@mechanai/deepreview 2.4.2 → 2.5.0

package/.opencode/agents/deepreview-applier.md

@@ -6,10 +6,17 @@ permission:
   edit: allow
   bash:
     "git diff*": allow
+    "mise tasks ls*": allow
     "mise run fmt*": allow
     "mise run lint*": allow
     "mise run check*": allow
     "mise run test*": allow
+    "npm run format*": allow
+    "npm run lint*": allow
+    "npm run test*": allow
+    "make fmt*": allow
+    "make lint*": allow
+    "make test*": allow
     "*": deny
 ---
 
@@ -44,13 +51,25 @@ If a fix cannot be applied (file doesn't exist, code doesn't match what was expe
 
 ## Verification (after all fixes are applied)
 
-After applying all fixes, run verification if `mise.toml` exists in the project root:
+After applying all fixes, run verification using the project's configured tooling.
 
-1. Run `mise run fmt` (auto-fix formatting — this is expected to modify files)
-2. Run `mise run lint` or `mise run check` (whichever exists)
-3. Run `mise run test`
+**Determining commands:** Check `AGENTS.md` (or `CLAUDE.md`) in the project root for explicit format/lint/test commands. If found, use those. Otherwise, auto-detect:
 
-If lint/check/test fails:
+- `mise.toml` exists: run `mise tasks ls`, then `mise run <task>` for tasks matching `fmt`/`format`, `lint`/`check`, `test`
+- `package.json` exists with matching scripts: `npm run format`, `npm run lint`, `npm run test`
+- `Makefile` exists with matching targets: `make fmt`, `make lint`, `make test`
+
+If no commands are found and no config files exist, skip verification and report `VERIFICATION: SKIPPED — no fmt/lint/test commands found`.
+
+**Steps:**
+
+1. **Format:** Run the format command (expected to modify files)
+2. **Lint:** Run the lint command
+   - If lint fails, fix errors **only in files you modified** this session. Do not fix pre-existing lint issues.
+   - Re-run format then lint. Attempt up to 2 fix-and-recheck cycles; if lint still fails, stop and report remaining errors.
+3. **Test:** Run the test command
+
+If lint/check/test still fails:
 
 - Include the error output in your response
 - Mark the relevant fix as FAILED with the error

package/README.md

@@ -83,6 +83,35 @@ its own context, keeping token usage minimal.
 - `git`
 - `gh` CLI (only for PR commands)
 
+## Configuration
+
+### Verification (formatting, linting, tests)
+
+After applying fixes, the applier agent runs formatting, linting, and tests. It auto-detects
+commands based on what exists in your project root:
+
+| File detected  | Format           | Lint                               | Test            |
+| -------------- | ---------------- | ---------------------------------- | --------------- |
+| `mise.toml`    | `mise run fmt`   | `mise run lint` / `mise run check` | `mise run test` |
+| `package.json` | `npm run format` | `npm run lint`                     | `npm run test`  |
+| `Makefile`     | `make fmt`       | `make lint`                        | `make test`     |
+
+If your project uses different commands (e.g., `cargo fmt`, `ruff check --fix`),
+specify them in `AGENTS.md`. The applier looks for commands labeled **Format**, **Lint**, and
+**Test** (or similar). For example:
+
+```markdown
+- **Format:** `cargo fmt`
+- **Lint:** `cargo clippy -- -D warnings`
+- **Tests:** `cargo test`
+```
+
+The applier checks `AGENTS.md` (or `CLAUDE.md`) first, falling back to auto-detection.
+If no commands are found and no config files exist, verification is skipped.
+
+When lint fails, the applier attempts to fix errors in the files it modified (up to 2 retry
+cycles) before reporting the failure.
+
 ## Upgrade
 
 OpenCode caches plugins on first install and does not automatically check for newer versions.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mechanai/deepreview",
-  "version": "2.4.2",
+  "version": "2.5.0",
   "description": "Multi-agent parallel code/spec review for OpenCode",
   "license": "MIT",
   "repository": {