Skip to content

spec(t27f): petal D HEALTH -- matrix, list and badge from dashboard.t27 (Closes #12322, Refs #12173) - #12324

Merged
github-actions[bot] merged 1 commit into
masterfrom
claude/viz-petal-health
Oct 25, 2026
Merged

github-actions[bot] merged 1 commit into
masterfrom
claude/viz-petal-health

Conversation

@gHashTag

Copy link
Copy Markdown
Collaborator

Closes #12322
Refs #12173

What

Petal D HEALTH, row 3 of specs/t27f/petal_viz_plan.t27. It answers: "Which project checks pass now, and which one should I open first?"

One new file, specs/t27f/petal_health_viz.t27. It is written against viz_contract.t27 in the order of the reference petal S and follows docs/nona-03-manifest/VIZ-PETAL-GUIDE.md (#12282).

  • Source, not copies. The rules come from specs/t27mcp/dashboard.t27, the session dashboard of the t27 MCP server:

    Shown Source
    Grade of each measured check health_tone, calls_tone, reply_tone
    Severity join worst (and worst(OK, UNKNOWN) is UNKNOWN)
    Which check to open first next_step (N_ERRORS .. N_READY)
    Titles, grade names, unknown text PANEL_TITLES, TONE_NAME, UNKNOWN_TEXT
  • Views. A matrix (six checks by ok, unknown, warn, bad; one cell lit per check), a list ordered by what to open first, and a badge, all over one VizState. Every contract event goes through viz_apply.

  • Inputs. A flip chip on each of the three measured checks (corpus, session, reply) moves that check from its recorded reading to the what-if the source's own tests grade. Example: the session flips to 4 calls with 1 error, its grade falls to bad, next_step answers N_ERRORS, the session moves to the top of the list and the badge drops to DEMO. The flip is the one petal event, numbered above VIZ_EV_COUNT.

  • Plan. pdv_reps() equals pvp_reps(3) = 262, and viz_plan_ok holds. The plan is not edited.

Glue the petal owns, called out

  • Recorded fixture. dashboard.t27 keeps no data: it grades numbers the server counts at run time. So the readings are numbers the petal records, the ones the source's tests grade (2402 specs, 113 failing, 257 warning; 7 calls, no error; a 57,084-byte t27_status reply), and their what-ifs (121 failing; 4 calls, 1 error; a reply of exactly REPLY_WARN_BYTES). Nothing is read from a running server. The header says so.
  • Order. The source picks one next step. The petal builds the list by asking again with each pick held fixed (up to three times), then sorts the rest by severity.
  • Unknown checks. Seal frontier, test receipts and CI state are always UNKNOWN here, and UNKNOWN does not count as passing.
  • Not used. The HTML templates (PAGE_TEXT, PANEL_TEXT, ...) are the server's own page; the petal draws its own views. No host draws this petal yet.

Checks

  • t27c test-report specs/t27f/petal_health_viz.t27 --specs-dir specs: 31 tests, 31 pass, 0 fail, 0 vacuous, 1 invariant.
  • python3 tools/dupe_scan.py: no new duplicate body.
  • ASCII only, no home paths, one file changed.

🤖 Generated with Claude Code

…27 (Closes #12322, Refs #12173)

Petal D, row 3 of specs/t27f/petal_viz_plan.t27: which project checks
pass now, and which one to open first.

A matrix of six checks by four statuses (ok, unknown, warn, bad) is
graded by health_tone, calls_tone and reply_tone of
specs/t27mcp/dashboard.t27; the list is ordered by next_step, asked
again with each pick held fixed, then by severity. The three checks
the page cannot know read UNKNOWN, and worst(OK, UNKNOWN) is UNKNOWN,
so UNKNOWN never counts as passing. A chip per measured check flips
its reading to the what-if the source's own tests grade, and the badge
drops from VERIFIED to DEMO.

dashboard.t27 keeps no data, so the readings are a recorded fixture
(the numbers the source's tests grade), stated in the header. Nothing
is read from a running server.

Reps 262 = matrix + list + badge, equal to pvp_reps(3).

test-report: 31 tests, 31 pass, 0 fail, 0 vacuous, 1 invariant.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-25 07:16:00 UTC

Summary

Status Count
Total Open PRs 50
PRs with Failing Checks 25
PRs with All Checks Green 25
READY 12
FAILING 25
PENDING 0
NO CHECKS YET 0

These columns do not partition: 12 + 25 + 0 + 0 = 37, and there are 50 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=55c0f1b1b32d != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

@tri-claw

tri-claw Bot commented Oct 25, 2026

Copy link
Copy Markdown

Summary by t27-bees

Checked by the t27 compiler

t27c parse, parse-complete and typecheck on each changed .t27 file at 1b86f06. Nothing was generated, built or run. A pass means the file is well formed, not that it is right.

  • specs/t27f/petal_health_viz.t27: parses and typechecks.

Read by the model

This PR adds specs/t27f/petal_health_viz.t27, a new petal (D HEALTH) that visualizes project health checks from the MCP dashboard. It provides a matrix, a priority-ordered list, and a badge driven by recorded fixture data and what-if flip chips for corpus, session, and reply checks.

Worth a look

  • specs/t27f/petal_health_viz.t27 (entire file): The diff was truncated ("Diff (cut: the rest was not shown)"), so I cannot verify the implementation details against the described behavior (matrix/list/badge views, flip-chip semantics, viz_apply wiring, unknown-check handling, or the recorded fixture values).
  • specs/t27f/petal_health_viz.t27 (plan conformance): The description claims pdv_reps() == pvp_reps(3) == 262 and viz_plan_ok holds, but without seeing the pdv_reps and viz_plan_ok definitions in the diff, I cannot confirm the plan math is correct.
  • specs/t27f/petal_health_viz.t27 (source coupling): The petal reads constants (PANEL_TITLES, TONE_NAME, UNKNOWN_TEXT, health_tone, calls_tone, reply_tone, worst, next_step) from specs/t27mcp/dashboard.t27. If that source file changes its API or grading logic, this petal could silently show stale or wrong grades.
  • specs/t27f/petal_health_viz.t27 (unknown severity): The description notes "UNKNOWN does not count as passing" and that three checks are always UNKNOWN. If the matrix/list/badge treat UNKNOWN as neutral rather than a distinct severity, users may misinterpret project health.
  • specs/t27f/petal_health_viz.t27 (flip-chip boundaries): The flip chips move checks to "what-if the source's own tests grade." Without seeing the event wiring and grade recomputation, I cannot verify that flips correctly propagate through worst/next_step and re-sort the list.

Read: 1 files, +1790 -0, the whole diff up to 60000 characters (the rest was not read). Head 1b86f06. Model: openai-compatible/nvidia/nemotron-3-ultra-550b-a55b. Commands: @t27-bees review | summary | help | pause | resume. Free for every repository.

This was referenced Oct 25, 2026
This was referenced Oct 25, 2026
@github-actions
github-actions Bot merged commit ed677d8 into master Oct 25, 2026
37 checks passed
@gHashTag
gHashTag deleted the claude/viz-petal-health branch October 26, 2026 03:42
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

spec(t27f): petal D HEALTH interactive visualization (petal_health_viz.t27)

1 participant