feat: seed a TEPP analysis run through tepp_client (v0.84.0) by seonghobae · Pull Request #111 · ContextualWisdomLab/LineageWeave

seonghobae · 2026-08-16T15:08:06Z

Buyer gap: home Analysis runs only showed lineage reconstruction.

`make seed` writes `analysis_run_tepp` through `tepp_client`
Default transport raises `TeppNotAvailable` → Failed / `tepp_not_available`
Never invents a theta or a local psychometric model
Stacked on feat: apply analysis-run knowledge cutoff to visible posts (v0.83.0) #110. Do not merge before Role, responsibility, and agent ontology (v2.10.0) #74 → feat: apply analysis-run knowledge cutoff to visible posts (v0.83.0) #110.

Verified locally: tepp/seed pytest (4 passed); frontend 43 tests + oxlint.

cursor · 2026-08-16T15:08:10Z

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

coderabbitai · 2026-08-16T15:08:16Z

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: e80a4cad-1a14-4030-ac3b-03a014a88011

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

🔍 Trigger review

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

seonghobae · 2026-08-16T15:08:26Z

Stacked on #110. Do not merge before #74 → #110. Seeded TEPP run uses tepp_client; default transport is Failed / tepp_not_available. Never a fabricated theta.

cursor

Stale comment

The TEPP path is the right ecosystem move: tepp_seed_outcome() calls TeppClient and treats TeppNotAvailable as Failed / tepp_not_available instead of inventing a theta. Keep that. Do not merge this onto #110 at c2c8407.

Do this before the next review:

Rebase onto the #110 repair (#114 at 8fb81d0, or the same two commits on feat/analysis-run-cutoff-posts). #110 still has the shared-fixture post-list break, unlabeled snapshot counts, and the lineage-run re-seed freeze.

Skip frozen analysis_source_count inserts on the TEPP snapshot the same way #114 skips them on the lineage snapshot. ON CONFLICT DO NOTHING still fires BEFORE INSERT freeze.

Move tepp_client and hashlib imports to the module top.

Make AnalysisRunRequest.snapshot_id name the same capture this helper inserts. Today the request says demo-source-contract-v1 while the row is demo-tepp-contract-v1.

Add a doctoring/ADR note that seed records a failed TEPP run when the transport is missing, and that a later live transport must not write a local psychometric substitute.

test_tepp_seed_outcome_is_unavailable_not_a_fake_score is the right shape for the default transport. It does not prove make seed inserted the Failed run. After the rebase, add one authz_db or seed-script assertion that the Demo TEPP run exists with tepp_not_available.

Do not merge before #74 → repaired #110. Product CI does not run on this stacked base.

_{Sent by Cursor Automation: Fix Issues}

cursor · 2026-08-16T15:19:57Z

+    Never invents a psychometric score. ``tepp_not_available`` means the
+    channel was dropped, not a calibrated negative result.
+    """
+    from lineageweave.tepp_client import AnalysisRunRequest, TeppClient, TeppNotAvailable


Move AnalysisRunRequest, TeppClient, and TeppNotAvailable to the module top with the other lineageweave imports. Same for hashlib in _seed_demo_tepp_run (and the existing lineage-run helper). Inline imports are disallowed unless there is a documented circular-import reason.

snapshot_id="demo-source-contract-v1" also does not match the TEPP snapshot this helper inserts (demo-tepp-contract-v1 / lineageweave-synthetic-tepp-snapshot-v1). If a transport is ever supplied, the request will name a different capture than the registry row. Use the same snapshot identity the insert uses.

cursor · 2026-08-16T15:19:57Z

+        insert into analysis_source_count
+            (analysis_source_snapshot_id, count_type_code, count_value)
+        values (%s, 'analysis_count_document', 3)
+        on conflict do nothing


This is the same freeze hole as #110 / #107. BEFORE INSERT on analysis_source_count runs before ON CONFLICT. A second make seed after the TEPP run exists raises analysis_source_count_frozen_after_run.

Select existing counts and skip the insert when the snapshot already has rows — the same pattern as #114's lineage-run seed fix. Do not land this slice on #110 at c2c8407; rebase onto that repair first.

cursor

Stale comment

Review of 87ebe13 (stacked on #110)

The buyer cut is right: home should show a TEPP kind, and a missing transport must be Failed / tepp_not_available, never a fabricated theta. Default TeppClient fail-closed behavior matches that rule.

