overkill 0.2.1__tar.gz → 0.3.0__tar.gz

{overkill-0.2.1 → overkill-0.3.0}/.gitignore

@@ -3,7 +3,9 @@
 
 # Legacy log directory from previous versions
 .ai-review-logs/
-# review-loop self-hosted install
+# overkill self-hosted install
+.overkill/
+# Legacy directory (pre-migration)
 .review-loop/
 
 # Python

overkill-0.2.1/.reviewlooprc.example → overkill-0.3.0/.overkillrc.example

@@ -1,5 +1,5 @@
-# .reviewlooprc — project-level configuration for review-loop.sh
-# Copy this file to your project root as ".reviewlooprc" and customize.
+# .overkillrc — project-level configuration for overkill review-loop
+# Place this file in .overkill/.overkillrc and customize.
 # CLI arguments always take precedence over these values.
 
 # Target branch to diff against (default: develop)

{overkill-0.2.1 → overkill-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: overkill
-Version: 0.2.1
+Version: 0.3.0
 Summary: AI-powered code review loop — automates Codex review + Claude fix cycles
 Project-URL: Repository, https://github.com/modocai/mr-overkill
 Project-URL: Issues, https://github.com/modocai/mr-overkill/issues
@@ -161,7 +161,7 @@ Examples:
 
 ## Usage: overkill init
 
-Initialize `.review-loop/` in a project directory. Safe to re-run — prompts are refreshed, user-edited configs are preserved.
+Initialize `.overkill/` in a project directory. Safe to re-run — prompts are refreshed, user-edited configs are preserved.
 
 ```
 overkill init [TARGET_DIR]   # default: current directory
@@ -170,9 +170,9 @@ overkill init [TARGET_DIR]   # default: current directory
 Creates:
 
 ```
-.review-loop/
+.overkill/
 ├── prompts/active/          # 10 prompt templates
-├── .reviewlooprc            # review-loop config
+├── .overkillrc              # review-loop config
 ├── .refactorsuggestrc       # refactor-suggest config
 ├── logs/                    # runtime logs
 │   └── refactor/            # refactor-suggest logs
@@ -207,9 +207,9 @@ Recommended workflow: start with `--dry-run` to review findings, then re-run wit
 
 ## Configuration
 
-After running `overkill init`, config files live in `.review-loop/`:
+After running `overkill init`, config files live in `.overkill/`:
 
-### .review-loop/.reviewlooprc
+### .overkill/.overkillrc
 
 ```bash
 TARGET_BRANCH="main"
@@ -220,9 +220,9 @@ REVIEWER_BACKEND="codex"    # or "claude"
 PROMPTS_DIR="./custom-prompts"
 ```
 
-See `.review-loop/.reviewlooprc` for all available options.
+See `.overkill/.overkillrc` for all available options.
 
-### .review-loop/.refactorsuggestrc
+### .overkill/.refactorsuggestrc
 
 ```bash
 SCOPE="auto"
@@ -243,7 +243,7 @@ PROMPTS_DIR="./custom-prompts"
 
 ```
 1. Check prerequisites (git, codex, claude, jq, envsubst, target branch)
-2. Create .review-loop/logs/ directory
+2. Create .overkill/logs/ directory
 3. Loop (iteration 1..N):
    a. Generate diff: git diff $TARGET...$CURRENT
    b. Empty diff → exit
@@ -258,14 +258,14 @@ PROMPTS_DIR="./custom-prompts"
    h. Push to remote (updates PR)
    i. Post review/fix/self-review summary as PR comment
    j. Next iteration reviews the updated committed state
-4. Write summary to .review-loop/logs/summary.md
+4. Write summary to .overkill/logs/summary.md
 ```
 
 ## Output Files
 
-All logs are git-ignored by default (inside `.review-loop/`).
+All logs are git-ignored by default (inside `.overkill/`).
 
-### review-loop logs (`.review-loop/logs/`)
+### review-loop logs (`.overkill/logs/`)
 
 | File | Description |
 |------|-------------|
@@ -277,7 +277,7 @@ All logs are git-ignored by default (inside `.review-loop/`).
 | `refix-N-M.md` | Claude re-fix log (iteration N, sub-iteration M) |
 | `summary.md` | Final summary with status and per-iteration results |
 
-### refactor-suggest logs (`.review-loop/logs/refactor/`)
+### refactor-suggest logs (`.overkill/logs/refactor/`)
 
 | File | Description |
 |------|-------------|
@@ -316,7 +316,7 @@ Go/no-go decision based on current usage percentage:
 
 ## Customizing Prompts
 
-Edit the templates in `.review-loop/prompts/active/`.
+Edit the templates in `.overkill/prompts/active/`.
 
 ### review-loop prompts
 
@@ -364,7 +364,7 @@ The loop terminates when any of these occur:
 
