Merge PR #12: langfuse evaluation layer

Adds the evaluation layer on top of the review traces: five deterministic
scores describing how the reviewer behaved, a bridge that turns human reactions
into ground truth, and a dataset seeded from the reviews already run.

The two are kept apart on purpose. feedback.db has recorded 113 reviews and
zero reactions, resolutions or replies — nobody has ever responded to a bot
comment — so an accuracy metric cannot be built yet. The scorers therefore
measure behaviour, which is computable from data in hand, and feedback_scores
turns verdicts into scores the moment any arrive.

eval_scores.py emits finding_rate, severity_info_ratio, severity_max,
dropped_findings and cost_per_finding into the same ingestion batch as the
trace. Undefined values are omitted rather than reported as zero: an info ratio
over a silent review is undefined, and charting it as 0 would read as perfect
calibration.

dropped_findings needed a parser change. Both parsers silently discard findings
with an unusable path/line, which made a model emitting garbage locations
indistinguishable from one that found nothing. last_parse_dropped() exposes the
delta, read at parse time — after apply_repo_config the drops are the config
working as intended, not the model misbehaving.

feedback_scores.py scores the session ("{repo}#{pr}"), because feedback arrives
days later against a PR and nothing records which re-run produced which
comment. review_acceptance is absent rather than 0 when nothing was engaged.

eval_bootstrap.py registers the score configs, seeds the pragent-reviews
dataset, and can backfill scores onto traces that predate the scorers.
expectedOutput is the reviewer's own prior output, flagged
labelled_by_human: false — a regression baseline, not verified truth.

Also fixes a silent telemetry failure: the ingestion endpoint answers 207 when
only some events succeed, so a batch with every event rejected still looked
like success. Score events were missing the required per-event timestamp and
ingested nothing while reporting 207. _warn_on_rejected_events now logs the
per-event errors under LANGFUSE_DEBUG.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

pilot/README-eval.md

@@ -0,0 +1,123 @@
+# Evaluation — scorers, ground truth, and the dataset
+
+Langfuse already receives one trace per review (`README-langfuse.md`). This is
+the layer on top: numbers attached to those traces that say how the reviewer
+*behaved*, and the beginnings of a ground-truth signal that says whether it was
+*right*.
+
+Those two things are deliberately kept apart, because only one of them exists
+yet.
+
+## What could and could not be built
+
+`feedback.db` has recorded 113 reviews across 4 repos. It has recorded **zero**
+reactions, zero thread resolutions and zero replies. The harvester, the schema
+and the daily analyzer are all working; nobody has ever reacted to a bot
+comment.
+
+That rules out an accuracy metric today. Correctness needs labels, and a
+judge scored against no labels is theatre. So the scorers here measure
+behaviour, which is computable from data already in hand, and a separate
+bridge exists to turn human reactions into scores the moment any arrive.
+
+## The five behavioural scores
+
+Emitted with every review by `eval_scores.py`, folded into the same ingestion
+batch as the trace so they cost no extra request.
+
+| score | type | what a change in it means |
+|---|---|---|
+| `finding_rate` | NUMERIC | Findings posted. 0 is the restraint case — good on clean code, a failure when the run degraded. Only the rate over time separates those. |
+| `severity_info_ratio` | NUMERIC 0–1 | Share of findings the model rated `info`/`trivial`. Rising = the model is hedging rather than committing. `None` when the review was silent: a ratio over an empty set is undefined, and charting it as 0 would read as perfect calibration. |
+| `severity_max` | CATEGORICAL | Highest severity surfaced, `none` when silent. Categorical because "did this ever surface something serious" is the real question, and a mean of severity ranks answers nothing. |
+| `dropped_findings` | NUMERIC | Findings the model emitted that the parser rejected for an unusable `path`/`line`. This is the only score here that measures the model's raw output. |
+| `cost_per_finding` | NUMERIC | Equivalent USD per finding. A cheaper model that finds nothing is not cheaper. |
+
+### Why `dropped_findings` needed a change to the parser
+
+`parse_findings` and `parse_review_output` discard any finding with a missing or
+unusable location. That happens silently, so a model emitting ten findings at
+invalid locations was indistinguishable from a model that found nothing — both
+produce an empty list. `ai_review.last_parse_dropped()` exposes the delta,
+recorded at parse time.
+
+It must be read at parse time specifically: by the time findings reach
+`_emit_langfuse`, `apply_repo_config` has already filtered them by
+`severity_threshold` and `max_findings`, and those drops are the config working
+as intended, not the model misbehaving.
+
+## Ground truth: `feedback_scores.py`
+
+Turns `feedback.db` into two session-level scores, keyed on `"{repo}#{pr}"`
+(which is what `langfuse_trace` already sets as `sessionId`).
+
+| score | meaning |
+|---|---|
+| `review_engagement` | Share of a PR's findings that drew any human reaction, resolution or reply. **Watch this first** — every quality number is vapour until it moves off 0. |
+| `review_acceptance` | Net verdict over engaged findings, −1 to +1. Absent, not 0, when nothing was engaged: zero would claim humans judged the review neutral, when the truth is nobody looked. |
+
+Session-level rather than trace-level because feedback arrives days later
+against a PR, and nothing in `feedback.db` records which re-run of the reviewer
+produced which comment. The session is both the available join and the honest
+granularity.
+
+Score ids are `uuid5(namespace, repo#pr#name)`, so the daily backfill updates
+rather than duplicates.
+
+## The dataset
+
+`pragent-reviews`, one item per PR the reviewer has run on, seeded by
+`eval_bootstrap.py` from `feedback.db`.
+
+`expectedOutput` is **the reviewer's own prior output**, not human-verified
+truth — every item carries `metadata.labelled_by_human: false`. Read it as a
+regression baseline: re-run a candidate model over these PRs and the diff
+against this column is the behaviour change. Promoting an item to real ground
+truth means a human editing it in the dataset view after re-reading the PR.
+
+## Running it
+
+```bash
+# once per project: score configs + dataset (+ score historical traces)
+python3 pilot/eval_bootstrap.py --db /data/feedback.db --backfill-traces
+
+# ship feedback verdicts (runs daily from the feedback CronJob)
+python3 pilot/feedback_scores.py --db /data/feedback.db
+```
+
+Both need `LANGFUSE_HOST`, `LANGFUSE_PUBLIC_KEY`, `LANGFUSE_SECRET_KEY`. In
+cluster they come from the `pragent-langfuse` Secret and point at the ClusterIP
+— never the NodePort, whose oauth2-proxy 302s ingestion to Logto and drops it.
+
+## Gotcha: HTTP 207 is not success
+
+The ingestion endpoint answers `207 Multi-Status` when *some* events failed, so
+a batch where **every** event was rejected still returns 207. An early version
+of these scorers omitted the required per-event `timestamp` and silently
+ingested nothing while reporting success. `langfuse_trace._warn_on_rejected_events`
+now logs the per-event errors under `LANGFUSE_DEBUG=1`. If scores are missing,
+check that before anything else.
+
+## What the first run showed
+
+Backfilled over 42 existing traces and 13 PRs:
+
+```
+cost_per_finding     n=42  mean=0.3133  min=0.0880  max=0.9042
+finding_rate         n=42  mean=0.4762  min=0.0000  max=4.0000
+severity_info_ratio  n=14  mean=0.0000
+review_engagement    n=14  mean=0.0000
+severity_max         {none: 28, medium: 11, high: 1, critical: 2}
+```
+
+Two things worth keeping:
+
+- **The reviewer is not info-heavy.** `feedback.db` shows 61 of 62 findings at
+  `INFO`, which looked like a badly calibrated model. It is not: `severity_max`
+  reads `medium`/`high`/`critical` on every trace that found anything, and
+  `severity_info_ratio` is flat 0. The `INFO` in the DB comes from
+  `feedback_harvest._parse_severity`, which defaults to `INFO` when its regex
+  misses the severity badge in the rendered comment. The DB severity is a
+  re-parse artifact; the score reads the model's structured output directly.
+- **28 of 42 reviews found nothing** (67%), and **engagement is flat zero**. The
+  first is not yet interpretable without the second.

pilot/ai_review.py

@@ -678,6 +678,19 @@ def _strip_path_prefix(p: str) -> str:
 # ---------------------------------------------------------------------------
 
