feat(board): Pulse strip + profiling conductor reads the Pulse board (phase 3, PyAutoBrain#450)

agents/conductors/profiling/AGENTS.md

@@ -110,7 +110,19 @@ as malformed — only a record missing *some* of its key fields is corruption.
   (`triage`) stays the only judge of what a timing or a drift means, with the
   human. The projects keep their producers, results, pins and drift policy.
   Same split as Heart ↔ vitals and Eyes ↔ the Eyes conductor. (Organ row
-  born in PyAutoMind#463; its reader and board arrive in phase 2.)
+  born in PyAutoMind#463; its reader and board arrived in phase 2.)
+- **Where drift candidates are read** — the
+  [Pulse board](https://pyautolabs.github.io/PyAutoPulse/) (strip on the
+  Brain board's Resume section, card on the organ cockpit) is where
+  cross-project drift candidates and their provenance (the resolved commit
+  per project, the pair of records, the evidence paths) are read.
+  `autolens_profiling` owns the measurement semantics: its producers, pins
+  and drift policy. A `/profiling triage <comparison_key>` prompt copied from
+  a Pulse drift item names one comparison row on the Pulse board: open that
+  row, read the pair of records and their evidence paths, then judge with the
+  `triage` verb. The key is a pointer for the human/agent reading the board —
+  `triage` itself takes no target argument and still sweeps every pinned
+  finding.
 - **vs build** — campaigns are not releases; `profile.yml`'s on-release runs
   stay CI/Build territory.
 - **release-validation script cost is NOT ours.** The compile axis was

board/AGENTS.md

@@ -21,7 +21,7 @@ copy-for-Claude payload:
 | ⏱ Test performance | the Heart board's published `performance` block — rendered verbatim, never re-derived | each row's own prompt |
 | 🏷️ Version consistency | the coupled-set stamps (`board: version_stamps`) | `/bug version drift: …` |
 | 💬 Community | the Ears (`community scan`, reused wholesale) — every open conversation gets a row | `/community`, `/community triage <ref>` |
-| 🔄 Resume | the Mind's registry + generated counts; pending-release PRs | `/start_dev …`, `/prm <url>` |
+| 🔄 Resume | the Mind's registry + generated counts; the Cortex, Eyes and Pulse strips (each organ's own `dashboard.md` head counts, display-only, never in `state.json`); pending-release PRs | `/start_dev …`, `/prm <url>` |
 | 🧹 Upkeep | open-issue count; the cleanup doors | `/issue_cleanup`, `/hygiene`, `/repo_cleanup` |
 | 🧼 Hygiene | the hygiene conductor's own `--json` pre-scan, run IN this render (BOARD_HYGIENE_SCAN=1; brain_board.yml checks out the body-map scan set blobless+sparse first) — no machine involved | each row's own delegate door |
 | 🖥️ Dev box | `state/devbox_board.json` — worktree state (unpushed/dirty/stashes — the one thing only the dev box can see; its hygiene rows render only when no cloud scan ran), pushed by `board publish` (morning.sh's last step); age-stamped, stale at 48h, dropped after 7d | — |

board/_board.py

@@ -725,6 +725,21 @@ def collect_eyes(repo="PyAutoEyes"):
     return _dashboard_counts(repo, head_only=True)
 