 ```bash
 # Quick — just nuke the directory
-rm -rf .review-loop
+rm -rf .overkill
 
 # Also remove the Python package
 pip uninstall overkill       # or: uv tool uninstall overkill / pipx uninstall overkill

{overkill-0.2.1 → overkill-0.3.0}/README.md

@@ -139,7 +139,7 @@ Examples:
 
 ## Usage: overkill init
 
-Initialize `.review-loop/` in a project directory. Safe to re-run — prompts are refreshed, user-edited configs are preserved.
+Initialize `.overkill/` in a project directory. Safe to re-run — prompts are refreshed, user-edited configs are preserved.
 
 ```
 overkill init [TARGET_DIR]   # default: current directory
@@ -148,9 +148,9 @@ overkill init [TARGET_DIR]   # default: current directory
 Creates:
 
 ```
-.review-loop/
+.overkill/
 ├── prompts/active/          # 10 prompt templates
-├── .reviewlooprc            # review-loop config
+├── .overkillrc              # review-loop config
 ├── .refactorsuggestrc       # refactor-suggest config
 ├── logs/                    # runtime logs
 │   └── refactor/            # refactor-suggest logs
@@ -185,9 +185,9 @@ Recommended workflow: start with `--dry-run` to review findings, then re-run wit
 
 ## Configuration
 
-After running `overkill init`, config files live in `.review-loop/`:
+After running `overkill init`, config files live in `.overkill/`:
 
-### .review-loop/.reviewlooprc
+### .overkill/.overkillrc
 
 ```bash
 TARGET_BRANCH="main"
@@ -198,9 +198,9 @@ REVIEWER_BACKEND="codex"    # or "claude"
 PROMPTS_DIR="./custom-prompts"
 ```
 
-See `.review-loop/.reviewlooprc` for all available options.
+See `.overkill/.overkillrc` for all available options.
 
-### .review-loop/.refactorsuggestrc
+### .overkill/.refactorsuggestrc
 
 ```bash
 SCOPE="auto"
@@ -221,7 +221,7 @@ PROMPTS_DIR="./custom-prompts"
 
 ```
 1. Check prerequisites (git, codex, claude, jq, envsubst, target branch)
-2. Create .review-loop/logs/ directory
+2. Create .overkill/logs/ directory
 3. Loop (iteration 1..N):
    a. Generate diff: git diff $TARGET...$CURRENT
    b. Empty diff → exit
@@ -236,14 +236,14 @@ PROMPTS_DIR="./custom-prompts"
    h. Push to remote (updates PR)
    i. Post review/fix/self-review summary as PR comment
    j. Next iteration reviews the updated committed state
-4. Write summary to .review-loop/logs/summary.md
+4. Write summary to .overkill/logs/summary.md
 ```
 
 ## Output Files
 
-All logs are git-ignored by default (inside `.review-loop/`).
+All logs are git-ignored by default (inside `.overkill/`).
 
-### review-loop logs (`.review-loop/logs/`)
+### review-loop logs (`.overkill/logs/`)
 
 | File | Description |
 |------|-------------|
@@ -255,7 +255,7 @@ All logs are git-ignored by default (inside `.review-loop/`).
 | `refix-N-M.md` | Claude re-fix log (iteration N, sub-iteration M) |
 | `summary.md` | Final summary with status and per-iteration results |
 
-### refactor-suggest logs (`.review-loop/logs/refactor/`)
+### refactor-suggest logs (`.overkill/logs/refactor/`)
 
 | File | Description |
 |------|-------------|
@@ -294,7 +294,7 @@ Go/no-go decision based on current usage percentage:
 
 ## Customizing Prompts
 
-Edit the templates in `.review-loop/prompts/active/`.
+Edit the templates in `.overkill/prompts/active/`.
 
 ### review-loop prompts
 
@@ -342,7 +342,7 @@ The loop terminates when any of these occur:
 