 
+# How many raw findings the last `parse_review_output` / `parse_findings` call
+# rejected for an unusable path/line. A side channel rather than a return value
+# because both parsers already return fixed-width tuples that several callers
+# and their tests unpack positionally; widening them to carry a telemetry
+# number would be a breaking change for a fail-open signal.
+_LAST_PARSE_DROPPED: dict[str, int] = {"n": 0}
+
+
+def last_parse_dropped() -> int:
+    """Findings the last parse discarded. Read it immediately after parsing."""
+    return int(_LAST_PARSE_DROPPED.get("n") or 0)
+
+
 def _normalize_finding(f: dict) -> dict | None:
     """Validate + normalize one raw finding dict. Returns None if it's unusable
     (missing path/line). Normalises severity, keeps `reference` (default "")."""
@@ -754,6 +767,7 @@ def parse_findings(text: str) -> list[dict]:
     Also accepts a bare JSON array as the outer value: ``[{...}, {...}]`` —
     some agents skip the ``{"summary":..., "findings":[...]}`` wrapper.
     """
+    _LAST_PARSE_DROPPED["n"] = 0
     data = _parse_json_tolerant(text)
     if isinstance(data, dict):
         findings = data.get("findings")
@@ -768,6 +782,7 @@ def parse_findings(text: str) -> list[dict]:
         n = _normalize_finding(f)
         if n is not None:
             out.append(n)
+    _LAST_PARSE_DROPPED["n"] = len(findings) - len(out)
     return out
 
 
@@ -823,6 +838,7 @@ def parse_review_output(
     block), with a tolerant fallback that scans for the last balanced
     object/array in the prose tail. Never raises.
     """
+    _LAST_PARSE_DROPPED["n"] = 0
     blob = _last_json_block(text)
     if blob is None:
         return "", [], [], [], [], "", ""
@@ -856,6 +872,12 @@ def parse_review_output(
             n = _normalize_finding(f)
             if n is not None:
                 out.append(n)
+        # A model that emits findings at unusable locations is indistinguishable
+        # from one that found nothing, because both end up with an empty `out`.
+        # Stash the delta so the caller can score it (see `eval_scores`).
+        _LAST_PARSE_DROPPED["n"] = len(findings_raw) - len(out)
+    else:
+        _LAST_PARSE_DROPPED["n"] = 0
     return summary, out, summary_changes, risks, walkthrough, risk_verdict, test_coverage
 
 
@@ -2080,6 +2102,7 @@ def _emit_langfuse(
     summary: str,
     engine: str,
     config: dict | None = None,
+    dropped_count: float | None = None,
 ) -> None:
     """Ship this review's usage to Langfuse, if one is configured.
 
@@ -2105,7 +2128,7 @@ def _emit_langfuse(
             repo=repo, index=index, sha=sha, title=title, model=model,
             usage=usage, findings=findings, summary=summary or "",
             engine=engine, lenses=(usage or {}).get("lenses"),
-            price_target=price_target,
+            price_target=price_target, dropped_count=dropped_count,
         )
     except Exception as e:
         print(f"pragent: langfuse emit skipped: {e}", file=sys.stderr)
@@ -2235,6 +2258,7 @@ def review_pr(
                     additional_context=additional_context,
                 )
             review_summary, findings, summary_changes, risks, _walkthrough, _risk_verdict, _test_coverage = parse_review_output(stdout)
+            parse_dropped = last_parse_dropped()
             if not findings and not review_summary:
                 # The findings JSON was missing or malformed. Don't discard the
                 # run: salvage the prose, keep the usage report (the tokens were
@@ -2255,12 +2279,14 @@ def review_pr(
                     repo=repo, index=index, sha=sha, title=title,
                     model=display_model, usage=usage, findings=[],
                     summary=salvaged, engine=engine, config=config,
+                    dropped_count=parse_dropped,
                 )
                 return True
         else:
             user_prompt = build_user_prompt(title, body + compression_note, diff, config, prior, additional_context)
             raw_findings = call_model(ollama_url, model, SYSTEM_PROMPT, user_prompt, max_tokens)
             findings = parse_findings(raw_findings)
+            parse_dropped = last_parse_dropped()
             usage = None
 
         # Filter / cap findings per `.pr-review.json` (style, threshold, max,
@@ -2343,6 +2369,7 @@ def review_pr(
             repo=repo, index=index, sha=sha, title=title,
             model=display_model, usage=usage, findings=findings,
             summary=review_summary, engine=engine, config=config,
+            dropped_count=parse_dropped,
         )
         print(
             f"pragent: reviewed {repo}#{index} sha={sha[:8]} "

