feat: debug mode — hypothesis-driven debugging (Cursor-style)#45
Merged
kienbui1995 merged 4 commits intomainfrom Apr 12, 2026
Merged
feat: debug mode — hypothesis-driven debugging (Cursor-style)#45kienbui1995 merged 4 commits intomainfrom
kienbui1995 merged 4 commits intomainfrom
Conversation
- New 'debug' tool with 4 phases: hypothesize, instrument, analyze, fix - Guides agent through evidence-based debugging workflow - TUI '/debug' template activates debug mode in conversation - Agent generates hypotheses → adds logging → analyzes evidence → targeted fix - 30 tools total, 192 tests pass, 0 warnings
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
📝 WalkthroughWalkthroughAdds a new synchronous "debug" tool: spec, core handler ( Changes
Sequence DiagramsequenceDiagram
participant User as User/TUI
participant Runtime as Runtime
participant Registry as Tool Registry
participant Debug as execute_debug
User->>Runtime: invoke tool "debug" with input (action + params)
Runtime->>Registry: identify tool "debug" (sequential)
Registry->>Runtime: confirm dispatch
Runtime->>Debug: call execute_debug(input)
alt action = "hypothesize"
Debug->>Debug: read `bug_description`
Debug->>User: return hypothesis message, is_error=false
else action = "instrument"
Debug->>Debug: validate `file` and optional `hypotheses`
alt missing required fields
Debug->>User: return error message, is_error=true
else
Debug->>User: return instrumentation message, is_error=false
end
else action = "analyze"
Debug->>Debug: validate `evidence`
alt evidence missing
Debug->>User: return error message, is_error=true
else
Debug->>User: return analysis message, is_error=false
end
else action = "fix"
Debug->>Debug: read `root_cause` and `file`
Debug->>User: return fix message, is_error=false
else unknown action
Debug->>User: return "Unknown debug action" error, is_error=true
end
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
Owner
Author
|
/review |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
mc/crates/mc-tui/src/commands.rs (1)
383-396:⚠️ Potential issue | 🟡 MinorExpose
debugin the/templatehelp list.
/templatenow supportsdebug(Line 396), but the list shown when no args are provided still omits it (Line 384), so users won’t discover it.💡 Proposed fix
- app.push("Templates: review, refactor, test, explain, document, optimize, security"); + app.push("Templates: review, refactor, test, explain, document, optimize, security, debug");🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@mc/crates/mc-tui/src/commands.rs` around lines 383 - 396, The templates help list omits the new "debug" template when arg.is_empty() is true; update the string pushed to the UI (the first app.push call that currently lists "Templates: review, refactor, test, explain, document, optimize, security") to include "debug" (e.g., append ", debug" or insert at desired position) so the /template help output exposes the Debug Mode entry; keep the Usage line as-is and ensure the change is made near the arg.is_empty() branch in commands.rs where app.push is called.
🧹 Nitpick comments (2)
mc/crates/mc-core/src/runtime.rs (1)
1704-1719: Validatefixinputs like other debug phases.
fixaccepts emptyroot_causeandfile, which weakens the workflow and can cause extra loops. Consider enforcing non-empty values (similar toinstrumentandanalyze).🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@mc/crates/mc-core/src/runtime.rs` around lines 1704 - 1719, The "fix" debug branch currently allows empty root_cause and file (variables root_cause and file in the "fix" match arm), so update runtime.rs to validate them the same way as the instrument/analyze phases: extract with input.get(...).and_then(|v| v.as_str()).filter(|s| !s.is_empty()) and if either is missing/empty return or propagate an error (or produce a validation prompt) instead of defaulting to ""; ensure the match arm for "fix" fails fast on missing values and logs/returns a clear validation error so the prompt is only built when both root_cause and file are non-empty.mc/crates/mc-tools/src/spec.rs (1)
391-405: Tightendebugschema with action-specific required fields.Right now only
actionis required, but runtime rejects missingfileforinstrumentand missingevidenceforanalyze. Encoding that in schema (oneOf/if-then) would reduce avoidable tool-call failures.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@mc/crates/mc-tools/src/spec.rs` around lines 391 - 405, The debug ToolSpec's input_schema currently only requires "action", causing runtime failures when action-specific fields are missing; update the ToolSpec with an action-specific schema (use JSON Schema oneOf or if/then branches) inside the input_schema for the ToolSpec named "debug" so that: when action == "hypothesize" require "bug_description"; when action == "instrument" require "file" (and optionally "hypotheses"); when action == "analyze" require "evidence"; when action == "fix" require "root_cause" (and "file" if fixes target a file). Modify the input_schema definition in the ToolSpec block to include these conditional/oneOf subschemas so the runtime validates required fields per action.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@mc/crates/mc-core/src/runtime.rs`:
- Around line 1208-1209: The "debug" tool is registered in runtime
(execute_debug) but run_turn currently doesn't classify "debug" as a sequential
tool, so calls go through the parallel path (ToolRegistry::execute) where no
debug handler exists; update the classification in run_turn to treat "debug" as
sequential (or alternatively add a debug branch in ToolRegistry::execute) so
that execute_debug is reached when name == "debug" and Debug Mode no longer
returns NotFound; reference run_turn and execute_debug (or ToolRegistry::execute
if you choose the alternate fix) when making the change.
