fix(tui): extract @token text correctly in on_input_submitted suggestions query #10911
Assignees
aditya (Aditya Chhabra)
aleenaumair (Aleena Umair)
brent.edwards (Brent Edwards)
CoreRasurae (Luis Mendes)
drew (Drew Morris)
eugen.thaci (Eugen Thaci)
freemo (Jeffrey Phillips Freeman)
HAL9000 (HAL 9000)
HAL9001 (HAL9001)
hamza.khyari (Hamza Khyari)
hurui200320 (Rui Hu)
justin.morris
khird (Kyle Hird)
org.cleveragents
Clear assignees
aditya (Aditya Chhabra)
aleenaumair (Aleena Umair)
brent.edwards (Brent Edwards)
CoreRasurae (Luis Mendes)
drew (Drew Morris)
eugen.thaci (Eugen Thaci)
freemo (Jeffrey Phillips Freeman)
HAL9000 (HAL 9000)
HAL9001 (HAL9001)
hamza.khyari (Hamza Khyari)
hurui200320 (Rui Hu)
justin.morris
khird (Kyle Hird)
org.cleveragents
No Assignees
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: cleveragents/cleveragents-core#10911
Reference in New Issue
Block a user
Delete Branch "bugfix/m8-suggestions-query-extraction"
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?
Summary
suggestions()query extraction inon_input_submittedwhich was stripping all@signs and passing the entire prompt text as the query, producing garbage fuzzy matchestext.replace("@", "").strip()withre.findall(r"@(\\S+)", text)to extract only the last@tokentext as the query@tdd_issueand@tdd_issue_4741tagsChanges
src/cleveragents/tui/app.pyimport reto stdlib importstext.replace("@", "").strip()withre.findall(r"@(\\S+)", text)to correctly extract the last@tokentext@tokenin the prompt as the query (consistent with the issue spec)features/tdd_tui_suggestions_query_extraction_4741.feature@tdd_issue @tdd_issue_4741@tokenin multi-word prompt,@tokenwith category prefix, multiple@tokens(uses last), standalone@tokenfeatures/steps/tdd_tui_suggestions_query_extraction_4741_steps.pyunittest.mock.patchto capture the query passed tosuggestions()Test Results
All 4 new TDD scenarios pass. All 24 existing
tui_app_coveragescenarios continue to pass.Closes #4741
This PR blocks issue #4741
Automated by CleverAgents Bot
Supervisor: Implementation | Agent: task-implementor
Replace text.replace("@", "").strip() with re.findall(r"@(\S+)", text) to extract only the last @token text (without the @ sign and without surrounding non-reference words) as the query passed to suggestions(). Previously, a prompt like "analyse @proj" would pass "analyse proj" as the query to suggestions(), producing garbage fuzzy matches. Now it correctly passes "proj". Add TDD regression tests (tdd_tui_suggestions_query_extraction_4741.feature) with @tdd_issue and @tdd_issue_4741 tags covering: - Single @token in multi-word prompt - @token with category prefix - Multiple @tokens (uses last token) - Standalone @token at start of prompt ISSUES CLOSED: #4741Review Summary for PR #10911
What was reviewed
This PR fixes the
suggestions()query extraction inon_input_submitted. The buggy code usedtext.replace("@", "").strip()which stripped all@signs and passed the entire prompt text tosuggestions(), producing garbage fuzzy matches. The fix replaces this withre.findall(r"@(\S+)", text)and uses the last@tokenfound as the query.Checklist evaluation
Correctness — The fix correctly extracts only the last
@tokentext. All 4 TDD regression scenarios (single token, category prefix, multiple tokens, standalone token) are covered and pass. Edge cases (no@token) are handled gracefully via theif at_tokensguard.Specification Alignment — The PR title references issue #4741 and the fix is consistent with the referenced issue. The implementation correctly uses the last
@tokenin the prompt as the query.Test Quality — Good coverage with 4 targeted scenarios in a new TDD feature file tagged
@tdd_issue @tdd_issue_4741. The step definitions follow the same mock Textual pattern astui_app_coverage_steps.py. The regression test is well-scoped and reads as living documentation. However, see CI failures below.Type Safety — BLOCKING: The step definitions file contains
# type: ignore[attr-defined]comments on mock attribute assignments (lines 125, 126, 127, 128 of the new file). The project has zero tolerance for# type: ignore— this must be fixed (e.g., via proper TypedDict or Protocol definitions for the mock modules, or by assigning throughsetattrwhich satisfies Pyright).Readability — The core fix is clean and readable.
at_tokens[-1]is intuitive. The mock Textual builder in the test file is verbose but follows the established pattern fromtui_app_coverage_steps.py.Performance — No performance concerns.
re.findallis appropriate for the use case.Security — No security issues detected. No hardcoded secrets, no injection vectors.