pilot/eval_bootstrap.py

@@ -0,0 +1,299 @@
+#!/usr/bin/env python3
+"""pragent pilot — one-time Langfuse project setup for evaluation.
+
+Three jobs, each idempotent so it can be re-run after any change:
+
+  1. **Score configs.** Registers the schema for every score pragent emits
+     (`eval_scores.SCORE_CONFIGS` + `feedback_scores.SCORE_CONFIGS`). Without
+     these the scores still ingest, but nothing stops a later scorer writing
+     `severity_max="HIGH"` beside today's `"high"` and quietly splitting one
+     series into two. Configs are immutable in Langfuse — a name that already
+     exists is left alone rather than updated.
+
+  2. **Dataset.** Seeds `pragent-reviews` from `feedback.db`: one item per PR
+     the reviewer has actually run on, carrying the repo/PR/sha as input and
+     the findings it posted as `expectedOutput`.
+
+     Read `expectedOutput` here as "what the reviewer said last time", not "what
+     is correct" — no human has labelled any of it. It is a regression baseline:
+     re-run a candidate model over these PRs and the diff against this column is
+     the behaviour change. Promoting an item to real ground truth means a human
+     editing it after reviewing the PR, which is what the dataset view is for.
+
+  3. **Trace backfill** (`--backfill-traces`). Scores only ride along with new
+     reviews, so without this the charts stay empty until the next PR lands.
+     Every trace `langfuse_trace` has ever written already carries the finding
+     count, the severity histogram and the cost in its metadata, which is
+     everything four of the five scorers need. `dropped_findings` is absent from
+     historical traces and is left unscored rather than backfilled as zero.
+
+  4. **Reports** what it found, so the gap between "reviews recorded" and
+     "reviews with human feedback" is visible rather than assumed.
+
+Usage:
+    LANGFUSE_HOST=... LANGFUSE_PUBLIC_KEY=... LANGFUSE_SECRET_KEY=... \\
+        python3 eval_bootstrap.py --db /data/feedback.db
+"""
+from __future__ import annotations
+
+import argparse
+import base64
+import json
+import os
+import sqlite3
+import sys
+import urllib.error
+import urllib.request
+
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+
+import eval_scores  # noqa: E402
+import feedback_scores  # noqa: E402
+
+DATASET_NAME = "pragent-reviews"
+
+
+def _conf() -> tuple[str, str, str]:
+    host = (os.environ.get("LANGFUSE_HOST") or "").strip().rstrip("/")
+    pk = (os.environ.get("LANGFUSE_PUBLIC_KEY") or "").strip()
+    sk = (os.environ.get("LANGFUSE_SECRET_KEY") or "").strip()
+    if not host or not pk or not sk:
+        raise SystemExit("LANGFUSE_HOST / LANGFUSE_PUBLIC_KEY / LANGFUSE_SECRET_KEY must be set")
+    return host, pk, sk
+
+
+def _call(method: str, path: str, body: dict | None = None, timeout: float = 20.0):
+    host, pk, sk = _conf()
+    auth = base64.b64encode(f"{pk}:{sk}".encode()).decode("ascii")
+    data = json.dumps(body).encode() if body is not None else None
+    req = urllib.request.Request(
+        host + path,
+        data=data,
+        headers={
+            "Content-Type": "application/json",
+            "Authorization": f"Basic {auth}",
+            "User-Agent": "pragent-pilot/1.0",
+        },
+        method=method,
+    )
+    try:
+        with urllib.request.urlopen(req, timeout=timeout) as resp:
+            raw = resp.read()
+            return resp.status, (json.loads(raw) if raw else None)
+    except urllib.error.HTTPError as e:
+        return e.code, e.read()[:400].decode("utf-8", "replace")
+
+
+# ---------------------------------------------------------------------------
+# 1. Score configs
+# ---------------------------------------------------------------------------
+
+def ensure_score_configs() -> dict:
+    status, existing = _call("GET", "/api/public/score-configs?limit=100")
+    have = set()
+    if status == 200 and isinstance(existing, dict):
+        have = {c.get("name") for c in existing.get("data", [])}
+
+    created, skipped, failed = [], [], []
+    for cfg in list(eval_scores.SCORE_CONFIGS) + list(feedback_scores.SCORE_CONFIGS):
+        if cfg["name"] in have:
+            skipped.append(cfg["name"])
+            continue
+        st, resp = _call("POST", "/api/public/score-configs", cfg)
+        if st in (200, 201):
+            created.append(cfg["name"])
+        else:
+            failed.append({"name": cfg["name"], "status": st, "error": resp})
+    return {"created": created, "already_present": skipped, "failed": failed}
+
+
+# ---------------------------------------------------------------------------
+# 2. Dataset from recorded reviews
+# ---------------------------------------------------------------------------
+
+def read_review_items(db_path: str) -> list[dict]:
+    """One dataset item per (repo, pr) the reviewer has run on.
+
+    Keyed on the PR rather than on each individual review row: the same PR is
+    re-reviewed on every push, and 113 rows over 26 PRs would make a benchmark
+    that is 4x redundant and weighted towards whichever PR churned most.
+    """
+    conn = sqlite3.connect(db_path)
+    conn.row_factory = sqlite3.Row
+    try:
+        prs = conn.execute(
+            """
+            SELECT repo, pr, MAX(posted_at) AS last_seen, COUNT(*) AS reviews,
+                   MAX(head_sha) AS head_sha
+            FROM review GROUP BY repo, pr ORDER BY repo, pr
+            """
+        ).fetchall()
+        items = []
+        for row in prs:
+            findings = conn.execute(
+                """
+                SELECT path, line, severity, problem, fix
+                FROM inline_finding WHERE repo = ? AND pr = ?
+                ORDER BY path, line
+                """,
+                (row["repo"], row["pr"]),
+            ).fetchall()
+            items.append(
+                {
+                    "id": f'{row["repo"]}#{row["pr"]}',
+                    "input": {
+                        "repo": row["repo"],
+                        "pr": int(row["pr"]),
+                        "head_sha": row["head_sha"],
+                    },
+                    "expectedOutput": {
+                        "findings": [dict(f) for f in findings],
+                        "finding_count": len(findings),
+                    },
+                    "metadata": {
+                        "reviews_run": int(row["reviews"]),
+                        "last_reviewed_at": int(row["last_seen"]),
+                        # Flags that this row is the reviewer's own past output,
+                        # not a human judgement. Filter on it before anyone
+                        # treats the dataset as ground truth.
+                        "labelled_by_human": False,
+                    },
+                }
+            )
+        return items
+    finally:
+        conn.close()
+
+
+def ensure_dataset(items: list[dict], name: str = DATASET_NAME) -> dict:
+    st, _ = _call(
+        "POST",
+        "/api/public/datasets",
+        {
+            "name": name,
+            "description": (
+                "PRs the pragent pilot has reviewed, seeded from feedback.db. "
+                "expectedOutput is the reviewer's own prior output — a regression "
+                "baseline, not human-verified ground truth."
+            ),
+            "metadata": {"source": "feedback.db", "seeded_by": "eval_bootstrap.py"},
+        },
+    )
+    # A duplicate name is fine: the dataset already exists from an earlier run.
+    dataset_ok = st in (200, 201, 409)
+
+    created, failed = 0, []
+    for item in items:
+        body = {
+            "datasetName": name,
+            "id": item["id"],  # idempotent: same PR updates rather than duplicates
+            "input": item["input"],
+            "expectedOutput": item["expectedOutput"],
+            "metadata": item["metadata"],
+        }
+        ist, resp = _call("POST", "/api/public/dataset-items", body)
+        if ist in (200, 201):
+            created += 1
+        else:
+            failed.append({"item": item["id"], "status": ist, "error": resp})
+    return {"dataset": name, "dataset_created": dataset_ok, "items_upserted": created, "failed": failed}
+
+
+# ---------------------------------------------------------------------------
+# 3. Backfill scores onto traces that predate the scorers
+# ---------------------------------------------------------------------------
+
+def _synth_findings(severities: dict) -> list[dict]:
+    """Rebuild a findings list from a trace's severity histogram.
+
+    Only severity matters to the scorers, and that is all the histogram kept.
+    Reconstructing placeholders is honest here because every scorer being
+    backfilled reads nothing else off a finding.
+    """
+    out = []
+    for sev, count in (severities or {}).items():
+        out.extend({"severity": sev} for _ in range(int(count)))
+    return out
+
+
+def backfill_traces(limit_pages: int = 20) -> dict:
+    import eval_scores as es
+
+    scored, skipped, events = 0, 0, []
+    page = 1
+    while page <= limit_pages:
+        st, resp = _call("GET", f"/api/public/traces?limit=50&page={page}&name=pr-review")
+        if st != 200 or not isinstance(resp, dict):
+            break
+        rows = resp.get("data") or []
+        if not rows:
+            break
+        for tr in rows:
+            meta = tr.get("metadata") or {}
+            severities = meta.get("severities") or {}
+            count = meta.get("findings")
+            if count is None:
+                skipped += 1
+                continue
+            findings = _synth_findings(severities)
+            # The histogram is authoritative when present; a trace that recorded
+            # a count but no histogram still scores its rate.
+            if not findings and count:
+                findings = [{"severity": "medium"} for _ in range(int(count))]
+            batch = es.build_scores(
+                trace_id=tr["id"],
+                findings=findings,
+                environment=tr.get("environment") or "default",
+                cost_usd=(tr.get("totalCost") or meta.get("provider_cost_usd")),
+                timestamp=tr.get("timestamp"),
+                comment="backfilled from trace metadata",
+            )
+            events.extend(batch)
+            scored += 1
+        page += 1
+
+    posted = False
+    status = None
+    if events:
+        import langfuse_trace
+
+        host, pk, sk = _conf()
+        # Chunked: one 2000-event POST is refused, and a partial backfill that
+        # reports success is worse than a slow one.
+        for i in range(0, len(events), 200):
+            status = langfuse_trace._post(host, pk, sk, events[i:i + 200], 30.0)
+            posted = status in (200, 201, 207)
+            if not posted:
+                break
+    return {"traces_scored": scored, "traces_skipped": skipped, "scores": len(events),
+            "posted": posted, "http_status": status}
+
+
+def main() -> int:
+    ap = argparse.ArgumentParser(description="Bootstrap Langfuse evaluation for the pragent pilot")
+    ap.add_argument("--db", default=os.environ.get("PRAGENT_FEEDBACK_DB", "/data/feedback.db"))
+    ap.add_argument("--skip-dataset", action="store_true")
+    ap.add_argument("--skip-configs", action="store_true")
+    ap.add_argument("--backfill-traces", action="store_true",
+                    help="score traces written before the scorers existed")
+    args = ap.parse_args()
+
+    out: dict = {}
+    if not args.skip_configs:
+        out["score_configs"] = ensure_score_configs()
+    if not args.skip_dataset:
+        items = read_review_items(args.db)
+        out["dataset"] = ensure_dataset(items)
+        out["dataset"]["items_read"] = len(items)
+    if args.backfill_traces:
+        out["trace_backfill"] = backfill_traces()
+    print(json.dumps(out, indent=2))
+
+    failed = (out.get("score_configs", {}).get("failed") or []) + (
+        out.get("dataset", {}).get("failed") or []
+    )
+    return 1 if failed else 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