Do not merge before #74 → #110. #110 still has open cutoff-test, live-list, and "3 documents" badge threads; this PR repeats 3 documents in App.test.tsx.

Requested changes

Do not mark TEPP succeeded without a persisted TEPP payload. tepp_seed_outcome discards submit_analysis_run and returns succeeded if any transport accepts the request. Fail closed instead until a later slice stores the real measurement.

Bind the TEPP request to the snapshot this run persists. snapshot_id="demo-source-contract-v1" is the other seed's contract string, not this run's digest / demo-tepp-contract-v1.

Stop writing dummy d… / e… reproducibility digests. The detail pane from #110 will show them as approved revision/config prefixes.

Prove the no-fake-score contract in a seed/persistence test, not only TeppClient()'s default raise.

Independent review is still required; this comment does not approve or merge.

_{Sent by Cursor Automation: Fix Issues}

cursor · 2026-08-16T15:20:54Z

+        tenant_workspace_id="demo-workspace",
+        snapshot_id="demo-source-contract-v1",
+        knowledge_cutoff="2026-01-12T12:00:00Z",
+        model_contract_version="tepp-analysis-run-v1",


This snapshot_id is the lineage seed's source_contract_version string, not the snapshot this function persists (sha256(lineageweave-synthetic-tepp-snapshot-v1) / demo-tepp-contract-v1). A later real TEPP transport would measure a different capture than the run row. Pass the persisted snapshot digest (or UUID, if that is TEPP's published identifier) into tepp_seed_outcome.

cursor · 2026-08-16T15:20:54Z

+        return "analysis_status_failed", "tepp_not_available"
+    return "analysis_status_succeeded", None
+
+


Any injected transport that does not raise marks the run succeeded and discards submit_analysis_run's return. The home list would then show a successful TEPP measurement with no theta, no calibration, and no stored response — a fabricated completion.

Keep the default TeppNotAvailable → Failed / tepp_not_available path. Do not return succeeded unless a later slice persists the real TEPP payload. Until then, unexpected success should fail closed (still Failed, distinct machine code) rather than look like a calibrated result.

cursor · 2026-08-16T15:20:54Z

+            """,
+            (snapshot_id, requested_by_account_id, "d" * 64, "e" * 40),
+        )
+        run_id = cur.fetchone()[0]


"d" * 64 / "e" * 40 are not the code or configuration that produced this run. #110 already surfaces revision and config prefixes on the detail pane, so make seed would show those dummy hex strings as if they were approved digests.

Use the same digest source as the lineage seed run, or omit prefixes until a real revision is known. Do not invent reproducibility evidence.

cursor · 2026-08-16T15:20:54Z

+from scripts.seed_demo_data import tepp_seed_outcome
+
+
+def test_tepp_seed_outcome_is_unavailable_not_a_fake_score() -> None:


This only proves the process-default TeppClient() raises. It still passes if _seed_demo_tepp_run writes succeeded, stores a dummy theta, or stops calling tepp_client.

Add a seed/Postgres case (or a monkeypatched insert helper) that: (1) default transport persists Failed / tepp_not_available and no measurement payload; (2) a fake successful transport does not persist analysis_status_succeeded without a stored TEPP response. Accuracy here is "no fabricated psychometric result," not "the helper returned a string."

cursor

Stale comment

REQUEST_CHANGES on 87ebe13.

Fail-closed TEPP usage is the right default: TeppClient is called, the missing transport becomes Failed / tepp_not_available, and no theta is invented. Keep that. Do not merge before #74 → #110, and do not merge this head until the items below are fixed.

Fix before merge:

Move the new tepp_client and hashlib imports to the top of scripts/seed_demo_data.py (workspace no-inline-imports rule).

Rewrite the 0.84.0 changelog. The home list caption is kind · status · entity only. After seed it shows TEPP measurement · Failed · Demo Corp. tepp_not_available lives on detail status history (ADR 0014), not the list.

Prove tepp_client was used. The new test passes if tepp_seed_outcome hard-codes the failed tuple. Spy submit_analysis_run and assert the seeded AnalysisRunRequest.

Add a Failed TEPP detail mock and click that row. Today only list text is asserted; opening the TEPP run in this test 404s, so the machine code that distinguishes “not wired” from “measurement failed” is untested.

