feat: add --effort support via model[:effort] syntax#285
Merged
Conversation
Extends task_model/review_model config and --task-model/--review-model CLI flags to carry Claude's --effort level using model[:effort] syntax (e.g., opus:high, :medium). Both parts are optional and independently usable. - ClaudeExecutor gains Effort field; --effort is injected alongside --model and deduped from claude_args (handles --effort=val and bare --effort) - ParseModelEffort helper splits spec on first colon; exported for external test package access - Runner.New compares parsed (model, effort) tuples when deciding whether to build a separate review executor, so "opus" and "opus:" don't produce redundant executors - Integration test verifies New wiring across 7 spec combinations, including same-model-different-effort (the main feature motivation) - docs/custom-providers.md command template updated to show --effort - Updated README, CLAUDE.md, llms.txt, config defaults, field godoc
Deploying ralphex with
|
| Latest commit: |
70fb73a
|
| Status: | ⚡️ Build in progress... |
There was a problem hiding this comment.
Pull request overview
Adds support for carrying Claude “reasoning effort” alongside the selected model by extending task_model / review_model (and corresponding CLI flags) to accept model[:effort], then wiring that into ClaudeExecutor so --effort <level> is injected next to --model <model>.
Changes:
- Introduces
ClaudeExecutor.Effortand injects/dedupes--effortin the generated Claude CLI args. - Adds
processor.ParseModelEffortand updates runner construction to compare parsed(model, effort)tuples (avoids redundant review executors for equivalent specs likeopusvsopus:). - Extends tests and updates docs/config defaults/README/CLAUDE.md/llms.txt to document the new syntax and behavior.
Reviewed changes
Copilot reviewed 12 out of 12 changed files in this pull request and generated 1 comment.
Show a summary per file
| File | Description |
|---|---|
| pkg/processor/runner.go | Parses model[:effort] and wires model/effort into task + review executors (tuple comparison for reuse). |
| pkg/processor/runner_test.go | Adds tests for parsing and runner wiring across model/effort combinations. |
| pkg/processor/export_test.go | Adds test-only accessors to inspect chosen executors from external tests. |
| pkg/executor/executor.go | Adds Effort field and --effort injection with dedupe behavior. |
| pkg/executor/executor_test.go | Adds coverage for --effort injection and dedupe (including --effort=... form). |
| pkg/config/defaults/config | Documents model[:effort] syntax and effort levels in embedded defaults. |
| pkg/config/config.go | Updates config field doc comments to reflect model[:effort]. |
| cmd/ralphex/main.go | Updates CLI flag descriptions to reflect model[:effort]. |
| docs/custom-providers.md | Updates provider command template and explains --model/--effort injection + stripping. |
| README.md | Updates CLI/config docs tables to document model[:effort] and effort levels. |
| CLAUDE.md | Updates user-facing documentation for model/effort configuration and wiring. |
| llms.txt | Updates configuration documentation and examples to include effort. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+251
to
+256
| // inject --effort flag if an effort override is configured; | ||
| // strip any existing --effort from args to avoid duplicate/conflicting flags | ||
| if e.Effort != "" { | ||
| args = stripFlag(args, "--effort") | ||
| args = append(args, "--effort", e.Effort) | ||
| } |
There was a problem hiding this comment.
stripFlag is used here to remove any pre-existing --effort before injecting the override, but its current implementation always skips the token following --effort when it appears as a standalone arg. If a user (or wrapper) has a bare --effort followed by another flag (e.g. --effort --verbose), stripFlag will incorrectly drop --verbose too. Consider updating stripFlag to only skip the next token when it’s a value (e.g., next token doesn’t start with -), so the documented “bare --effort” case is handled safely.
addresses copilot inline review on PR #285: stripFlag was too aggressive when encountering a bare flag in the middle of args — it always consumed the next token as a value, eating unrelated flags like "--effort --verbose". now only consumes the next token when it doesn't start with a dash.
|
@umputun whats your personal preferences on this with opus 4.7? |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Extends
task_model/review_model(and their CLI flags) to carry Claude's--effortlevel alongside the model, usingmodel[:effort]syntax. Single flag, both knobs.Examples:
ralphex --task-model=opus:high docs/plans/foo.md ralphex --review-model=:medium # effort onlyEffort levels:
low,medium,high,xhigh,max. Either part is optional:opus,opus:high,:medium,opus:, all work.Changes:
ClaudeExecutor.Effortfield; injects--effort <val>alongside existing--model, with the same dedupe treatment (handles--effort=valand bare--effort)ParseModelEfforthelper in pkg/processor (exported for external test package)Runner.Newcompares parsed(model, effort)tuples when deciding whether to build a separate review executor — soopusandopus:don't produce redundant executorsTestRunner_New_ModelEffortWiringcovers 7 spec combinations, including the same-model-different-effort case (the main feature motivation)docs/custom-providers.mdcommand template now shows--effort; README, CLAUDE.md, llms.txt, config defaults, field godoc all mention the new syntaxNot in a release yet, so no compat shim.