pilot/eval_scores.py

@@ -0,0 +1,233 @@
+#!/usr/bin/env python3
+"""pragent pilot — deterministic review scorers.
+
+Four numbers computed from a review that already happened, shipped to Langfuse
+as scores on the review's trace. All are derived from data the reviewer already
+has in hand: no LLM judge, no ground truth, no extra token spend.
+
+Why these four and not `helpfulness`/`quality`
+----------------------------------------------
+They come from what the recorded reviews actually did, not from a generic eval
+checklist:
+
+  * `severity_info_ratio` — of the findings ever posted to a PR, effectively all
+    landed at `info`. Either the model will not commit to a severity or the
+    per-repo `severity_threshold` is filtering the rest out. Trending the ratio
+    per model says which.
+  * `finding_rate` — most reviews post nothing at all. Silence on clean code is
+    the goal; silence because the run degraded is a failure. Same output, two
+    causes, and only the rate over time separates them.
+  * `dropped_findings` — `ai_review.parse_findings` discards any finding whose
+    `path`/`line` is unusable. That happens silently, so a model that emits ten
+    findings at invalid locations is indistinguishable from one that found
+    nothing. This is the only signal here that measures the *model's* output
+    rather than the review's.
+  * `cost_per_finding` — the equivalent-cost number is already trended per
+    review; per finding is what actually compares two models, since a cheaper
+    model that finds nothing is not cheaper.
+
+None of these say whether a finding was *correct*. That needs labels, and the
+labels come from `feedback_scores.py` once maintainers start reacting to review
+comments. Read these as behavioural drift detectors, not as accuracy.
+
+Fail-open, like every other telemetry path here: a scorer that raises returns no
+score rather than failing the review.
+"""
+from __future__ import annotations
+
+import uuid
+from datetime import datetime, timezone
+
+# Mirrors ai_review.SEVERITY_RANK. Duplicated rather than imported because this
+# module is also run standalone (backfill) where ai_review's import side effects
+# are unwanted.
+SEVERITY_RANK = {"info": -1, "trivial": 0, "low": 1, "medium": 2, "high": 3, "critical": 4}
+
+# Findings at or below this rank are "the model declined to commit". `trivial`
+# and `info` are advisory by the reviewer's own prompt contract.
+_ADVISORY_MAX_RANK = 0
+
+# Score names. Named for what is measured, not for the mechanism producing it —
+# these land on every trace and become the axis of every chart.
+FINDING_RATE = "finding_rate"
+SEVERITY_INFO_RATIO = "severity_info_ratio"
+SEVERITY_MAX = "severity_max"
+DROPPED_FINDINGS = "dropped_findings"
+COST_PER_FINDING = "cost_per_finding"
+
+
+def _sev(f: dict) -> str:
+    return str(f.get("severity") or "medium").strip().lower()
+
+
+def finding_rate(findings: list[dict] | None) -> float:
+    """How many findings this review posted. 0.0 is the restraint case."""
+    return float(len(findings or []))
+
+
+def severity_info_ratio(findings: list[dict] | None) -> float | None:
+    """Share of findings the model rated advisory (`info`/`trivial`).
+
+    `None` for a review with no findings — a ratio over an empty set is not 0,
+    it is undefined, and charting it as 0 would read as "perfectly calibrated".
+    """
+    fs = findings or []
+    if not fs:
+        return None
+    advisory = sum(1 for f in fs if SEVERITY_RANK.get(_sev(f), 2) <= _ADVISORY_MAX_RANK)
+    return round(advisory / len(fs), 4)
+
+
+def severity_max(findings: list[dict] | None) -> str:
+    """Highest severity present, or `none` when the review was silent.
+
+    Categorical on purpose: the useful question is "did this review ever surface
+    something serious", and an average of severity ranks answers nothing.
+    """
+    fs = findings or []
+    if not fs:
+        return "none"
+    top = max(fs, key=lambda f: SEVERITY_RANK.get(_sev(f), 2))
+    sev = _sev(top)
+    return sev if sev in SEVERITY_RANK else "medium"
+
+
+def dropped_findings(raw_count: int | None, kept_count: int | None) -> float | None:
+    """Findings the model emitted that the parser could not use.
+
+    `raw_count` is what came back in the JSON; `kept_count` is what survived
+    `_normalize_finding`. `None` when the caller could not determine the raw
+    count — better no score than a fabricated zero.
+    """
+    if raw_count is None or kept_count is None:
+        return None
+    return float(max(0, int(raw_count) - int(kept_count)))
+
+
+def cost_per_finding(cost_usd: float | None, findings: list[dict] | None) -> float | None:
+    """Equivalent USD spent per finding posted.
+
+    `None` when nothing could be priced. A silent review divides by one, not by
+    zero: the run still cost money, and attributing that whole cost to "found
+    nothing" is the honest reading.
+    """
+    if cost_usd is None:
+        return None
+    try:
+        c = float(cost_usd)
+    except (TypeError, ValueError):
+        return None
+    return round(c / max(1, len(findings or [])), 6)
+
+
+def build_scores(
+    *,
+    trace_id: str,
+    findings: list[dict] | None,
+    environment: str,
+    cost_usd: float | None = None,
+    dropped_count: float | None = None,
+    timestamp: str | None = None,
+    comment: str = "",
+) -> list[dict]:
+    """The `score-create` ingestion events for one review.
+
+    `dropped_count` must be measured at parse time, not here: by the time
+    `findings` reaches this function the per-repo config has already filtered it
+    by severity threshold and `max_findings`, and those drops are the config
+    working as intended, not the model emitting garbage.
+
+    Returns [] rather than raising if something is unscoreable — scores are
+    telemetry and must never cost a review.
+    """
+    # The ingestion envelope requires a timestamp on every event; omitting it
+    # gets the whole batch rejected with an HTTP 207 whose per-event 400s are
+    # easy to mistake for success.
+    ts = timestamp or datetime.now(timezone.utc).isoformat().replace("+00:00", "Z")
+    out: list[dict] = []
+
+    def add(name: str, value, data_type: str) -> None:
+        if value is None:
+            return
+        body = {
+            "id": str(uuid.uuid4()),
+            "traceId": trace_id,
+            "name": name,
+            "dataType": data_type,
+            "environment": environment,
+        }
+        if data_type == "CATEGORICAL":
+            body["value"] = str(value)
+        else:
+            body["value"] = float(value)
+        if comment:
+            body["comment"] = comment
+        out.append(
+            {
+                "id": str(uuid.uuid4()),
+                "type": "score-create",
+                "timestamp": ts,
+                "body": body,
+            }
+        )
+
+    try:
+        add(FINDING_RATE, finding_rate(findings), "NUMERIC")
+        add(SEVERITY_INFO_RATIO, severity_info_ratio(findings), "NUMERIC")
+        add(SEVERITY_MAX, severity_max(findings), "CATEGORICAL")
+        add(DROPPED_FINDINGS, dropped_count, "NUMERIC")
+        add(COST_PER_FINDING, cost_per_finding(cost_usd, findings), "NUMERIC")
+    except Exception:  # pragma: no cover - defensive
+        return out
+    return out
+
+
+# ---------------------------------------------------------------------------
+# Score configs — the schema these scores must comply with
+# ---------------------------------------------------------------------------
+
+# Registered once per project via `eval_bootstrap.py`. Without configs the
+# scores still ingest, but nothing constrains a future scorer from writing
+# `severity_max="HIGH"` next to today's `"high"` and silently splitting the
+# series in two.
+SCORE_CONFIGS = [
+    {
+        "name": FINDING_RATE,
+        "dataType": "NUMERIC",
+        "minValue": 0,
+        "description": "Findings posted by one review. 0 = the reviewer stayed silent.",
+    },
+    {
+        "name": SEVERITY_INFO_RATIO,
+        "dataType": "NUMERIC",
+        "minValue": 0,
+        "maxValue": 1,
+        "description": "Share of a review's findings rated info/trivial. High = the model is not committing to a severity.",
+    },
+    {
+        "name": SEVERITY_MAX,
+        "dataType": "CATEGORICAL",
+        "categories": [
+            {"label": "none", "value": 0},
+            {"label": "info", "value": 1},
+            {"label": "trivial", "value": 2},
+            {"label": "low", "value": 3},
+            {"label": "medium", "value": 4},
+            {"label": "high", "value": 5},
+            {"label": "critical", "value": 6},
+        ],
+        "description": "Highest severity surfaced by one review; 'none' when it posted nothing.",
+    },
+    {
+        "name": DROPPED_FINDINGS,
+        "dataType": "NUMERIC",
+        "minValue": 0,
+        "description": "Findings the model emitted that the parser rejected for an unusable path/line.",
+    },
+    {
+        "name": COST_PER_FINDING,
+        "dataType": "NUMERIC",
+        "minValue": 0,
+        "description": "Equivalent USD per finding posted. Silent reviews divide by 1, not 0.",
+    },
+]