---
Outside diff comments:
In `@mc/crates/mc-tui/src/commands.rs`:
- Around line 383-396: The templates help list omits the new "debug" template
when arg.is_empty() is true; update the string pushed to the UI (the first
app.push call that currently lists "Templates: review, refactor, test, explain,
document, optimize, security") to include "debug" (e.g., append ", debug" or
insert at desired position) so the /template help output exposes the Debug Mode
entry; keep the Usage line as-is and ensure the change is made near the
arg.is_empty() branch in commands.rs where app.push is called.
---
Nitpick comments:
In `@mc/crates/mc-core/src/runtime.rs`:
- Around line 1704-1719: The "fix" debug branch currently allows empty
root_cause and file (variables root_cause and file in the "fix" match arm), so
update runtime.rs to validate them the same way as the instrument/analyze
phases: extract with input.get(...).and_then(|v| v.as_str()).filter(|s|
!s.is_empty()) and if either is missing/empty return or propagate an error (or
produce a validation prompt) instead of defaulting to ""; ensure the match arm
for "fix" fails fast on missing values and logs/returns a clear validation error
so the prompt is only built when both root_cause and file are non-empty.
In `@mc/crates/mc-tools/src/spec.rs`:
- Around line 391-405: The debug ToolSpec's input_schema currently only requires
"action", causing runtime failures when action-specific fields are missing;
update the ToolSpec with an action-specific schema (use JSON Schema oneOf or
if/then branches) inside the input_schema for the ToolSpec named "debug" so
that: when action == "hypothesize" require "bug_description"; when action ==
"instrument" require "file" (and optionally "hypotheses"); when action ==
"analyze" require "evidence"; when action == "fix" require "root_cause" (and
"file" if fixes target a file). Modify the input_schema definition in the
ToolSpec block to include these conditional/oneOf subschemas so the runtime
validates required fields per action.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 092baf62-5463-4577-80ac-f7077a6341ab
📒 Files selected for processing (4)
mc/crates/mc-core/src/runtime.rsmc/crates/mc-tools/src/registry.rsmc/crates/mc-tools/src/spec.rsmc/crates/mc-tui/src/commands.rs
1. Add 'debug' and 'browser' to sequential tool list (was going through parallel path where no handler exists — critical bug) 2. Add 'debug' to /template help list 3. Validate root_cause and file in debug 'fix' action 192 tests, 0 warnings.
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
mc/crates/mc-core/src/runtime.rs (1)
1640-1732: Consider moving Debug Mode into its ownmc-coremodule.This handler is already feature-sized, and
runtime.rsis doing a lot of orchestration work already. Extractingexecute_debuginto something likemc-core/src/debug.rswould keep dispatch here thin and make the new workflow easier to evolve and test.Based on learnings: When adding a new core feature: create module in
mc-core/src/{feature}.rs, integrate into ConversationRuntime in runtime.rs, export frommc-core/src/lib.rs, add to config if needed, add slash command in TUI🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@mc/crates/mc-core/src/runtime.rs` around lines 1640 - 1732, The execute_debug handler is large and should be extracted into its own module named debug: create a new module debug that exposes a pub fn execute_debug(input: &serde_json::Value) -> (String, bool) (migrating the entire match body), then replace the inline implementation in runtime.rs with a thin dispatch that calls debug::execute_debug; update the crate root to pub mod debug / pub use debug::execute_debug so the function is exported, wire the ConversationRuntime dispatch to call the new function, and update any config/TUI slash-command registration and tests that referenced the old inline handler to point to the new module.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@mc/crates/mc-core/src/runtime.rs`:
- Around line 1661-1717: The runtime's execute_debug branches ("instrument",
"analyze", "fix") enforce required fields (file, hypotheses/evidence/root_cause)
but the registered debug ToolSpec still only marks action as required; update
the ToolSpec registration for the "debug" tool in mc/crates/mc-tools/src/spec.rs
(ToolSpec for debug) to reflect per-action required parameters: require "file"
for action="instrument" and action="fix", require "hypotheses" (or mark its
expected type) for "instrument", require "evidence" for action="analyze", and
require "root_cause" for action="fix" so the schema matches execute_debug's
validations and prevents the model from invoking the tool without those fields.