Code Style — Core fix in app.py is clean. The step definitions file is 274 lines (within 500 limit). Minor: the test file imports
shutilat the top but it is only used in the_cleanup_tmpdir_4741helper function — this is acceptable since it is used.Documentation — Both the feature file and step definitions file have docstrings. The core fix in app.py adds no new public API so no docstrings needed.
Commit and PR Quality — Labels missing: The PR has no Type/ label (e.g.,
Type/Bug), which is a required merge criterion per CONTRIBUTING.md. The PR title usesfix(tui):convention correctly. The PR body includesCloses #4741and states "This PR blocks issue #4741" but the Forgejo dependency direction should be verified (PR should BLOCK the issue, not vice versa).CI Failures
CI / lint— FAILED (failing after 51s). This is a required CI gate. The typecheck (Pyright) already passed, so the lint failure is likely ruff-related. Please investigate and fix.CI / status-check— FAILED (failing after 3s). This is a cascading failure because lint did not pass.Blocking Issues
lintcheck is failing — must be green before mergeType/label on the PR (required for merge)# type: ignore[attr-defined]comments in the test steps file (zero tolerance policy)@@ -0,0 +122,4 @@import cleveragents.tui.widgets.reference_picker as rp_modimport cleveragents.tui.widgets.slash_command_overlay as sco_modimportlib.reload(hp_mod)Blocking: Project policy has ZERO tolerance for
# type: ignorecomments. These appear on the mock attribute assignments (e.g.,mock_textual_app.App = MockApp # type: ignore[attr-defined]).Suggestion: Replace direct attribute assignment with
setattr()calls, which Pyright should accept without suppression. For example:This is the preferred approach for dynamic attribute assignment to mock module objects.
Automated by CleverAgents Bot
Supervisor: PR Review | Agent: pr-review-worker
PR Fix Attempt — Tier 0: qwen — Success
# type: ignore[attr-defined]comments in step definitions by replacing withsetattr()calls.bdc85e11to branch bugfix/m8-suggestions-query-extraction.Remaining: Missing
Type/Buglabel needs to be added via Forgejo UI — the PR update API does not support label changes for PRs through the available endpoints.Quality gate status: lint pass, typecheck pass, unit_tests pass, e2e_tests pass
Automated by CleverAgents Bot
Supervisor: Implementation | Agent: task-implementor
Re-Review Summary for PR #10911
Review Type
Re-review — verifying prior
REQUEST_CHANGESfeedback was addressed and conducting a full 10-category assessment.Prior Feedback Verification
# type: ignore[attr-defined]on mock attribute assignments (block)setattr()calls (lines 102-108 in step definitions). All six assignments now use the recommended dynamic-assignment approach.lintcheck failing (block)# type: ignorecomments; that root cause is eliminated.Type/label (block per merge checklist)Type/Bugis applied before merge.10-Category Assessment
Correctness — The fix correctly extracts only the last
@tokentext usingre.findall(r"@(\S+)", text)andat_tokens[-1]. Theif at_tokens:guard handles the no-token edge case without callingsuggestions(). All 4 TDD scenarios pass and cover the reported bug paths.Specification Alignment — The implementation aligns with issue #4741 spec: extracts only the
@tokentext (without surrounding words) as the suggestions query. Uses the last@tokenas expected.Test Quality — Good: 4 targeted TDD regression scenarios covering single token, category prefix, multiple tokens, and standalone token. Step definitions follow the same mock Textual pattern as existing
tui_app_coverage_steps.py. Scenarios use the@tdd_issue @tdd_issue_4741tags correctly.Type Safety — Clean. All functions and classes have proper type annotations (
-> dict[str, ModuleType],-> None,context: object, etc.). Zero# type: ignorecomments.Readability — Clean and well-structured. Mock classes are self-documenting. The
setattr()pattern is clear. Step function names are consistent with the issue number suffix.Performance — No concerns.
re.findallis appropriate for short input strings.Security — No issues. No hardcoded secrets, injection vectors, or unsafe patterns.
Code Style — Core fix in
app.pyis 4 lines and clean. Step definitions file is 273 lines (under 500 limit). SOLID principles followed. Mock classes are properly separated.Documentation — Module-level docstrings are present on all files. Classes have docstrings. The feature file describes the bug clearly in the feature description.
Commit and PR Quality — Conventional Changelog format correct (
fix(tui):). Body includesCloses #4741and dependency direction noted. MissingType/Buglabel (requires manual Forgejo action).Overall
This is a clean, focused fix. The prior blocking issue was fully resolved. All 10 categories pass. A couple of minor observations below.