pilot/feedback_scores.py

@@ -0,0 +1,247 @@
+#!/usr/bin/env python3
+"""pragent pilot — feedback DB to Langfuse scores.
+
+`feedback.db` already records every reaction, thread resolution and reply a
+maintainer leaves on a bot comment. That is the only ground truth pragent has
+about whether a finding was any good, and until now it went to a markdown report
+nobody reads and nowhere else. This ships it to Langfuse as session-level
+scores, so "was the reviewer right" sits on the same axis as "what did it cost".
+
+Session, not trace
+------------------
+`langfuse_trace` sets `sessionId` to `"{repo}#{pr}"` and lets the trace id be a
+fresh uuid per review. Feedback arrives days later against a PR, not against one
+particular re-run of the reviewer, and nothing in `feedback.db` records which
+trace produced which comment. Scoring the session is therefore both the
+available join and the honest granularity: this is feedback on the review of
+this PR, not on one invocation.
+
+Two scores, deliberately separated
+----------------------------------
+  * `review_engagement` — the share of a PR's findings that got any human
+    response at all. This is a signal about the *feedback loop*, not the
+    reviewer: at the time of writing it is 0.0 across all 113 recorded reviews,
+    which is exactly the fact that makes an accuracy metric impossible today.
+    It must be watched first, because every other quality number is vapour
+    until it moves.
+  * `review_acceptance` — net verdict over the findings that *did* get a
+    response: (upvotes + resolved) - (downvotes + negation replies), normalised
+    to -1..1. Computed only over engaged findings, so an ignored review scores
+    `None` rather than 0. Zero would read as "humans judged this exactly
+    neutral"; the truth is nobody looked.
+
+Fail-open and idempotent. Score ids are derived from (repo, pr, name) so a
+re-run overwrites rather than duplicates.
+"""
+from __future__ import annotations
+
+import argparse
+import json
+import os
+import sqlite3
+import sys
+import uuid
+from datetime import datetime, timezone
+
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+
+from feedback_harvest import classify_reaction, _is_negation_reply  # noqa: E402
+
+REVIEW_ENGAGEMENT = "review_engagement"
+REVIEW_ACCEPTANCE = "review_acceptance"
+
+# Stable namespace so the same (repo, pr, score) always produces the same score
+# id — Langfuse treats a repeated id as an update, which is what a backfill of a
+# still-accumulating PR should do.
+_NS = uuid.UUID("6f1d9c2e-4a77-4f2a-9c1a-0d3b5e8a7c41")
+
+
+def _score_id(repo: str, pr: int, name: str) -> str:
+    return str(uuid.uuid5(_NS, f"{repo}#{pr}#{name}"))
+
+
+def collect_pr_feedback(conn: sqlite3.Connection, repo: str, pr: int) -> dict:
+    """Tally one PR's findings and the human responses attached to them.
+
+    Returns counts only — the scoring maths lives in `score_pr` so it can be
+    tested without a database.
+    """
+    rows = conn.execute(
+        "SELECT id, comment_id FROM inline_finding WHERE repo = ? AND pr = ?",
+        (repo, pr),
+    ).fetchall()
+    total = len(rows)
+    engaged = 0
+    positive = 0
+    negative = 0
+
+    for row in rows:
+        fid = row["id"] if isinstance(row, sqlite3.Row) else row[0]
+        cid = row["comment_id"] if isinstance(row, sqlite3.Row) else row[1]
+        pos = neg = 0
+
+        if cid is not None:
+            for r in conn.execute(
+                "SELECT content FROM reaction WHERE comment_id = ?", (cid,)
+            ):
+                kind = classify_reaction(r[0])
+                if kind == "positive":
+                    pos += 1
+                elif kind == "negative":
+                    neg += 1
+
+        for r in conn.execute(
+            "SELECT resolved FROM thread_state WHERE finding_id = ?", (fid,)
+        ):
+            # A resolved thread means the maintainer acted on the finding.
+            if r[0]:
+                pos += 1
+
+        # A reply counts as engagement either way; only a negation phrase makes
+        # it a vote against. A neutral reply ("done", "good catch, but…") is
+        # deliberately not a positive vote — it says someone looked, not that
+        # they agreed.
+        replied = 0
+        for r in conn.execute(
+            "SELECT body FROM reply WHERE finding_id = ?", (fid,)
+        ):
+            replied += 1
+            if _is_negation_reply(r[0]):
+                neg += 1
+
+        if pos or neg or replied:
+            engaged += 1
+        positive += pos
+        negative += neg
+
+    return {"total": total, "engaged": engaged, "positive": positive, "negative": negative}
+
+
+def score_pr(tally: dict) -> dict:
+    """Turn one PR's tally into score values.
+
+    `review_acceptance` is `None` when nothing was engaged — see the module
+    docstring on why that is not 0.
+    """
+    total = int(tally.get("total") or 0)
+    engaged = int(tally.get("engaged") or 0)
+    pos = int(tally.get("positive") or 0)
+    neg = int(tally.get("negative") or 0)
+
+    engagement = round(engaged / total, 4) if total else None
+    acceptance = None
+    if pos or neg:
+        acceptance = round((pos - neg) / (pos + neg), 4)
+    return {REVIEW_ENGAGEMENT: engagement, REVIEW_ACCEPTANCE: acceptance}
+
+
+def build_score_events(
+    repo: str, pr: int, values: dict, environment: str = "default",
+    timestamp: str | None = None,
+) -> list[dict]:
+    """`score-create` events for one PR's feedback.
+
+    Every event carries a timestamp: the ingestion endpoint rejects those that
+    do not, and it reports the rejection as a per-event 400 inside an HTTP 207,
+    which reads as success to a caller that only checks the status code.
+    """
+    ts = timestamp or datetime.now(timezone.utc).isoformat().replace("+00:00", "Z")
+    events = []
+    for name, value in values.items():
+        if value is None:
+            continue
+        events.append(
+            {
+                "id": str(uuid.uuid4()),
+                "type": "score-create",
+                "timestamp": ts,
+                "body": {
+                    "id": _score_id(repo, pr, name),
+                    "sessionId": f"{repo}#{pr}",
+                    "name": name,
+                    "value": float(value),
+                    "dataType": "NUMERIC",
+                    "environment": environment,
+                    "comment": f"from feedback.db · {repo}#{pr}",
+                },
+            }
+        )
+    return events
+
+
+SCORE_CONFIGS = [
+    {
+        "name": REVIEW_ENGAGEMENT,
+        "dataType": "NUMERIC",
+        "minValue": 0,
+        "maxValue": 1,
+        "description": "Share of a PR's findings that drew any human reaction, resolution or reply. 0 = nobody engaged with the review.",
+    },
+    {
+        "name": REVIEW_ACCEPTANCE,
+        "dataType": "NUMERIC",
+        "minValue": -1,
+        "maxValue": 1,
+        "description": "Net human verdict over engaged findings: +1 all accepted, -1 all rejected. Absent when nothing was engaged.",
+    },
+]
+
+
+def iter_prs(conn: sqlite3.Connection):
+    for row in conn.execute(
+        "SELECT DISTINCT repo, pr FROM inline_finding ORDER BY repo, pr"
+    ):
+        yield row[0], int(row[1])
+
+
+def backfill(db_path: str, *, environment: str = "default", dry_run: bool = False) -> dict:
+    """Score every PR in the feedback DB. Returns a summary dict."""
+    conn = sqlite3.connect(db_path)
+    conn.row_factory = sqlite3.Row
+    events: list[dict] = []
+    scanned = 0
+    engaged_prs = 0
+    try:
+        for repo, pr in iter_prs(conn):
+            scanned += 1
+            tally = collect_pr_feedback(conn, repo, pr)
+            values = score_pr(tally)
+            if (values.get(REVIEW_ENGAGEMENT) or 0) > 0:
+                engaged_prs += 1
+            events.extend(build_score_events(repo, pr, values, environment))
+    finally:
+        conn.close()
+
+    summary = {"prs_scanned": scanned, "prs_with_engagement": engaged_prs, "scores": len(events)}
+    if dry_run or not events:
+        summary["posted"] = False
+        return summary
+
+    import langfuse_trace
+
+    conf = langfuse_trace._enabled()
+    if conf is None:
+        summary["posted"] = False
+        summary["error"] = "Langfuse not configured (LANGFUSE_HOST / keys unset)"
+        return summary
+    host, pk, sk = conf
+    status = langfuse_trace._post(host, pk, sk, events, 15.0)
+    summary["posted"] = status in (200, 201, 207)
+    summary["http_status"] = status
+    return summary
+
+
+def main() -> int:
+    ap = argparse.ArgumentParser(description="Ship feedback.db verdicts to Langfuse as scores")
+    ap.add_argument("--db", default=os.environ.get("PRAGENT_FEEDBACK_DB", "/data/feedback.db"))
+    ap.add_argument("--environment", default="default")
+    ap.add_argument("--dry-run", action="store_true")
+    args = ap.parse_args()
+
+    summary = backfill(args.db, environment=args.environment, dry_run=args.dry_run)
+    print(json.dumps(summary, indent=2))
+    return 0 if summary.get("posted") or args.dry_run else 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

pilot/langfuse_trace.py

@@ -218,11 +218,17 @@ def build_batch(
     trace_id: str | None = None,
     release: str = "",
     price_target: str | None = None,
+    dropped_count: float | None = None,
 ) -> list[dict]:
-    """The ingestion batch for one review: a trace plus one generation.
+    """The ingestion batch for one review: a trace, a generation, and scores.
 
     Split out from `emit_review_trace` so the shape is testable without a
     Langfuse to POST to.
+
+    `dropped_count` is how many findings the parser rejected for an unusable
+    `path`/`line`, measured where the model output was parsed. Passing it turns
+    on the `dropped_findings` score; leaving it `None` omits that score rather
+    than reporting a zero the caller never measured.
     """
     usage = usage or {}
     tid = trace_id or str(uuid.uuid4())
@@ -316,9 +322,40 @@ def build_batch(
             }
         )
 
+    events.extend(
+        _score_events(
+            trace_id=tid,
+            findings=findings,
+            environment=env,
+            cost_usd=costs.get("total"),
+            dropped_count=dropped_count,
+            timestamp=ts,
+            cost_basis=cost_basis,
+        )
+    )
+
     return events
 
 
+def _score_events(*, cost_basis: str, **kwargs) -> list[dict]:
+    """Deterministic scores for this review, or [] if the scorer is missing.
+
+    Local import + blanket except for the same reason the rest of this module
+    swallows: `eval_scores` is optional, and a scoring bug must not cost the
+    trace it was supposed to annotate.
+    """
+    try:
+        import eval_scores
+
+        # The cost score is only meaningful next to its basis — a $/finding
+        # figure computed from an equivalent price is not money that was spent.
+        comment = f"cost basis: {cost_basis}" if cost_basis else ""
+        return eval_scores.build_scores(comment=comment, **kwargs)
+    except Exception as e:  # pragma: no cover - defensive
+        _debug(f"scoring failed: {e}")
+        return []
+
+
 def _post(host: str, pk: str, sk: str, batch: list[dict], timeout: float) -> int:
     payload = json.dumps({"batch": batch}).encode("utf-8")
     auth = base64.b64encode(f"{pk}:{sk}".encode("utf-8")).decode("ascii")
@@ -333,9 +370,31 @@ def _post(host: str, pk: str, sk: str, batch: list[dict], timeout: float) -> int
         method="POST",
     )
     with urllib.request.urlopen(req, timeout=timeout) as resp:
+        _warn_on_rejected_events(resp.read())
         return resp.status
 
