fix(cli): per-language commands, honest scorecard tiers, tests-aware detection#20
Merged
Merged
Conversation
…detection
Post-v5 follow-ups so generated commands, the scorecard, and language
auto-detection are correct and consistent on multi-language repos.
Per-language commands (docs are `docs <lang> <subject>`, not `--doc`):
- thread the resolved language through the AI fix-prompt (`compose_prompt`),
the `check` cycle/metric `fix`, the scorecard next-step hint, the `docs`
index hints, and the HTML viewer's Prompt Generator — every emitted
`code-ranker docs …` / `report --prompt …` now names the language.
- `resolve_language_snap` returns the resolved language key (was snapshot only).
- `docs <lang> <subject>` localizes `<lang>` placeholders in served docs to the
concrete language (`docs rust hk`, `--plugins rust`); `base` keeps the generic
placeholder. Compact one-line `docs <lang> ai` catalog.
- templates carry a `{lang}` placeholder (prompt.md, builtin.toml remediation).
Scorecard `--focus <metric>` tiers are now honest: each ranked module shows its
own tier for that metric (`warn`/`info`/`—`) instead of a blanket `warn`; the
header/next-step hint name the resolved language, not the first plugin.
Auto-detect honors `[ignore] tests`: marker-less, extension-detected plugins
(c/cpp/csharp/markdown) walk with the same tests/gitignore filters analysis uses,
so a project whose only such files are test fixtures is no longer detected then
warned about as empty.
Docs synced (CLI.md, DESIGN.md, templates.md, ai-skill.md, USE-CASES.md,
customization/README.md, plugins/base/{AI,HK}.md, contrib) + sample goldens.
No format/CLI-surface contract changed, so no version bump.
Claude-Session: https://claude.ai/code/session_01Jcdvq3iTsqk74KfrXxZzcP
|
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 (2)
✅ Files skipped from review due to trivial changes (1)
📝 WalkthroughWalkthroughThreads a resolved language key through docs rendering, prompt composition, scorecard output, remediation strings, fixtures, tests, and documentation so ChangesLanguage parameter threading
ignore_tests propagation into plugin detection
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #20 +/- ##
=======================================
Coverage 98.36% 98.37%
=======================================
Files 128 128
Lines 16034 16079 +45
=======================================
+ Hits 15772 15817 +45
Misses 262 262 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
code-ranker View diff report ↗rust: ok
baseline main @f79d519 2026-06-28 17:14 UTC · updated 2026-06-28 21:43 UTC |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (2)
crates/code-ranker-cli/src/recommend_test.rs (1)
944-945: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winKeep asserting the canonical language key here.
resolve_language_snapnow returns the resolved key that prompt/docs output uses. Dropping it from this alias test means a regression from canonicalpythonback to rawpywould still pass here.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@crates/code-ranker-cli/src/recommend_test.rs` around lines 944 - 945, The alias test in resolve_language_snap is only checking the returned snap and currently drops the resolved language key, which could miss a regression back to raw py. Update the assertion in recommend_test.rs to keep validating the canonical resolved key from resolve_language_snap for the py alias, alongside the existing SRP principle check, so the test fails if the returned key is not python.crates/code-ranker-cli/src/docs_test.rs (1)
57-57: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winAssert the rendered
docs <lang>command text, not just pass the arg.These tests now call the lang-aware helpers, but they never verify the language-scoped header each function renders. A regression that drops
{lang}from the command hint would still pass.Also applies to: 123-123, 133-141
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@crates/code-ranker-cli/src/docs_test.rs` at line 57, The docs tests using render_category, render_name, and render_spec should assert the full rendered docs command text including the language-scoped header, not just pass "rust" as an argument. Update the affected assertions to verify the output contains the expected "docs <lang>" hint from the lang-aware helpers, so a regression that drops the language placeholder would fail. Use the existing render_category and related test cases in docs_test.rs to locate the checks that need strengthening.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@contrib/prompting-self-improve.md`:
- Around line 154-159: Update the run example in the self-improve prompt so it
stays language-scoped: the “BEFORE” step currently uses bare code-ranker report
., which can rely on auto-detection and pick the wrong prompt/artifacts in
multi-language repos. Change the example to use the same language selector as
the surrounding flow, referencing the code-ranker report invocation in the run
steps and keeping it aligned with the earlier code-ranker docs <lang> ai / docs
<lang> <FOCUS> commands.
In `@crates/code-ranker-cli/src/docs.rs`:
- Around line 415-441: The new docs headers in render_metrics_index,
render_principles_index, and render_category are formatting lang directly, which
bypasses the placeholder-preserving behavior expected for the base locale.
Update these helpers to use the same localized language value that
localize_lang() provides for base so the rendered commands keep `<lang>` generic
in the base docs flow, while still showing the real language for non-base
locales.
---
Nitpick comments:
In `@crates/code-ranker-cli/src/docs_test.rs`:
- Line 57: The docs tests using render_category, render_name, and render_spec
should assert the full rendered docs command text including the language-scoped
header, not just pass "rust" as an argument. Update the affected assertions to
verify the output contains the expected "docs <lang>" hint from the lang-aware
helpers, so a regression that drops the language placeholder would fail. Use the
existing render_category and related test cases in docs_test.rs to locate the
checks that need strengthening.
In `@crates/code-ranker-cli/src/recommend_test.rs`:
- Around line 944-945: The alias test in resolve_language_snap is only checking
the returned snap and currently drops the resolved language key, which could
miss a regression back to raw py. Update the assertion in recommend_test.rs to
keep validating the canonical resolved key from resolve_language_snap for the py
alias, alongside the existing SRP principle check, so the test fails if the
returned key is not python.
