@zephyrdeng/pi-review 0.17.2 → 0.17.3
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json
CHANGED
|
@@ -171,6 +171,12 @@ Screening accumulates two kinds of state under the pi-review dir (siblings of `c
|
|
|
171
171
|
- **`screen-patterns.json`** — user-declared patterns merged over the builtin catalog: `{ "patterns": { "<id>": { "title", "severity" (critical|major|minor), "category" (correctness|security|data-loss|performance), "recommendation" } }, "disabled": ["<id>"] }`. An id matching a builtin overrides it; `disabled` retires entries that misfire on this codebase. `PI_REVIEW_SCREEN_PATTERNS=<path>` loads one extra file last (e.g. a project catalog committed to the repo — that is how a team shares patterns).
|
|
172
172
|
- **`screen-memory.jsonl`** — every flagged hunk is appended with a code hash; `pi-review screen-memory` aggregates it into hit frequencies and recurring **unmatched signals** (catch-all fired, no catalog hit). Those are the evidence for new patterns — promotion is a human/agent edit of `screen-patterns.json`, screen never rewrites its own catalog. `PI_REVIEW_SCREEN_MEMORY=0` disables recording; `PI_REVIEW_SCREEN_MEMORY_FILE` moves the log.
|
|
173
173
|
|
|
174
|
+
A finding whose summary ends `matched no known pattern` is an unmatched signal: escalate that file to a full `pi-review` run, and the hunk is already in `screen-memory.jsonl`. Reach for `pi-review screen-memory` when:
|
|
175
|
+
|
|
176
|
+
- The same unmatched signal recurs (`seen` ≥ 2) — promote it into `screen-patterns.json`.
|
|
177
|
+
- The user reports screen missing a defect class — check `Unmatched` for the accumulated candidates.
|
|
178
|
+
- A pattern misfires on this codebase — retire it via `disabled` instead of deleting history.
|
|
179
|
+
|
|
174
180
|
## CLI arguments: when to use each
|
|
175
181
|
|
|
176
182
|
| Argument | When to use |
|