Also align the seed with ADR 0013 before calling this idempotent: reuse the lineage snapshot (one capture, many runs) instead of demo-source-contract-v1 on the wire and demo-tepp-contract-v1 in Postgres, and do not INSERT counts after a run already points at the snapshot (analysis_source_count_frozen_after_run fires BEFORE INSERT, so ON CONFLICT DO NOTHING still aborts make seed on re-run).

Do not invent a local Unavailable status, and do not fabricate a theta to make the list look green. Next action: apply the four fixes on this stack, then re-request review.

_{Sent by Cursor Automation: Fix Issues}

cursor · 2026-08-16T15:21:01Z

+    Never invents a psychometric score. ``tepp_not_available`` means the
+    channel was dropped, not a calibrated negative result.
+    """
+    from lineageweave.tepp_client import AnalysisRunRequest, TeppClient, TeppNotAvailable


Move this tepp_client import to the module top with psycopg2 / http_client. New code must not add inline imports.

cursor · 2026-08-16T15:21:01Z

+    Uses :func:`tepp_seed_outcome`. Default transport is unavailable, so
+    the run ends Failed / ``tepp_not_available`` -- never a fake theta.
+    """
+    import hashlib


Same rule: import hashlib at the top of the file. _seed_demo_analysis_run already inlines it; do not copy that pattern here.

cursor · 2026-08-16T15:21:01Z

+    request = AnalysisRunRequest(
+        idempotency_key="demo-tepp-seed-2026-w02",
+        tenant_workspace_id="demo-workspace",
+        snapshot_id="demo-source-contract-v1",


This wire snapshot_id is the lineage contract version, but the inserted snapshot uses demo-tepp-contract-v1 and a different digest. Reuse the lineage snapshot (ADR 0013: one capture, many runs) or set snapshot_id to the persisted snapshot identity so a later real transport submits the capture that seed stored.

cursor · 2026-08-16T15:21:01Z

+    else:
+        snapshot_id = snapshot_row[0]
+    cur.execute(
+        """


After the first seed, a run already points at this snapshot. enforce_analysis_source_count_freeze runs BEFORE INSERT, so ON CONFLICT DO NOTHING still raises analysis_source_count_frozen_after_run and rolls back the whole seed() transaction. Skip the count insert when the snapshot already has counts or is referenced by a run.

cursor · 2026-08-16T15:21:01Z

+from scripts.seed_demo_data import tepp_seed_outcome
+
+
+def test_tepp_seed_outcome_is_unavailable_not_a_fake_score() -> None:


This passes if the function hard-codes analysis_status_failed / tepp_not_available. Spy TeppClient.submit_analysis_run (or inject a transport) and assert it was called with the seeded AnalysisRunRequest. A hardcoded skip is not proof the client was used.

cursor · 2026-08-16T15:21:01Z

+
+- `make seed` records a Demo Corp TEPP measurement run through
+  `tepp_client`. The default transport is unavailable, so the home
+  list shows "TEPP measurement · Failed · Demo Corp" with


The list caption is kind · status · entity only (analysisRunCaption). After seed the home list shows TEPP measurement · Failed · Demo Corp and does not show tepp_not_available. Say the list shows Failed, and that detail history keeps the machine code.

cursor · 2026-08-16T15:21:01Z

                  },
                ],
              },
+              {


This list fixture has no matching GET /api/analysis-runs/run-demo-tepp stub. The lineage-only detail mock is at /api/analysis-runs/run-demo-lineage. Clicking the TEPP row in this test 404s.

cursor · 2026-08-16T15:21:01Z

    expect(await screen.findByRole("heading", { name: "Analysis runs" })).toBeInTheDocument();
    const list = screen.getByRole("list", { name: "Analysis runs" });
    expect(list).toHaveTextContent("Lineage reconstruction · Succeeded · Demo Corp");
+    expect(list).toHaveTextContent("TEPP measurement · Failed · Demo Corp");


Only list text is new. Click the Failed TEPP row and assert status history includes tepp_not_available. That is the buyer-visible difference between “channel dropped” and “measurement failed.”

Buyer gap: home Analysis runs only showed lineage reconstruction. make seed now records a Demo Corp TEPP measurement via tepp_client. The default transport is unavailable, so the row is Failed / tepp_not_available -- never a fabricated theta. TEPP stays a wire client, not a local psychometric engine.

cursor

REQUEST_CHANGES on 34fdb892 (now stacked on #74, not #110).