🪄 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: e4ec7ea4-02ef-4db4-aaf5-9fbd7058eece
📒 Files selected for processing (37)
contrib/prompting-self-improve.mdcrates/code-ranker-cli/src/config/rules.rscrates/code-ranker-cli/src/docs.rscrates/code-ranker-cli/src/docs_test.rscrates/code-ranker-cli/src/recommend.rscrates/code-ranker-cli/src/recommend/prompt.rscrates/code-ranker-cli/src/recommend/scorecard.rscrates/code-ranker-cli/src/recommend_test.rscrates/code-ranker-cli/src/report.rscrates/code-ranker-cli/src/templates.rscrates/code-ranker-cli/src/templates_test.rscrates/code-ranker-graph/metrics/builtin.tomlcrates/code-ranker-graph/metrics/prompt.mdcrates/code-ranker-graph/src/builtin_test.rscrates/code-ranker-plugins/src/languages/c/mod.rscrates/code-ranker-plugins/src/languages/c/tests/sample/code-ranker-report.jsoncrates/code-ranker-plugins/src/languages/cfamily/mod.rscrates/code-ranker-plugins/src/languages/cpp/mod.rscrates/code-ranker-plugins/src/languages/cpp/tests/sample/code-ranker-report.jsoncrates/code-ranker-plugins/src/languages/csharp/mod.rscrates/code-ranker-plugins/src/languages/csharp/structure.rscrates/code-ranker-plugins/src/languages/csharp/tests/sample/code-ranker-report.jsoncrates/code-ranker-plugins/src/languages/go/tests/sample/code-ranker-report.jsoncrates/code-ranker-plugins/src/languages/js/tests/sample/code-ranker-report.jsoncrates/code-ranker-plugins/src/languages/md/tests/sample/code-ranker-report.jsoncrates/code-ranker-plugins/src/languages/python/tests/sample/code-ranker-report.jsoncrates/code-ranker-plugins/src/languages/rust/tests/sample/code-ranker-report.jsoncrates/code-ranker-plugins/src/languages/ts/tests/sample/code-ranker-report.jsoncrates/code-ranker-viewer/src/assets/export-popup.jsdocs/DESIGN.mddocs/ai-skill.mddocs/code-ranker-cli/CLI.mddocs/code-ranker-cli/USE-CASES.mddocs/customization/README.mddocs/templates.mdplugins/base/AI.mdplugins/base/HK.md
… hints generic - contrib: the BEFORE-snapshot step pinned `--plugins <lang>` (was bare `report .`, could auto-detect the wrong language on a multi-language repo). - docs index hints (`docs <lang> metrics/principles/<category>`) print a concrete language for a real plugin but keep the generic `<lang>` for `base`, matching `localize_lang` (via a small `command_lang` helper). Claude-Session: https://claude.ai/code/session_01Jcdvq3iTsqk74KfrXxZzcP
… of command_lang The CodeRabbit fix added a `command_lang` helper (+6 SLOC) that pushed docs.rs to 403 > the repo's own 400 sloc gate, failing the dogfood self-check in CI. Drop the helper: index hints now carry the literal `<lang>` token and are localized once at emit time via `localize_lang` (concrete language for a real plugin, generic for `base`) — same behavior, one substitution point, back under the gate. Claude-Session: https://claude.ai/code/session_01Jcdvq3iTsqk74KfrXxZzcP
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.
Post-v5 follow-ups so generated commands, the scorecard, and language
auto-detection are correct and consistent on multi-language repos. No
format/CLI-surface contract changed → no version bump.
Per-language commands
Docs are
docs <lang> <subject>now (the old--docflag is gone), but severalgenerated/runnable commands still omitted the language — broken on multi-language
repos. Fixed across the board:
compose_prompt), thecheckcycle/metricfix, the scorecardnext-step hint, the
docsindex hints, and the HTML viewer's Prompt Generatornow all name the resolved language (
code-ranker docs rust hk,report --plugins rust --prompt hk).resolve_language_snapreturns the resolved language key (was snapshot only).docs <lang> <subject>localizes<lang>placeholders in served docs to theconcrete language (
docs rust hk,--plugins rust);basekeeps the genericplaceholder. Compact one-line
docs <lang> aicatalog.{lang}placeholder (prompt.md,builtin.tomlremediation).ai-skill.md,USE-CASES.md,CLI.md,templates.md,customization/README.md,plugins/base/{AI,HK}.md,contrib); fixed stale--plugin/plugin = "…"in the AI playbook.Honest scorecard tiers
report --output.scorecard --focus <metric>tagged every ranked modulewarnregardless of threshold. Now each row shows its own tier for that metric
(
warn> warning,info> info,—under both / no threshold). Also the headerand next-step hint name the resolved language, not the alphabetically-first plugin.
Tests-aware auto-detection
Marker-less, extension-detected plugins (
c/cpp/csharp/markdown) now walkwith the same
[ignore] tests/ gitignore filters analysis uses. A repo whose onlysuch files are test fixtures is no longer auto-detected and then warned about as
"produced no nodes".
Validation
make all: PASS (build, test, clippy, lychee + markdownlint, self-check,coverage 96.97% ≥ 90%)
make e2e: PASS (62/62)scorecard_focus_metric_tags_each_module_by_its_actual_tier,localize_lang_substitutes_concrete_language_but_not_base, plus updatedrecommend/docs/templates/builtin tests and the 9 sample goldens.
https://claude.ai/code/session_01Jcdvq3iTsqk74KfrXxZzcP
Summary by CodeRabbit
New Features
code-ranker docs <lang> <subject>/code-ranker docs <lang> ai.Bug Fixes
Documentation
--plugins <lang>command shapes.