 ```bash
 # Quick — just nuke the directory
-rm -rf .review-loop
+rm -rf .overkill
 
 # Also remove the Python package
 pip uninstall overkill       # or: uv tool uninstall overkill / pipx uninstall overkill

{overkill-0.2.1 → overkill-0.3.0}/bin/lib/common.sh

@@ -277,7 +277,7 @@ _changed_files_since_snapshot() {
   _out=$(mktemp)
   _git_all_dirty_nul | while IFS= read -r -d '' _f; do
     [[ -n "$_f" ]] || continue
-    [[ "$_f" == .review-loop/logs/* ]] && continue
+    [[ "$_f" == .overkill/logs/* || "$_f" == .review-loop/logs/* ]] && continue
     if [[ -f "$_f" ]]; then
       _cur_hash=$(git hash-object "$_f" 2>/dev/null || echo UNHASHABLE)
       if [[ -x "$_f" ]]; then _cur_mode="100755"; else _cur_mode="100644"; fi

{overkill-0.2.1 → overkill-0.3.0}/prompts/active/claude-review.prompt.md

@@ -8,15 +8,23 @@ You are a code reviewer analyzing a proposed change.
 
 ## Instructions
 
-Run the following command to get the diff:
+Start by running the following command to get the diff:
 
 ```
 git diff ${TARGET_BRANCH}...${CURRENT_BRANCH}
 ```
 
+Then **actively explore the codebase** to understand context before flagging issues:
+
+- Use `Read` to examine surrounding code when a change touches a function or module.
+- Use `Grep` to check if a pattern, variable, or function is used elsewhere before calling it dead code or flagging a naming issue.
+- Use `Glob` to find related files (e.g. tests, configs) that may clarify intent.
+
 Review the diff according to the guidelines below.
 
-## Review Guidelines
+## Review Guidelines — HIGH SIGNAL ONLY
+
+We only want findings where you are confident the issue is real. **If you are not certain an issue is real, do not flag it.** False positives erode trust and waste reviewer time.
 
 1. Only flag issues the original author would fix if they knew about them.
 2. The issue must be **introduced by this diff** — do not flag pre-existing problems.
@@ -24,6 +32,16 @@ Review the diff according to the guidelines below.
 4. Do not flag trivial style issues unless they obscure meaning or violate documented standards.
 5. Do not speculate — you must identify the exact code location and explain why it is a problem.
 6. If this is iteration > 1, focus on whether issues from prior reviews have been properly fixed, and identify any new issues introduced by the fixes.
+7. **Only flag issues where your confidence is >= 0.85.** If you cannot reach that threshold after investigating, do not include the finding.
+
+### Do NOT flag (false positive filter)
+
+- Pre-existing issues not introduced by this diff.
+- Something that appears to be a bug but is actually correct when you read the surrounding code.
+- Pedantic nitpicks that a senior engineer would not flag in a real review.
+- Issues that a linter or formatter will catch — do not run the linter to verify.
+- Potential issues that depend on specific inputs or state that are not demonstrated in the diff.
+- General code quality concerns (e.g. lack of test coverage, general security hygiene) unless they are concretely exploitable.
 
 ## Priority Levels
 

{overkill-0.2.1 → overkill-0.3.0}/prompts/active/codex-review.prompt.md

@@ -20,10 +20,16 @@ Review the diff according to the guidelines below.
 