 
+def _warn_on_rejected_events(raw: bytes) -> None:
+    """Surface per-event rejections hiding inside a 207.
+
+    The ingestion endpoint answers 207 Multi-Status when *some* events failed,
+    so a caller that only checks the status code reads a batch where every
+    single event was rejected as a success. That failure mode is invisible
+    exactly when it matters — the traces simply never appear.
+    """
+    try:
+        body = json.loads(raw or b"{}")
+        errors = body.get("errors") or []
+        if errors:
+            first = errors[0]
+            _debug(
+                f"{len(errors)} event(s) rejected by ingestion; "
+                f"first: status={first.get('status')} {first.get('error')}"
+            )
+    except Exception:  # pragma: no cover - never let logging break emission
+        pass
+
+
 def emit_review_trace(**kwargs) -> bool:
     """Ship one review's trace. Returns True if Langfuse accepted it.
 

tests/pilot/test_eval_scores.py

@@ -0,0 +1,200 @@
+"""Tests for the deterministic review scorers."""
+import os
+import sys
+
+import pytest
+
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", "..", "pilot"))
+
+import eval_scores as es  # noqa: E402
+
+
+def f(sev, path="a.py", line=1):
+    return {"severity": sev, "path": path, "line": line, "problem": "p", "fix": ""}
+
+
+# --- finding_rate ---------------------------------------------------------
+
+def test_finding_rate_counts_findings():
+    assert es.finding_rate([f("high"), f("low")]) == 2.0
+
+
+def test_finding_rate_zero_for_silent_review():
+    assert es.finding_rate([]) == 0.0
+    assert es.finding_rate(None) == 0.0
+
+
+# --- severity_info_ratio --------------------------------------------------
+
+def test_info_ratio_all_advisory():
+    assert es.severity_info_ratio([f("info"), f("trivial")]) == 1.0
+
+
+def test_info_ratio_mixed():
+    assert es.severity_info_ratio([f("info"), f("high")]) == 0.5
+
+
+def test_info_ratio_none_when_no_findings():
+    # Undefined, not zero — zero would read as perfectly calibrated.
+    assert es.severity_info_ratio([]) is None
+
+
+def test_info_ratio_unknown_severity_treated_as_medium():
+    # Matches _normalize_finding's fallback, so an odd severity is not
+    # silently counted as advisory.
+    assert es.severity_info_ratio([f("bogus")]) == 0.0
+
+
+# --- severity_max ---------------------------------------------------------
+
+def test_severity_max_picks_highest():
+    assert es.severity_max([f("info"), f("critical"), f("low")]) == "critical"
+
+
+def test_severity_max_none_when_silent():
+    assert es.severity_max([]) == "none"
+
+
+def test_severity_max_case_insensitive():
+    assert es.severity_max([f("HIGH")]) == "high"
+
+
+# --- dropped_findings -----------------------------------------------------
+
+def test_dropped_findings_delta():
+    assert es.dropped_findings(5, 2) == 3.0
+
+
+def test_dropped_findings_never_negative():
+    assert es.dropped_findings(1, 3) == 0.0
+
+
+def test_dropped_findings_none_when_unknown():
+    assert es.dropped_findings(None, 2) is None
+
+
+# --- cost_per_finding -----------------------------------------------------
+
+def test_cost_per_finding_divides():
+    assert es.cost_per_finding(1.0, [f("high"), f("low")]) == 0.5
+
+
+def test_cost_per_finding_silent_review_divides_by_one():
+    # The run still cost money; attributing all of it to "found nothing" is
+    # the honest reading, and it avoids a division by zero.
+    assert es.cost_per_finding(0.25, []) == 0.25
+
+
+def test_cost_per_finding_none_when_unpriced():
+    assert es.cost_per_finding(None, [f("high")]) is None
+
+
+def test_cost_per_finding_none_on_garbage():
+    assert es.cost_per_finding("abc", [f("high")]) is None
+
+
+# --- build_scores ---------------------------------------------------------
+
+def _by_name(events):
+    return {e["body"]["name"]: e["body"] for e in events}
+
+
+def test_build_scores_emits_expected_set():
+    events = es.build_scores(
+        trace_id="t1", findings=[f("high"), f("info")], environment="claude",
+        cost_usd=0.5, dropped_count=2, timestamp="2026-01-01T00:00:00Z",
+    )
+    names = _by_name(events)
+    assert set(names) == {
+        es.FINDING_RATE, es.SEVERITY_INFO_RATIO, es.SEVERITY_MAX,
+        es.DROPPED_FINDINGS, es.COST_PER_FINDING,
+    }
+    assert names[es.FINDING_RATE]["value"] == 2.0
+    assert names[es.SEVERITY_MAX]["value"] == "high"
+    assert names[es.DROPPED_FINDINGS]["value"] == 2.0
+    assert names[es.COST_PER_FINDING]["value"] == 0.25
+
+
+def test_build_scores_all_events_are_score_create_on_the_trace():
+    events = es.build_scores(
+        trace_id="t9", findings=[f("low")], environment="ollama", cost_usd=1.0,
+    )
+    assert all(e["type"] == "score-create" for e in events)
+    assert all(e["body"]["traceId"] == "t9" for e in events)
+    assert all(e["body"]["environment"] == "ollama" for e in events)
+
+
+def test_build_scores_omits_undefined_scores():
+    # No cost and no drop count measured -> those scores are absent, not zero.
+    events = es.build_scores(trace_id="t2", findings=[], environment="ollama")
+    names = set(_by_name(events))
+    assert es.COST_PER_FINDING not in names
+    assert es.DROPPED_FINDINGS not in names
+    assert es.SEVERITY_INFO_RATIO not in names
+    assert names == {es.FINDING_RATE, es.SEVERITY_MAX}
+
+
+def test_build_scores_categorical_value_is_string():
+    events = es.build_scores(trace_id="t3", findings=[f("high")], environment="claude")
+    sev = _by_name(events)[es.SEVERITY_MAX]
+    assert sev["dataType"] == "CATEGORICAL"
+    assert isinstance(sev["value"], str)
+
+
+def test_build_scores_numeric_values_are_floats():
+    events = es.build_scores(
+        trace_id="t4", findings=[f("high")], environment="claude", cost_usd=1,
+    )
+    for name, body in _by_name(events).items():
+        if body["dataType"] == "NUMERIC":
+            assert isinstance(body["value"], float), name
+
+
+def test_build_scores_comment_propagates():
+    events = es.build_scores(
+        trace_id="t5", findings=[f("high")], environment="claude",
+        cost_usd=1.0, comment="cost basis: equivalent:claude-sonnet-5",
+    )
+    assert all("equivalent" in e["body"]["comment"] for e in events)
+
+
+# --- score configs --------------------------------------------------------
+
+def test_every_emitted_score_has_a_config():
+    configured = {c["name"] for c in es.SCORE_CONFIGS}
+    events = es.build_scores(
+        trace_id="t6", findings=[f("high")], environment="claude",
+        cost_usd=1.0, dropped_count=0,
+    )
+    assert set(_by_name(events)) <= configured
+
+
+def test_severity_max_config_covers_every_severity_it_can_emit():
+    labels = {c["label"] for c in
+              next(c for c in es.SCORE_CONFIGS if c["name"] == es.SEVERITY_MAX)["categories"]}
+    assert set(es.SEVERITY_RANK) | {"none"} == labels
+
+
+# --- ingestion envelope ---------------------------------------------------
+
+def test_every_event_carries_a_timestamp():
+    # Ingestion rejects events without one, and reports the rejection as a
+    # per-event 400 inside an HTTP 207 that reads as success.
+    events = es.build_scores(
+        trace_id="t7", findings=[f("high")], environment="claude", cost_usd=1.0,
+    )
+    assert events
+    assert all(e.get("timestamp") for e in events)
+
+
+def test_timestamp_defaults_when_caller_omits_it():
+    events = es.build_scores(trace_id="t8", findings=[f("low")], environment="claude")
+    assert all(isinstance(e["timestamp"], str) and e["timestamp"].endswith("Z") for e in events)
+
+
+def test_explicit_timestamp_is_used():
+    events = es.build_scores(
+        trace_id="t9", findings=[f("low")], environment="claude",
+        timestamp="2026-01-02T03:04:05Z",
+    )
+    assert all(e["timestamp"] == "2026-01-02T03:04:05Z" for e in events)