Retargeting did not fix the seed holes. Fail-closed TEPP usage is still the right default — keep calling TeppClient and never invent a theta. Do not merge this head.

Land #118 instead. That repair reuses the shared Demo Corp snapshot, skips frozen count inserts, keeps a live unused envelope Failed / tepp_result_not_persisted, spies submit_analysis_run, and opens the Failed TEPP row so detail history shows tepp_not_available.

Do not merge #115: it still targets feat/analysis-run-cutoff-posts. Do not invent a local psychometric substitute. Independent review is still required; this comment does not approve or merge.

Next action: review and land #118 on #74, then close #111 and #115 as superseded.

_{Sent by Cursor Automation: Fix Issues}

cursor · 2026-08-16T15:32:02Z

+    Never invents a psychometric score. ``tepp_not_available`` means the
+    channel was dropped, not a calibrated negative result.
+    """
+    from lineageweave.tepp_client import AnalysisRunRequest, TeppClient, TeppNotAvailable


Move AnalysisRunRequest, TeppClient, and TeppNotAvailable to the module top. New code must not add inline imports.

snapshot_id="demo-source-contract-v1" is the lineage contract string, not the snapshot this helper inserts (demo-tepp-contract-v1). Pass the shared Demo Corp digest (sha256(lineageweave-synthetic-demo-snapshot-v1)). That repair is on #118.

cursor · 2026-08-16T15:32:02Z

+    except TeppNotAvailable:
+        return "analysis_status_failed", "tepp_not_available"
+    return "analysis_status_succeeded", None
+


Any injected transport that does not raise marks the run Succeeded and discards the envelope. The home list would then show a successful TEPP measurement with no stored payload.

Keep TeppNotAvailable → Failed / tepp_not_available. Until a later slice persists the real measurement, an unexpected accept must stay Failed (tepp_result_not_persisted).

cursor · 2026-08-16T15:32:02Z

+        insert into analysis_source_count
+            (analysis_source_snapshot_id, count_type_code, count_value)
+        values (%s, 'analysis_count_document', 3)
+        on conflict do nothing


enforce_analysis_source_count_freeze runs BEFORE INSERT, so ON CONFLICT DO NOTHING still raises analysis_source_count_frozen_after_run on a second make seed. Select existing counts and skip the insert when the snapshot already has rows.

cursor · 2026-08-16T15:32:02Z

+from scripts.seed_demo_data import tepp_seed_outcome
+
+
+def test_tepp_seed_outcome_is_unavailable_not_a_fake_score() -> None:


This passes if tepp_seed_outcome hard-codes the failed tuple. Spy submit_analysis_run and assert the seeded AnalysisRunRequest. Also prove an accepting transport does not persist analysis_status_succeeded.

cursor · 2026-08-16T15:32:02Z

+
+- `make seed` records a Demo Corp TEPP measurement run through
+  `tepp_client`. The default transport is unavailable, so the home
+  list shows "TEPP measurement · Failed · Demo Corp" with


The list caption is kind · status · entity only. After seed the home list shows TEPP measurement · Failed · Demo Corp and does not show tepp_not_available. Say the list shows Failed and the next action; detail history keeps the machine code.

cursor · 2026-08-16T15:32:02Z

                  },
                ],
              },
+              {


This list fixture has no matching GET /api/analysis-runs/run-demo-tepp stub. Clicking the TEPP row 404s. Add the Failed detail (status history includes tepp_not_available) and click that row.

* feat: seed a TEPP analysis run through tepp_client (v0.84.0) Buyer gap: home Analysis runs only showed lineage reconstruction. make seed now records a Demo Corp TEPP measurement via tepp_client. The default transport is unavailable, so the row is Failed / tepp_not_available -- never a fabricated theta. TEPP stays a wire client, not a local psychometric engine. * fix: fail-closed TEPP seed on the shared Demo Corp snapshot #111 still marked a live unused envelope Succeeded, named a different capture than the registry row, and re-inserted frozen counts. Seed now reuses the lineage snapshot (ADR 0013), skips count inserts after the first run, and keeps missing or unused TEPP Failed. The home list tells the operator to open the run and connect TEPP; detail history keeps tepp_not_available. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com> --------- Co-authored-by: Seongho Bae <me@seonghobae.me> Co-authored-by: Cursor Agent <cursoragent@cursor.com> Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