 
+def collect_pulse(repo="PyAutoPulse"):
+    """The Pulse's own generated counts (projects, records, comparisons,
+    drifted, refused pairs, cached, failed — the counts table at the head of
+    its `dashboard.md`), or None.
+
+    Composed exactly as the Eyes strip is: the organ's renderer decides the
+    numbers, the Brain board only shows them — a drifted count is the
+    producer's say-so, never a Brain verdict (judgment is the profiling
+    conductor's `triage`). Only the head of the page is read, because the
+    body carries one table per project. No Pulse checkout, or no rendered
+    board, is not a degraded morning either — the strip just does not appear.
+    """
+    return _dashboard_counts(repo, head_only=True)
+
+
 def collect_open_issues(org, degraded):
     """Total open issues across the org — the /issue_cleanup pointer count
     (the audit itself stays that skill's confirmation-gated job)."""
@@ -932,6 +947,7 @@ def collect():
     resume = collect_resume(org, degraded)
     cortex = collect_cortex()
     eyes = collect_eyes(board_family.get("eyes", "PyAutoEyes"))
+    pulse = collect_pulse(board_family.get("pulse", "PyAutoPulse"))
     open_issues = collect_open_issues(org, degraded)
     boards = {name: f"{pages_base}/{repo}/"
               for name, repo in board_family.items()}
@@ -952,6 +968,7 @@ def collect():
         "resume": resume,
         "cortex": cortex,
         "eyes": eyes,
+        "pulse": pulse,
         "open_issues": open_issues,
         "hygiene": collect_hygiene(degraded),
         "devbox": collect_devbox(),
@@ -1343,6 +1360,11 @@ def render_md(data):
                  + " · ".join(f"{k.lower()} {n}"
                               for k, n in data["eyes"].items())
                  + f" — [Eyes board]({data['boards'].get('eyes', '')})")
+    if data.get("pulse"):
+        L.append("- Pulse: "
+                 + " · ".join(f"{k.lower()} {n}"
+                              for k, n in data["pulse"].items())
+                 + f" — [Pulse board]({data['boards'].get('pulse', '')})")
     for t in data["resume"]["tasks"]:
         L.append(f"  - `Use the start-dev skill. {t['path']}` — {t['title'][:70]}")
     pending = data["resume"]["pending_prs"]
@@ -1727,6 +1749,13 @@ def community_row(e, note, tone):
             "board \u2197</a>"
             + pills(*[(f"{k.lower()} {n}", "" if i == 0 else "n")
                       for i, (k, n) in enumerate(data["eyes"].items())])))
+    if data.get("pulse"):
+        pulse_url = data["boards"].get("pulse", "")
+        H.append(_plain(
+            f'Run times on the <a href="{_attr(pulse_url)}">Pulse '
+            "board \u2197</a>"
+            + pills(*[(f"{k.lower()} {n}", "" if i == 0 else "n")
+                      for i, (k, n) in enumerate(data["pulse"].items())])))
     for t in data["resume"]["tasks"]:
         # An in-flight task wears the header facets the Mind gave it, so it
         # looks like itself on both pages — same pills, same order.

docs/organs/pulse.md

@@ -12,7 +12,8 @@ the Eyes: the Eyes show what the software *shows*; the Pulse feels how fast it
 **Repo:** [PyAutoLabs/PyAutoPulse](https://github.com/PyAutoLabs/PyAutoPulse)
 
 **Status:** organ row registered (phase 0 of the `profiling-organ-birth`
-epic); the registry, reader, receipts, board and workflows arrive in phase 2.
+epic); the registry, reader, receipts, board and workflows arrived in phase 2;
+phase 3 puts the organ on the Brain board and the organ cockpit.
 Design: `docs/research/profiling_inference_organs.md` in the Brain repo.
 
 ## The defining function: feeling the pulse
@@ -41,6 +42,13 @@ The board shows one row per registered project with scope, evidence time, last
 fetch, coverage and links, and keeps missing or refused evidence visible beside
 valid data. A valid empty feed says "no measurements", never "all passed".
 
+Beside the board, the organ publishes two machine surfaces: `badge.json`, a
+one-line headline, and `state.json`, the organ-cockpit feed (contract v1 in
+`PyAutoBrain/board/state_schema.json`). The feed is what puts the Pulse on the
+[organ cockpit](https://pyautolabs.github.io/cockpit/) alongside the other
+organs; the Brain board's Resume section carries a Pulse strip composed from
+the head counts of the organ's own `dashboard.md`.
+
 ## The driver split
 
 The project repos *produce and hold* timings and the Pulse *shows* them;
@@ -76,3 +84,7 @@ The organ is being born through the `profiling-organ-birth` epic:
   the lens profiling project.
 - **Phase 0** ([PyAutoMind#463](https://github.com/PyAutoLabs/PyAutoMind/issues/463))
   added the organ row to the body map.
+- **Phase 3** ([PyAutoBrain#450](https://github.com/PyAutoLabs/PyAutoBrain/issues/450))
+  added the Pulse strip to the Brain board and the Pulse card to the organ
+  cockpit, and named the Pulse board in the profiling conductor's prose as
+  where drift candidates are read.

skills/profiling/profiling.md

@@ -21,6 +21,14 @@ and board show how fast each `<lib>_profiling` project runs, but it never
 judges. This command's `triage` is the only judge of what a timing or a drift
 means; the projects keep their producers, results, pins and drift policy.
 