tests/pilot/test_feedback_scores.py

@@ -0,0 +1,207 @@
+"""Tests for the feedback.db -> Langfuse score bridge."""
+import os
+import sys
+
+import pytest
+
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", "..", "pilot"))
+
+import feedback  # noqa: E402
+import feedback_scores as fs  # noqa: E402
+
+
+@pytest.fixture
+def db(tmp_path):
+    conn = feedback.init(str(tmp_path / "fb.db"))
+    yield conn
+    conn.close()
+
+
+def _seed_finding(conn, repo="o/r", pr=1, comment_id=100, path="a.py", line=1):
+    cur = conn.execute(
+        "INSERT INTO review (repo, pr, head_sha, posted_at) VALUES (?,?,?,?)",
+        (repo, pr, "deadbeef", 1000),
+    )
+    review_id = cur.lastrowid
+    cur = conn.execute(
+        """INSERT INTO inline_finding
+           (review_id, repo, pr, path, line, severity, problem, comment_id, posthash, posted_at)
+           VALUES (?,?,?,?,?,?,?,?,?,?)""",
+        (review_id, repo, pr, path, line, "HIGH", "problem", comment_id, f"h{comment_id}", 1000),
+    )
+    conn.commit()
+    return cur.lastrowid
+
+
+# --- score_pr maths -------------------------------------------------------
+
+def test_engagement_zero_when_nobody_responded():
+    v = fs.score_pr({"total": 4, "engaged": 0, "positive": 0, "negative": 0})
+    assert v[fs.REVIEW_ENGAGEMENT] == 0.0
+
+
+def test_acceptance_absent_when_nobody_engaged():
+    # Not 0.0 — zero would claim humans judged it neutral.
+    v = fs.score_pr({"total": 4, "engaged": 0, "positive": 0, "negative": 0})
+    assert v[fs.REVIEW_ACCEPTANCE] is None
+
+
+def test_engagement_is_a_share_of_findings():
+    v = fs.score_pr({"total": 4, "engaged": 1, "positive": 1, "negative": 0})
+    assert v[fs.REVIEW_ENGAGEMENT] == 0.25
+
+
+def test_acceptance_all_positive():
+    v = fs.score_pr({"total": 2, "engaged": 2, "positive": 3, "negative": 0})
+    assert v[fs.REVIEW_ACCEPTANCE] == 1.0
+
+
+def test_acceptance_all_negative():
+    v = fs.score_pr({"total": 2, "engaged": 2, "positive": 0, "negative": 2})
+    assert v[fs.REVIEW_ACCEPTANCE] == -1.0
+
+
+def test_acceptance_mixed_is_normalised():
+    v = fs.score_pr({"total": 4, "engaged": 4, "positive": 3, "negative": 1})
+    assert v[fs.REVIEW_ACCEPTANCE] == 0.5
+
+
+def test_engagement_absent_when_no_findings_at_all():
+    v = fs.score_pr({"total": 0, "engaged": 0, "positive": 0, "negative": 0})
+    assert v[fs.REVIEW_ENGAGEMENT] is None
+
+
+# --- collect_pr_feedback over a real sqlite ------------------------------
+
+def test_collect_counts_nothing_on_untouched_findings(db):
+    _seed_finding(db)
+    tally = fs.collect_pr_feedback(db, "o/r", 1)
+    assert tally == {"total": 1, "engaged": 0, "positive": 0, "negative": 0}
+
+
+def test_collect_counts_positive_reaction(db):
+    _seed_finding(db, comment_id=101)
+    db.execute(
+        "INSERT INTO reaction (comment_id, user, content, created_at) VALUES (?,?,?,?)",
+        (101, "alice", "+1", 1),
+    )
+    db.commit()
+    tally = fs.collect_pr_feedback(db, "o/r", 1)
+    assert tally["positive"] == 1 and tally["engaged"] == 1
+
+
+def test_collect_counts_negative_reaction(db):
+    _seed_finding(db, comment_id=102)
+    db.execute(
+        "INSERT INTO reaction (comment_id, user, content, created_at) VALUES (?,?,?,?)",
+        (102, "bob", "-1", 1),
+    )
+    db.commit()
+    tally = fs.collect_pr_feedback(db, "o/r", 1)
+    assert tally["negative"] == 1 and tally["engaged"] == 1
+
+
+def test_resolved_thread_counts_positive(db):
+    fid = _seed_finding(db, comment_id=103)
+    db.execute(
+        "INSERT INTO thread_state (finding_id, resolved, checked_at) VALUES (?,?,?)",
+        (fid, 1, 1),
+    )
+    db.commit()
+    tally = fs.collect_pr_feedback(db, "o/r", 1)
+    assert tally["positive"] == 1 and tally["engaged"] == 1
+
+
+def test_unresolved_thread_is_not_a_vote(db):
+    fid = _seed_finding(db, comment_id=104)
+    db.execute(
+        "INSERT INTO thread_state (finding_id, resolved, checked_at) VALUES (?,?,?)",
+        (fid, 0, 1),
+    )
+    db.commit()
+    tally = fs.collect_pr_feedback(db, "o/r", 1)
+    assert tally == {"total": 1, "engaged": 0, "positive": 0, "negative": 0}
+
+
+def test_negation_reply_counts_negative(db):
+    fid = _seed_finding(db, comment_id=105)
+    db.execute(
+        "INSERT INTO reply (finding_id, author, body, created_at) VALUES (?,?,?,?)",
+        (fid, "carol", "this is a false positive", 1),
+    )
+    db.commit()
+    tally = fs.collect_pr_feedback(db, "o/r", 1)
+    assert tally["negative"] == 1 and tally["engaged"] == 1
+
+
+def test_neutral_reply_is_engagement_but_not_a_vote(db):
+    fid = _seed_finding(db, comment_id=106)
+    db.execute(
+        "INSERT INTO reply (finding_id, author, body, created_at) VALUES (?,?,?,?)",
+        (fid, "dave", "done", 1),
+    )
+    db.commit()
+    tally = fs.collect_pr_feedback(db, "o/r", 1)
+    assert tally["engaged"] == 1
+    assert tally["positive"] == 0 and tally["negative"] == 0
+
+
+# --- event shape ----------------------------------------------------------
+
+def test_build_score_events_shape():
+    events = fs.build_score_events("o/r", 7, {fs.REVIEW_ENGAGEMENT: 0.5}, "claude")
+    assert len(events) == 1
+    body = events[0]["body"]
+    assert events[0]["type"] == "score-create"
+    assert body["sessionId"] == "o/r#7"
+    assert body["value"] == 0.5
+    assert body["environment"] == "claude"
+
+
+def test_build_score_events_skips_none():
+    events = fs.build_score_events("o/r", 7, {fs.REVIEW_ACCEPTANCE: None})
+    assert events == []
+
+
+def test_score_ids_are_stable_across_runs():
+    # A backfill re-run must update, not duplicate.
+    a = fs.build_score_events("o/r", 7, {fs.REVIEW_ENGAGEMENT: 0.5})[0]["body"]["id"]
+    b = fs.build_score_events("o/r", 7, {fs.REVIEW_ENGAGEMENT: 0.9})[0]["body"]["id"]
+    assert a == b
+
+
+def test_score_ids_differ_per_pr_and_name():
+    e1 = fs.build_score_events("o/r", 7, {fs.REVIEW_ENGAGEMENT: 1})[0]["body"]["id"]
+    e2 = fs.build_score_events("o/r", 8, {fs.REVIEW_ENGAGEMENT: 1})[0]["body"]["id"]
+    e3 = fs.build_score_events("o/r", 7, {fs.REVIEW_ACCEPTANCE: 1})[0]["body"]["id"]
+    assert len({e1, e2, e3}) == 3
+
+
+def test_backfill_dry_run_reports_without_posting(db, tmp_path):
+    _seed_finding(db, comment_id=107)
+    db.commit()
+    path = db.execute("PRAGMA database_list").fetchone()[2]
+    summary = fs.backfill(path, dry_run=True)
+    assert summary["prs_scanned"] == 1
+    assert summary["prs_with_engagement"] == 0
+    assert summary["posted"] is False
+
+
+def test_every_emitted_score_has_a_config():
+    configured = {c["name"] for c in fs.SCORE_CONFIGS}
+    assert {fs.REVIEW_ENGAGEMENT, fs.REVIEW_ACCEPTANCE} == configured
+
+
+def test_every_event_carries_a_timestamp():
+    # Without one the ingestion endpoint 400s the event inside a 207 that the
+    # caller reads as success.
+    events = fs.build_score_events("o/r", 1, {fs.REVIEW_ENGAGEMENT: 0.0})
+    assert events
+    assert all(e.get("timestamp") for e in events)
+
+
+def test_explicit_timestamp_is_used():
+    events = fs.build_score_events(
+        "o/r", 1, {fs.REVIEW_ENGAGEMENT: 0.0}, timestamp="2026-01-02T03:04:05Z"
+    )
+    assert events[0]["timestamp"] == "2026-01-02T03:04:05Z"

tests/pilot/test_langfuse_trace.py

@@ -145,15 +145,18 @@ def test_unknown_comparison_target_yields_no_cost_block_rather_than_a_wrong_one(
 def test_batch_has_a_trace_and_a_generation_linked_by_trace_id():
     batch = lt.build_batch(model="headroom/claude-sonnet-5", **BASE)
     types = [e["type"] for e in batch]
-    assert types == ["trace-create", "generation-create"]
+    # Scores ride in the same batch; the trace and generation lead it.
-    trace, gen = batch
+    assert types[:2] == ["trace-create", "generation-create"]
+    trace, gen = batch[0], batch[1]
     assert gen["body"]["traceId"] == trace["body"]["id"]
     assert trace["body"]["environment"] == gen["body"]["environment"] == "claude"
 