 1. Only flag issues the original author would fix if they knew about them.
 2. The issue must be **introduced by this diff** — do not flag pre-existing problems.
-3. The issue must be discrete, actionable, and concretely provable.
-4. Do not flag trivial style issues unless they obscure meaning or violate documented standards.
-5. Do not speculate — you must identify the exact code location and explain why it is a problem.
-6. If this is iteration > 1, focus on whether issues from prior reviews have been properly fixed, and identify any new issues introduced by the fixes.
+3. The issue must meaningfully impact the accuracy, performance, security, or maintainability of the code.
+4. The issue must be discrete, actionable, and concretely provable.
+5. Fixing the bug does not demand a level of rigor not present in the rest of the codebase (e.g. one doesn't need detailed comments and input validation in a repository of one-off scripts).
+6. The bug does not rely on unstated assumptions about the codebase or author's intent.
+7. It is not enough to speculate that a change may disrupt another part of the codebase — you must identify the other parts of the code that are provably affected.
+8. The bug is clearly not just an intentional change by the original author.
+9. Do not flag trivial style issues unless they obscure meaning or violate documented standards.
+10. If this is iteration > 1, focus on whether issues from prior reviews have been properly fixed, and identify any new issues introduced by the fixes.
+
+Output all findings that the original author would fix if they knew about them. If there is no finding that a person would definitely want to fix, prefer outputting no findings. Do not stop at the first qualifying finding — continue until you've listed every qualifying finding.
 
 ## Priority Levels
 
@@ -36,9 +42,12 @@ Review the diff according to the guidelines below.
 
 - One comment per distinct issue.
 - Brief (at most 1 paragraph body).
-- Clearly state the scenarios/inputs required for the bug to manifest.
-- Matter-of-fact tone, no flattery.
+- Clearly state the scenarios, environments, or inputs required for the bug to manifest.
+- Appropriately communicate severity — do not overclaim. Immediately indicate if the issue's severity depends on specific conditions.
+- The comment should be written such that the original author can immediately grasp the idea without close reading.
+- Matter-of-fact tone, no flattery. Avoid "Great job …", "Thanks for …".
 - Code snippets max 3 lines, wrapped in markdown code tags.
+- Use ```suggestion blocks only for concrete replacement code (minimal lines; no commentary inside the block). Preserve exact leading whitespace.
 
 ## Output Format
 

{overkill-0.2.1 → overkill-0.3.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "overkill"
-version = "0.2.1"
+version = "0.3.0"
 description = "AI-powered code review loop — automates Codex review + Claude fix cycles"
 readme = "README.md"
 license = { text = "MIT" }
@@ -36,7 +36,7 @@ packages = ["src/mr_overkill"]
 
 [tool.hatch.build.targets.wheel.force-include]
 "prompts/active" = "mr_overkill/data/prompts/active"
-".reviewlooprc.example" = "mr_overkill/data/.reviewlooprc.example"
+".overkillrc.example" = "mr_overkill/data/.overkillrc.example"
 ".refactorsuggestrc.example" = "mr_overkill/data/.refactorsuggestrc.example"
 
 [tool.pytest.ini_options]

{overkill-0.2.1 → overkill-0.3.0}/src/mr_overkill/__main__.py

@@ -24,7 +24,7 @@ def main() -> None:
         print(
             "Usage: overkill <command>\n\n"
             "Commands:\n"
-            "  init              Initialize .review-loop/ in a project\n"
+            "  init              Initialize .overkill/ in a project\n"
             "  review-loop       Run AI review-fix loop\n"
             "  refactor-suggest  Run AI refactoring suggestions\n"
             "  check-budget      Show token budget usage summary",
@@ -43,7 +43,7 @@ def main() -> None:
 
         p = argparse.ArgumentParser(
             prog="overkill init",
-            description="Initialize .review-loop/ in a project",
+            description="Initialize .overkill/ in a project",
         )
         p.add_argument("target", nargs="?", default=".", help="Target directory")
         init_args = p.parse_args()

