflecting the high reasoning effort configuration.
opencode.json:
- Added new model aliases gpt-5.5-low and gpt-5.5-high with corresponding reasoning effort options.
- Updated various agents to use these new aliases, ensuring that the reasoning effort is preserved across different configurations.

The PR introduces model aliases for OpenAI's `gpt-5.5` to differentiate between low and high reasoning efforts, updating the configuration to use these aliases. This change appears safe to merge, as it aligns with the goal of preserving reasoning effort across agent switches. ## Walkthrough - **.opencode/agents/plan-advisor.md**: - Updated the model to use the `openai/gpt-5.5-high` alias, reflecting the high reasoning effort configuration. - **opencode.json**: - Added new model aliases `gpt-5.5-low` and `gpt-5.5-high` with corresponding reasoning effort options. - Updated various agents to use these new aliases, ensuring that the reasoning effort is preserved across different configurations.

jwilger referenced this pull request

2026-05-18 11:44:38 -07:00

Streamline Actions workflows with path-aware checks #252

jwilger merged commit bf25f76012 into main

2026-05-18 11:44:46 -07:00

jwilger referenced this pull request from a commit

2026-05-18 11:44:46 -07:00

chore(opencode): route reasoning effort through model aliases (#251)