+The Pulse board (strip on the Brain board, card on the organ cockpit) is where
+cross-project drift candidates and their provenance are read; `autolens_profiling`
+owns the producers, pins and drift policy. A `/profiling triage <comparison_key>`
+prompt copied from a Pulse drift item names one comparison row on the Pulse
+board: open that row, read the pair of records and their evidence paths, then
+judge with the `triage` verb (which takes no target — the key is where to look,
+not an argument).
+
 The Profiling Agent **reasons; it never runs sweeps or edits source.** The
 classification is the result for CPU-unusable cells; full timings for those
 belong to the A100 rows.

tests/test_board.py

@@ -37,6 +37,9 @@
     # The Eyes organ's own counts (the head of its dashboard.md), None when
     # no Eyes is checked out.
     "eyes",
+    # The Pulse organ's own counts (the head of its dashboard.md), None when
+    # no Pulse is checked out.
+    "pulse",
 }
 
 AUTONOMY_LOG = """\
@@ -208,6 +211,26 @@ def _default_fixtures(**overrides):
 | [fit.png](https://example.invalid/fit.png) | 5 |
 """
 
+# The head of the Pulse's generated board is its counts table; its body
+# carries one table per project, which the strip must not count.
+PULSE_DASHBOARD_MD = """\
+# PyAutoPulse — profiling dashboard
+
+| Where | Count |
+|-------|------:|
+| [Projects](#projects) | 1 |
+| [Records](#projects) | 159 |
+| [Comparisons](#projects) | 145 |
+| [Drifted](#projects) | 0 |
+| [Refused pairs](#projects) | 0 |
+| [Cached](#projects) | 0 |
+| [Failed](#projects) | 0 |
+
+## Projects
+
+| [alpha_profiling](#alpha_profiling) | 7 | 9 |
+"""
+
 
 def _fabricate(tmp_path, fixtures, heart_board=None):
     """A PYAUTO_ROOT with a fabricated Mind, file:// sibling-board badges, and
@@ -243,6 +266,9 @@ def _fabricate(tmp_path, fixtures, heart_board=None):
     eyes = tmp_path / "PyAutoEyes"
     eyes.mkdir()
     (eyes / "dashboard.md").write_text(EYES_DASHBOARD_MD)
+    pulse = tmp_path / "PyAutoPulse"
+    pulse.mkdir()
+    (pulse / "dashboard.md").write_text(PULSE_DASHBOARD_MD)
 
     # One badge per sibling board named in the declared config surface —
     # read from policy.yaml so no board repo name is hardcoded here.
@@ -357,6 +383,28 @@ def test_the_eyes_strip_composes_the_eyes_own_head_counts(tmp_path):
     assert json.loads(r.stdout)["eyes"] is None
 
 
+def test_the_pulse_strip_composes_the_pulse_own_head_counts(tmp_path):
+    """The Pulse decides its own numbers; the Brain board reads only the
+    counts table at the head of its page (the body's per-project tables are
+    not counts), links its board, and shows nothing when no Pulse is here."""
+    s, _ = _surface(tmp_path)
+    assert s["pulse"] == {"Projects": 1, "Records": 159, "Comparisons": 145,
+                          "Drifted": 0, "Refused pairs": 0, "Cached": 0,
+                          "Failed": 0}
+    assert s["boards"]["pulse"].endswith("/PyAutoPulse/")
+    stub = _fabricate(tmp_path / "md", _default_fixtures())
+    md = _run([], tmp_path / "md", stub).stdout
+    assert ("- Pulse: projects 1 · records 159 · comparisons 145 · "
+            "drifted 0 · refused pairs 0 · cached 0 · failed 0") in md
+    page = _run(["--html"], tmp_path / "md", stub)
+    assert page.returncode == 0, page.stderr
+    assert "Pulse board" in page.stdout and 'data-organ="pulse"' in page.stdout
+    stub = _fabricate(tmp_path / "no_pulse", _default_fixtures())
+    (tmp_path / "no_pulse" / "PyAutoPulse" / "dashboard.md").unlink()
+    r = _run(["--json"], tmp_path / "no_pulse", stub)
+    assert json.loads(r.stdout)["pulse"] is None
+
+
 def test_json_surface_is_complete_and_derives_org(tmp_path):
     s, _ = _surface(tmp_path)
     assert set(s) == SURFACE_KEYS