{overkill-0.2.1 → overkill-0.3.0}/src/mr_overkill/cli.py

@@ -7,6 +7,7 @@ mirroring the CLI interface of the bash scripts.
 from __future__ import annotations
 
 import argparse
+import logging
 import re
 import subprocess
 from dataclasses import dataclass
@@ -59,18 +60,45 @@ def _load_rc_file(rc_name: str) -> dict[str, str]:
         return {}
 
     git_root = Path(result.stdout.strip())
-    rc_path = git_root / ".review-loop" / rc_name
+    rc_path = git_root / ".overkill" / rc_name
     if not rc_path.is_file():
-        # Fall back to the legacy repo-root location
-        legacy = git_root / rc_name
-        if legacy.is_file():
-            import logging
+        # Fall back to the legacy .review-loop/ location
+        legacy_dir = git_root / ".review-loop" / rc_name
+        # Also check for legacy .reviewlooprc name in .review-loop/
+        legacy_old_name = (
+            git_root / ".review-loop" / ".reviewlooprc"
+            if rc_name == ".overkillrc" else None
+        )
+        # Legacy repo-root name: .overkillrc was .reviewlooprc
+        legacy_root_name = ".reviewlooprc" if rc_name == ".overkillrc" else rc_name
+        legacy_root = git_root / legacy_root_name
+        if legacy_dir.is_file():
+            logging.getLogger(__name__).warning(
+                "%s found at .review-loop/ (legacy location). "
+                "Please run 'overkill init' to migrate to .overkill/",
+                rc_name,
+            )
+            rc_path = legacy_dir
+        elif legacy_old_name and legacy_old_name.is_file():
+            logging.getLogger(__name__).warning(
+                ".reviewlooprc found at .review-loop/ (legacy location). "
+                "Please run 'overkill init' to migrate to .overkill/.overkillrc",
+            )
+            rc_path = legacy_old_name
+        elif (git_root / rc_name).is_file():
             logging.getLogger(__name__).warning(
                 "%s found at repo root (legacy location). "
-                "Please move it to .review-loop/%s",
+                "Please move it to .overkill/%s",
                 rc_name, rc_name,
             )
-            rc_path = legacy
+            rc_path = git_root / rc_name
+        elif legacy_root.is_file():
+            logging.getLogger(__name__).warning(
+                "%s found at repo root (legacy location). "
+                "Please move it to .overkill/%s",
+                legacy_root_name, rc_name,
+            )
+            rc_path = legacy_root
         else:
             return {}
 
@@ -118,7 +146,33 @@ def _resolve_prompts_dir(prompts_dir: str) -> Path:
         check=False,
     )
     if result.returncode == 0:
-        return Path(result.stdout.strip()) / prompts_dir
+        git_root = Path(result.stdout.strip())
+        resolved = git_root / prompts_dir
+        # Legacy fallback: .overkill/... → .review-loop/...
+        if not resolved.is_dir() and ".overkill/" in prompts_dir:
+            legacy = git_root / prompts_dir.replace(
+                ".overkill/", ".review-loop/", 1
+            )
+            if legacy.is_dir():
+                logging.getLogger(__name__).warning(
+                    "Prompts dir found at %s (legacy location). "
+                    "Please run 'overkill init' to migrate to .overkill/",
+                    legacy,
+                )
+                return legacy
+        # Inverse fallback: .review-loop/... → .overkill/...
+        if not resolved.is_dir() and ".review-loop/" in prompts_dir:
+            migrated = git_root / prompts_dir.replace(
+                ".review-loop/", ".overkill/", 1
+            )
+            if migrated.is_dir():
+                logging.getLogger(__name__).warning(
+                    "Prompts dir migrated to %s. "
+                    "Please update PROMPTS_DIR in your rc file.",
+                    migrated,
+                )
+                return migrated
+        return resolved
     return p
 
 
@@ -249,7 +303,7 @@ def parse_review_loop_args(
     args = parser.parse_args(argv)
 