 
-def test_batch_without_usage_is_trace_only():
+def test_batch_without_usage_has_no_generation():
     batch = lt.build_batch(model="headroom/glm-5.2:cloud", **{**BASE, "usage": None})
-    assert [e["type"] for e in batch] == ["trace-create"]
+    types = [e["type"] for e in batch]
+    assert "generation-create" not in types
+    assert types[0] == "trace-create"
 
 
 def test_trace_carries_repo_pr_session_and_severity_counts():
@@ -226,7 +229,9 @@ def test_configured_emit_posts_to_the_ingestion_endpoint(monkeypatch):
     assert lt.emit_review_trace(model="headroom/claude-sonnet-5", **BASE) is True
     # Trailing slash stripped so the path is not doubled.
     assert seen["host"] == "http://langfuse.test:3000"
-    assert len(seen["batch"]) == 2
+    kinds = [e["type"] for e in seen["batch"]]
+    assert kinds[:2] == ["trace-create", "generation-create"]
+    assert "score-create" in kinds
 
 
 def test_transport_failure_is_swallowed(monkeypatch):
@@ -243,3 +248,79 @@ def test_non_success_status_reports_failure_without_raising(monkeypatch):
     _configure(monkeypatch)
     monkeypatch.setattr(lt, "_post", lambda *a, **k: 401)
     assert lt.emit_review_trace(model="headroom/glm-5.2:cloud", **BASE) is False
+
+
+# ---------------------------------------------------------------------------
+# Scores folded into the review batch (added with eval_scores)
+# ---------------------------------------------------------------------------
+
+def _scores(events):
+    return {e["body"]["name"]: e["body"] for e in events if e["type"] == "score-create"}
+
+
+def test_build_batch_appends_scores():
+    events = lt.build_batch(
+        repo="o/r", index="1", sha="abc", title="t",
+        model="headroom/claude-sonnet-5",
+        usage={"input": 100, "output": 10},
+        findings=[{"severity": "high", "path": "a.py", "line": 1}],
+    )
+    names = set(_scores(events))
+    assert "finding_rate" in names
+    assert "severity_max" in names
+
+
+def test_scores_attach_to_the_same_trace():
+    events = lt.build_batch(
+        repo="o/r", index="1", sha="abc", title="t", model="m",
+        usage={"input": 1, "output": 1}, findings=[], trace_id="fixed-id",
+    )
+    for body in _scores(events).values():
+        assert body["traceId"] == "fixed-id"
+
+
+def test_scores_inherit_the_trace_environment():
+    events = lt.build_batch(
+        repo="o/r", index="1", sha="abc", title="t",
+        model="headroom/glm-5.2:cloud",
+        usage={"input": 1, "output": 1}, findings=[],
+    )
+    for body in _scores(events).values():
+        assert body["environment"] == "ollama"
+
+
+def test_dropped_findings_scored_when_provided():
+    events = lt.build_batch(
+        repo="o/r", index="1", sha="abc", title="t", model="m",
+        usage={"input": 1, "output": 1}, findings=[], dropped_count=3,
+    )
+    assert _scores(events)["dropped_findings"]["value"] == 3.0
+
+
+def test_dropped_findings_absent_when_not_measured():
+    events = lt.build_batch(
+        repo="o/r", index="1", sha="abc", title="t", model="m",
+        usage={"input": 1, "output": 1}, findings=[],
+    )
+    assert "dropped_findings" not in _scores(events)
+
+
+def test_cost_score_carries_its_basis_in_the_comment():
+    # An equivalent-cost $/finding must never be read as money spent.
+    events = lt.build_batch(
+        repo="o/r", index="1", sha="abc", title="t",
+        model="headroom/glm-5.2:cloud",
+        usage={"input": 1000, "output": 100}, findings=[{"severity": "low", "path": "a", "line": 1}],
+    )
+    cpf = _scores(events).get("cost_per_finding")
+    if cpf is not None:  # only when cost_model could price the comparison target
+        assert "equivalent" in cpf["comment"]
+
+
+def test_batch_without_usage_still_scores_findings():
+    # A run with no usage report still produced findings worth scoring.
+    events = lt.build_batch(
+        repo="o/r", index="1", sha="abc", title="t", model="m",
+        usage=None, findings=[{"severity": "critical", "path": "a", "line": 2}],
+    )
+    assert _scores(events)["severity_max"]["value"] == "critical"

tests/pilot/test_parse_dropped.py

@@ -0,0 +1,90 @@
+"""The parse-time drop counter feeding the `dropped_findings` score.
+
+A model that emits findings at unusable locations produces an empty findings
+list, exactly like a model that found nothing. These tests pin the signal that
+tells the two apart.
+"""
+import json
+import os
+import sys
+
+HERE = os.path.dirname(os.path.abspath(__file__))
+sys.path.insert(0, os.path.abspath(os.path.join(HERE, "..", "..", "pilot")))
+
+import ai_review  # noqa: E402
+
+
+def _payload(findings):
+    return "```json\n" + json.dumps({"summary": "s", "findings": findings}) + "\n```"
+
+
+GOOD = {"severity": "high", "path": "a.py", "line": 3, "problem": "p", "fix": "f"}
+NO_PATH = {"severity": "high", "line": 3, "problem": "p"}
+NO_LINE = {"severity": "high", "path": "a.py", "problem": "p"}
+BAD_LINE = {"severity": "high", "path": "a.py", "line": 0, "problem": "p"}
+
+
+def test_no_drops_on_clean_output():
+    _, findings, *_ = ai_review.parse_review_output(_payload([GOOD, GOOD]))
+    assert len(findings) == 2
+    assert ai_review.last_parse_dropped() == 0
+
+
+def test_counts_findings_missing_path():
+    _, findings, *_ = ai_review.parse_review_output(_payload([GOOD, NO_PATH]))
+    assert len(findings) == 1
+    assert ai_review.last_parse_dropped() == 1
+
+
+def test_counts_findings_missing_line():
+    _, findings, *_ = ai_review.parse_review_output(_payload([NO_LINE, NO_LINE]))
+    assert findings == []
+    assert ai_review.last_parse_dropped() == 2
+
+
+def test_counts_findings_with_unusable_line():
+    _, findings, *_ = ai_review.parse_review_output(_payload([BAD_LINE]))
+    assert findings == []
+    assert ai_review.last_parse_dropped() == 1
+
+
+def test_all_dropped_is_distinguishable_from_found_nothing():
+    ai_review.parse_review_output(_payload([NO_PATH, NO_PATH, NO_PATH]))
+    all_dropped = ai_review.last_parse_dropped()
+    ai_review.parse_review_output(_payload([]))
+    found_nothing = ai_review.last_parse_dropped()
+    assert all_dropped == 3 and found_nothing == 0
+
+
+def test_counter_resets_on_unparseable_output():
+    # Otherwise a salvage-path review inherits the previous review's count.
+    ai_review.parse_review_output(_payload([NO_PATH, NO_PATH]))
+    assert ai_review.last_parse_dropped() == 2
+    ai_review.parse_review_output("no json here at all")
+    assert ai_review.last_parse_dropped() == 0
+
+
+def test_counter_resets_on_malformed_json():
+    ai_review.parse_review_output(_payload([NO_PATH]))
+    ai_review.parse_review_output("```json\n{not valid json,,,}\n```")
+    assert ai_review.last_parse_dropped() == 0
+
+
+def test_parse_findings_tracks_drops_too():
+    # The non-opencode path must be scored on the same basis.
+    findings = ai_review.parse_findings(json.dumps({"findings": [GOOD, NO_PATH]}))
+    assert len(findings) == 1
+    assert ai_review.last_parse_dropped() == 1
+
+
+def test_parse_findings_resets_on_garbage():
+    ai_review.parse_findings(json.dumps({"findings": [NO_PATH]}))
+    assert ai_review.last_parse_dropped() == 1
+    ai_review.parse_findings("not json")
+    assert ai_review.last_parse_dropped() == 0
+
+
+def test_bare_array_output_is_counted():
+    _, findings, *_ = ai_review.parse_review_output("```json\n" + json.dumps([GOOD, NO_PATH]) + "\n```")
+    assert len(findings) == 1
+    assert ai_review.last_parse_dropped() == 1