---
Nitpick comments:
In `@mc/crates/mc-core/src/runtime.rs`:
- Around line 1640-1732: The execute_debug handler is large and should be
extracted into its own module named debug: create a new module debug that
exposes a pub fn execute_debug(input: &serde_json::Value) -> (String, bool)
(migrating the entire match body), then replace the inline implementation in
runtime.rs with a thin dispatch that calls debug::execute_debug; update the
crate root to pub mod debug / pub use debug::execute_debug so the function is
exported, wire the ConversationRuntime dispatch to call the new function, and
update any config/TUI slash-command registration and tests that referenced the
old inline handler to point to the new module.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: cddb14c3-4e6a-4cdc-ae01-22fbebe50265
📒 Files selected for processing (2)
mc/crates/mc-core/src/runtime.rsmc/crates/mc-tui/src/commands.rs
🚧 Files skipped from review as they are similar to previous changes (1)
- mc/crates/mc-tui/src/commands.rs
1. Extract execute_debug to mc-core/src/debug.rs (was inline in runtime.rs) 2. Add 4 unit tests for debug module 3. Tighten ToolSpec schema with per-action required fields (allOf/if-then) 4. Export debug module from mc-core/src/lib.rs 196 tests (4 new), 0 warnings.
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
mc/crates/mc-core/src/debug.rs (1)
1-2: Markexecute_debugas#[must_use].This is a public function returning a value, so it should carry the repo-required annotation.
Based on learnings: "Add `#[must_use]` attribute on all public functions returning values"♻️ Proposed fix
/// Structured debug mode — guides agent through hypothesis-driven debugging. +#[must_use] pub fn execute_debug(input: &serde_json::Value) -> (String, bool) {🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@mc/crates/mc-core/src/debug.rs` around lines 1 - 2, The public function execute_debug returns a value but lacks the required must_use annotation; add the #[must_use] attribute directly above the pub fn execute_debug(...) declaration so callers get a compiler warning if the returned (String, bool) tuple is ignored, preserving the function signature and visibility.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@mc/crates/mc-core/src/debug.rs`:
- Around line 6-9: The code currently sets bug to "unknown bug" when
input.get("bug_description") is missing; instead, make the "bug_description"
required by checking input.get("bug_description").and_then(|v| v.as_str()) and
returning an error (or propagating a failure) if it's None rather than using
unwrap_or; update the branch that sets the variable named bug in
mc/crates/mc-core/src/debug.rs so the "hypothesize" path fails fast when
bug_description is absent and include a clear error message referencing the
missing "bug_description".
---
Nitpick comments:
In `@mc/crates/mc-core/src/debug.rs`:
- Around line 1-2: The public function execute_debug returns a value but lacks
the required must_use annotation; add the #[must_use] attribute directly above
the pub fn execute_debug(...) declaration so callers get a compiler warning if
the returned (String, bool) tuple is ignored, preserving the function signature
and visibility.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: cd0310ea-f7b4-464e-8e00-4f59c66f340b
📒 Files selected for processing (4)
mc/crates/mc-core/src/debug.rsmc/crates/mc-core/src/lib.rsmc/crates/mc-core/src/runtime.rsmc/crates/mc-tools/src/spec.rs
✅ Files skipped from review due to trivial changes (2)
- mc/crates/mc-core/src/lib.rs
- mc/crates/mc-tools/src/spec.rs
- Add #[must_use] to execute_debug - Require bug_description for hypothesize (was defaulting to 'unknown bug') - Add test for missing bug_description 197 tests, 0 warnings.
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
1 participant
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.
Debug Mode
Structured debugging workflow inspired by Cursor's Debug Mode:
4 phases:
hypothesize— generate 3-5 hypotheses about root causeinstrument— add targeted logging/assertions to verify each hypothesisanalyze— compare evidence against hypotheses, identify root causefix— apply targeted fix, remove instrumentation, add regression testUsage:
/debugin TUI or agent callsdebugtool directlyWhy: Standard agent debugging = guess and fix. Debug Mode = evidence-based, systematic, fewer iterations.
30 tools, 192 tests, 0 warnings.
Summary by CodeRabbit
/template debugcommand to generate step-by-step debug prompts.