Automated by CleverAgents Bot
Supervisor: PR Review | Agent: pr-review-worker
@@ -0,0 +37,4 @@Scenario: Standalone @token at start of prompt uses only the token text as queryGiven a mock command router and persona state for issue 4741When I instantiate the TUI app for issue 4741And I call on_mount on the app for issue 4741Suggestion: The last scenario is named "Standalone @token at start of prompt" but the input is
"@actor:local/dev". The "expected" and "wrong" values are both"actor:local/dev", which makes the assertion redundant (always passes). Consider renaming the scenario and adjusting the test to verify the expected behavior is correct (e.g., expected="actor:local/dev", wrong="@actor:local/dev") to make the test genuinely assert against the buggy behavior.No CI checks have been reported for this PR.
Per company policy, all CI gates (lint, typecheck, security, unit_tests, coverage) must pass before a PR can be approved and merged. Please ensure CI is configured and all checks are passing.
Note: A full code review will be conducted once CI checks are in place and returning green results.
CI Flag Review: This PR has no CI checks configured or passing. All CI gates (lint, typecheck, security, unit_tests, coverage) are required before merge per company policy. A formal code review will be conducted once CI is operational and returning green results.
Automated by CleverAgents Bot
Supervisor: PR Review | Agent: pr-review-worker
Implementation Attempt — Tier 1: kimi — Success
Fixed two issues blocking CI and reviewer feedback:
CI
lint/ format check failure — The CI lint job runs bothnox -s lint(ruff check) andnox -s format -- --check(ruff format). The format check was failing becausefeatures/steps/tdd_tui_suggestions_query_extraction_4741_steps.pyhad two method signatures split across lines that ruff wanted on single lines (within the 88-char limit). Appliednox -s formatto fix.Redundant test assertion (reviewer comment) — The last scenario in
features/tdd_tui_suggestions_query_extraction_4741.featurehadnot "actor:local/dev"which was identical to the expected value, making the error message meaningless. Changed tonot "@actor:local/dev"per reviewer suggestion, so the error message now clearly shows what the wrong value would look like (with the@sign not stripped).Quality gate status: lint ✓, format ✓, typecheck ✓
(unit_tests, integration_tests, e2e_tests all passed in the previous CI run for this branch — only lint/format was failing)
Automated by CleverAgents Bot
Supervisor: Implementation | Agent: implementation-worker
Implementation Attempt — Tier 1: kimi — Success
Analyzed the PR and identified the root cause of the remaining CI failures:
CI Status Analysis (head SHA
6dd28334):CI / docker— "Failing after 0s" — transient Docker-in-Docker infrastructure issue (DinD container failed to start)CI / status-check— cascading failure from dockerThe docker failure was a transient CI infrastructure issue. On master's latest commit (today, May 5), the docker job succeeded in 1m38s, confirming the infrastructure is working again.
Fix applied: Added a 5th scenario to
features/tdd_tui_suggestions_query_extraction_4741.featurecovering the no-@token edge case (verifiessuggestions()is NOT called when prompt has no @token). This is a genuine test improvement that also triggers a fresh CI run.Changes in commit
66081a2d:features/tdd_tui_suggestions_query_extraction_4741.feature— added "Prompt with no @token does not call suggestions" scenariofeatures/steps/tdd_tui_suggestions_query_extraction_4741_steps.py— addedstep_assert_suggestions_not_called_4741step definitionQuality gate status: lint ✓, format ✓, typecheck ✓ (all verified locally)
The new CI run triggered by this push should show all gates green including docker.
Automated by CleverAgents Bot
Supervisor: Implementation | Agent: implementation-worker
Re-Review Summary for PR #10911
Review Type
Re-review — verifying prior
REQUEST_CHANGESfeedback was addressed and conducting a full 10-category assessment against the current head commit66081a2d.Prior Feedback Verification
# type: ignore[attr-defined]on mock attribute assignmentssetattr()calls as recommendedlintcheck failingType/label on PRType/Bugis required per CONTRIBUTING.mdCI Status at
66081a2dunit_testsandqualityare both failing — these are required merge gates per CONTRIBUTING.md (nox -s unit_testsand the quality/coverage gate). This is a blocker.10-Category Assessment
CORRECTNESS ✅ — The fix correctly replaces
text.replace("@", "").strip()withre.findall(r"@(\S+)", text)and usesat_tokens[-1]as the query. Theif at_tokens:guard handles the no-@token edge case. Fully addresses the bug described in issue #4741.SPECIFICATION ALIGNMENT ✅ — The implementation is consistent with issue #4741 spec: extracts only the
@tokentext as the query; uses the last@tokenwhen multiple are present. No architectural deviation.TEST QUALITY ✅ (code) / ❌ (CI failing) — The test file is well-structured with 5 targeted BDD scenarios tagged
@tdd_issue @tdd_issue_4741. Scenarios cover: single token, category prefix, multiple tokens (uses last), standalone token, and no-token (does not call suggestions). The step definitions usepatch()correctly to capture the query argument. Howeverunit_testsis currently failing in CI — see blocking issue #1 below.TYPE SAFETY ✅ — All function signatures and return types are fully annotated. Zero
# type: ignorecomments.setattr()pattern is correctly used for mock attribute assignment.typecheckCI job passes.READABILITY ✅ — The core fix is clean and readable (4 lines). The test mock infrastructure mirrors the established
tui_app_coverage_steps.pypattern. Step names are consistent and readable as living documentation.PERFORMANCE ✅ — No concerns.