     # Load rc file defaults
-    rc = _load_rc_file(".reviewlooprc")
+    rc = _load_rc_file(".overkillrc")
 
     # Resolve values with precedence: CLI > rc file > defaults
     target = args.target or rc.get("TARGET_BRANCH", "develop")
@@ -292,7 +346,7 @@ def parse_review_loop_args(
     budget_scope_str = rc.get("BUDGET_SCOPE", "module")
 
     prompts_dir = _resolve_prompts_dir(
-        rc.get("PROMPTS_DIR", ".review-loop/prompts/active")
+        rc.get("PROMPTS_DIR", ".overkill/prompts/active")
     )
 
     current_branch = _detect_current_branch()
@@ -306,7 +360,18 @@ def parse_review_loop_args(
         check=False,
     )
     git_root = Path(result.stdout.strip()) if result.returncode == 0 else Path(".")
-    log_dir = git_root / ".review-loop" / "logs"
+    log_dir = git_root / ".overkill" / "logs"
+    if not log_dir.is_dir():
+        legacy_log = git_root / ".review-loop" / "logs"
+        if legacy_log.is_dir():
+            log_dir = legacy_log
+
+    # In mixed-state repos, fall back to legacy logs when resume metadata
+    # is missing from the preferred directory.
+    if args.resume and not (log_dir / "max-loop.txt").is_file():
+        legacy_log = git_root / ".review-loop" / "logs"
+        if (legacy_log / "max-loop.txt").is_file():
+            log_dir = legacy_log
 
     # Restore saved values on resume when not explicitly given
     if args.resume:
@@ -539,7 +604,7 @@ def parse_refactor_suggest_args(
     budget_scope_str = rc.get("BUDGET_SCOPE", "module")
 
     prompts_dir = _resolve_prompts_dir(
-        rc.get("PROMPTS_DIR", ".review-loop/prompts/active")
+        rc.get("PROMPTS_DIR", ".overkill/prompts/active")
     )
 
     current_branch = _detect_current_branch()
@@ -555,7 +620,18 @@ def parse_refactor_suggest_args(
         Path(result.stdout.strip()) if result.returncode == 0
         else Path(".")
     )
-    log_dir = git_root / ".review-loop" / "logs" / "refactor"
+    log_dir = git_root / ".overkill" / "logs" / "refactor"
+    if not log_dir.is_dir():
+        legacy_log = git_root / ".review-loop" / "logs" / "refactor"
+        if legacy_log.is_dir():
+            log_dir = legacy_log
+
+    # In mixed-state repos, fall back to legacy logs when resume metadata
+    # is missing from the preferred directory.
+    if args.resume and not (log_dir / "max-loop.txt").is_file():
+        legacy_log = git_root / ".review-loop" / "logs" / "refactor"
+        if (legacy_log / "max-loop.txt").is_file():
+            log_dir = legacy_log
 
     # Restore saved values on resume when not explicitly given
     if args.resume:

{overkill-0.2.1 → overkill-0.3.0}/src/mr_overkill/git_ops.py

@@ -100,12 +100,14 @@ def changed_files_since_snapshot(
     snapshot: list[WorktreeSnapshot],
     cwd: Path | None = None,
     exclude_prefix: str | None = None,
+    exclude_prefixes: tuple[str, ...] | None = None,
 ) -> list[str]:
     """Compare current dirty files against a snapshot.
 
     Returns list of file paths that changed (new, modified, or deleted).
     Uses a dict for O(1) lookups (vs O(n²) awk in bash).
     """
+    _prefixes = exclude_prefixes or ((exclude_prefix,) if exclude_prefix else ())
     snap_map: dict[str, tuple[str, str]] = {
         s.path: (s.file_hash, s.mode) for s in snapshot
     }
@@ -113,7 +115,7 @@ def changed_files_since_snapshot(
     changed: list[str] = []
 
     for f in git_all_dirty(cwd):
-        if exclude_prefix and f.startswith(exclude_prefix):
+        if _prefixes and any(f.startswith(p) for p in _prefixes):
             continue
 
         fpath = base / f
@@ -210,7 +212,8 @@ def commit_and_push(
     Raises ``RuntimeError`` if ``git commit`` or ``git push`` fails.
     """
     changed = changed_files_since_snapshot(
-        snapshot, cwd=cwd, exclude_prefix=".review-loop/logs/",
+        snapshot, cwd=cwd,
+        exclude_prefixes=(".overkill/logs/", ".review-loop/logs/"),
     )
     if not changed:
         logger.info("No file changes after fix — nothing to commit.")