* feat: seed a TEPP analysis run through tepp_client (v0.84.0) Buyer gap: home Analysis runs only showed lineage reconstruction. make seed now records a Demo Corp TEPP measurement via tepp_client. The default transport is unavailable, so the row is Failed / tepp_not_available -- never a fabricated theta. TEPP stays a wire client, not a local psychometric engine. * fix: fail-closed TEPP seed on the shared Demo Corp snapshot #111 still marked a live unused envelope Succeeded, named a different capture than the registry row, and re-inserted frozen counts. Seed now reuses the lineage snapshot (ADR 0013), skips count inserts after the first run, and keeps missing or unused TEPP Failed. The home list tells the operator to open the run and connect TEPP; detail history keeps tepp_not_available. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com> * fix: keep failed-run next actions kind-specific A failed lineage row must not tell the operator to connect TEPP. Stacked PRs now run the same GitHub Checks as PRs to main. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com> * docs: keep TEPP next-action copy off failed lineage rows Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com> * fix: keep TEPP corpus hint off a succeeded measurement A calibrated TEPP row must not tell the operator to replace Failed. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com> --------- Co-authored-by: Seongho Bae <me@seonghobae.me> Co-authored-by: Cursor Agent <cursoragent@cursor.com> Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

seonghobae mentioned this pull request Aug 16, 2026

feat: apply analysis-run knowledge cutoff to visible posts (v0.83.0) #108

Closed

cursor Bot mentioned this pull request Aug 16, 2026

feat: add normalized analysis-run registry (Milestone 2.1, v0.76.0) #105

Draft

cursor Bot requested changes Aug 16, 2026

View reviewed changes

cursor Bot mentioned this pull request Aug 16, 2026

fix: seed TEPP on the shared snapshot and prove the client (v0.84.0) #115

Closed

seonghobae changed the base branch from feat/analysis-run-cutoff-posts to feat/role-responsibility-agent-ontology August 16, 2026 15:24

seonghobae force-pushed the feat/analysis-run-tepp-seed branch from 87ebe13 to 34fdb89 Compare August 16, 2026 15:25

cursor Bot mentioned this pull request Aug 16, 2026

fix: fail-closed TEPP seed on the shared Demo Corp snapshot #118

Merged

cursor Bot requested changes Aug 16, 2026

View reviewed changes

seonghobae marked this pull request as draft August 16, 2026 15:33

cursor Bot mentioned this pull request Aug 16, 2026

fix: keep failed-run next actions kind-specific #124

Merged

seonghobae closed this Aug 16, 2026

seonghobae mentioned this pull request Aug 16, 2026

Role, responsibility, and agent ontology (v2.10.0) #74

Open

5 tasks

cursor Bot mentioned this pull request Aug 16, 2026

fix: pin failed-run next actions to registered kinds #128

Merged

cursor Bot mentioned this pull request Aug 16, 2026

feat: let operators request a pending lineage reconstruction (v0.85.0) #133

Closed

This was referenced Aug 16, 2026

Milestone 2: port verified direct-PostgreSQL analysis into the protected product architecture #79

Open

fix: seed Late Demo public post after the analysis-run cutoff (v2.9.1) #242

Draft

		return "analysis_status_failed", "tepp_not_available"
		return "analysis_status_succeeded", None

		from scripts.seed_demo_data import tepp_seed_outcome


		def test_tepp_seed_outcome_is_unavailable_not_a_fake_score() -> None:

Uh oh!

Conversation

seonghobae commented Aug 16, 2026

Uh oh!

cursor Bot commented Aug 16, 2026

Uh oh!

coderabbitai Bot commented Aug 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review skipped

Uh oh!

seonghobae commented Aug 16, 2026

Uh oh!

cursor Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Review of 87ebe13 (stacked on #110)

Requested changes

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

cursor Bot Aug 16, 2026

Choose a reason for hiding this comment

Uh oh!

coderabbitai Bot commented Aug 16, 2026 •

edited

Loading

cursor Bot left a comment •

edited

Loading

cursor Bot left a comment •

edited

Loading

Review of `87ebe13` (stacked on #110)

cursor Bot left a comment •

edited

Loading