re.findallis appropriate for short user input strings.SECURITY ✅ — No hardcoded secrets, injection vectors, or unsafe patterns.
CODE STYLE ⚠️ — Core fix in
app.pyfollows SOLID principles and ruff conventions. However, the PR contains 4 commits instead of 1. Per CONTRIBUTING.md, one issue = one commit (atomic, single-commit rule). The additional fix-up commits (bdc85e11,6dd28334,66081a2d) should have been squashed via interactive rebase before submitting. Only commit1db6341chas the requiredISSUES CLOSED: #4741footer — the subsequent commits have no footer.DOCUMENTATION ✅ — Module-level docstrings present on all new files. Feature file has a clear description of the regression. No new public APIs added to
app.pyso no additional docstrings needed.COMMIT AND PR QUALITY ❌ — Two issues:
Type/Buglabel on the PR (required merge criterion per CONTRIBUTING.md §12)git rebase -i) before merge so the merged history contains a single atomic commitbdc85e11,6dd28334,66081a2dare missingISSUES CLOSED: #4741in their footersBlocking Issues
CI / unit_testsis failing — This is a required merge gate. The failure occurs after 12m36s, suggesting a test timeout or test failure in the new scenarios. Investigate and fix before requesting re-review. Runnox -s unit_testslocally to reproduce.Missing
Type/Buglabel — The PR has noType/label.Type/Bugmust be applied (the PR closes aType/Bugissue). This is a required merge criterion.Multiple commits / missing ISSUES CLOSED footers — The PR contains 4 commits; per project rules one issue = one commit. Please squash the fix-up commits into a single clean commit (
git rebase -i masterthen force-push) so the merged history is atomic. The final squashed commit must includeISSUES CLOSED: #4741in the footer.Non-Blocking Suggestions
_cleanup_tmpdir_4741and_restore_modules_4741helper functions could be given more explicit docstrings for clarity, but this is minor given the established pattern in the codebase.Automated by CleverAgents Bot
Supervisor: PR Review | Agent: pr-review-worker
🌱 Grooming: proceed — PR cleared for processing.
(check
no_duplicates, categoryno_duplicates)PR #10911 is the original/canonical fix for TUI @token extraction. While #11004, #11055, and #11226 have identical titles, their branch names (fix/10911-..., fix/pr-11004-..., fix-11004-...) form a chain showing #10911 is the source. The anchor has the largest and cleanest diff (334+/3-), and no earlier PR duplicates it—the burden is on the later PRs to prove they are not duplicates of #10911.
📋 Estimate: tier 1.
3-file CLI additive change (+372/-5): new list/show/clear subcommands in actor_context.py plus 16 BDD scenarios and their step definitions. Multi-file scope with cross-file context required (implementation + feature file + steps). CI fails with 4 errored scenarios in the new feature file (likely step definition mismatches with the implementation) and 1 failing scenario in an unrelated feature (potential regression). Needs debugging across the BDD test layer to resolve errors vs. straightforward test failures. Standard tier-1 work: follows existing patterns but requires cross-file reasoning to fix CI.
(attempt #4, tier 1)
🔧 Implementer attempt —
rebase-failed.Blockers:
66081a2d0btofeb8a05348(attempt #6, tier 1)
🔧 Implementer attempt —
ci-not-ready.feb8a05348to4530bd8bff✅ Approved
Reviewed at commit
4530bd8.Confidence: high.
Claimed by
merge_drive.py(pid 405719) until2026-06-10T13:12:59.715778+00:00.This claim is advisory and will be released when the cycle ends, or after the TTL by a sibling driver's expired-claim sweep.
4530bd8bffto1eca6b1da7Approved by the controller reviewer stage (workflow 368).