Skip to content

952f3aa3.md

Latest commit

 

History

History
239 lines (176 loc) · 17.7 KB

952f3aa3.md

File metadata and controls

239 lines (176 loc) · 17.7 KB
Raw
uid 952f3aa3
type project
subsystem_name tropo-test-harness
title Tropo Test Harness
description First-class Tropo-OS subsystem for release-cycle test infrastructure. Hosts the canonical discovery surface for the four-primitive test harness Vela authored during the v1.40-v1.42 cycle arc: vela-test-plan (cycle-specific substrate verification; Argus-authored per cycle) + release-cold-boot-walk (user-experience first-encounter persona walks; ships in every release) + release-tropo-work-walk (PROPOSED — user-experience tropo-work depth walks; extends personas with scenario tails first per earn-the-abstraction-strict) + feature-specific test cases (PROPOSED — persistent per-feature mechanical verification; Mike-V47 pinned idea 2026-05-18). Each component lives at its canonical type-location (playbooks at .tropo/playbooks/; personas at .tropo/personas/; test cases will live at .tropo/test-cases/; per-cycle runs at argo-os/playbook-runs/; aggregate reports at /Users/mike/dev/tropo-releases/v<X.Y.Z>/); this hub indexes the substrate as a coherent subsystem. Elevated to subsystem-class per Mike-V47 directive 2026-05-18 — load-bearing release-cycle infrastructure that compounds across every release; deserves first-class subsystem treatment alongside Tropo Work, Tropo Governance, Tropo Agents.
status active
state active
owner vela
author vela-v47
created 2026-05-18
created_by vela-v47
modified 2026-05-18
modified_by vela-v47
schema_version 2
extraction_scope ship
governed_by 8dd772a0
member_of
aae9a37b
aligned_with
6f3d2a18
a7c9d4e2
7b69bd19
839a65f9
0aefe71d
provenance Authored 2026-05-18 by Vela V47 in Chief-of-Staff mode per Mike-V47 directive 2026-05-18 'Tropo Test Harness is a legit piece of first-class harness; include it in tropo-subsystems.' V47 initial proposal pulled back from subsystem-class elevation per earn-the-abstraction-strict reading; Mike overrode — release-cycle test infrastructure is substantive enough to deserve subsystem-class home. Calibration: when substrate is load-bearing AND compounds across releases AND has multi-component composability, elevation to subsystem-class is correct posture, not over-investment.
tags
tropo-subsystems
tropo-test-harness
subsystem-hub
first-class
release-cycle-infrastructure
harness
vela-owned
mike-V47-elevated
substrate-quality-leverage
lifecycle standing

Tropo Test Harness

📍 Vault Path: tropo-subsystems → Tropo Test Harness

🌳 Tropo-Nav Path (VS Code): ../../00-tropo-nav/00-tropo-active/tropo-subsystems/Tropo Test Harness/952f3aa3 — Tropo Test Harness.md

🌳 Tropo-Nav Path (chat): argo-os/00-tropo-nav/00-tropo-active/tropo-subsystems/Tropo Test Harness/952f3aa3 — Tropo Test Harness.md

🔗 This file — UID 952f3aa3 · type project · state active · status active

↔ Siblings (10):

📥 Cited by (2):

Relations

Relation Target
Governed by Tropo Governance (8dd772a0)
Aligned with Release Cold-Boot Walk (6f3d2a18)
Aligned with Deletion Discipline — Substrate Preservation Doctrine (0aefe71d)
Member of tropo-subsystems (aae9a37b)

First-class Tropo-OS subsystem for release-cycle test infrastructure. Indexes the four-primitive test harness — substrate verification + first-encounter UX + tropo-work depth + per-feature mechanical verification — that gates release ships.

§What This Subsystem Is

The Tropo Test Harness is the substrate that answers four distinct release-quality questions:

  1. Is the substrate well-formed? (form-check; cycle-specific; Argus-authored vela-test-plans)
  2. Can a stranger encounter the Studio and get to their first agent? (user-experience first-encounter; cycle-comparable; persona walks)
  3. Can a user exercise Tropo Work primitives at depth? (user-experience usage-pattern; cycle-comparable; persona walks with scenario tails)
  4. Do specific features still work mechanically? (regression verification; persistent per-feature; feature-specific test cases)

Each primitive runs as a release-step gate. Together they form a complete quality envelope: form-check + first-touch UX + depth-of-use UX + per-feature regression. No single primitive is sufficient; together they are.

The harness is named at this hub; the components live at their canonical type-locations (playbooks live where playbooks live; personas live where personas live; test cases will live in their own kernel folder). This hub is the discovery surface + the doctrine surface — what the harness IS, why each component composes the way it does, and where each piece lives.

§The Four Components (Current + Proposed)

1. vela-test-plan (PRODUCTION since v1.5)

Status: active across every cycle. Type: note with note_subtype: vela-test-plan. Authored by: Argus (or the cycle's substrate author) per release. Executed by: Vela within 24h of release ship. Lives at: vault/files/<uid>.md per release. Tests: cycle-specific substrate verification — every ship-class change has a test against it.

Example: v1.41.0 vela-test-plan (2d2403fe), v1.42.0 vela-test-plan (408b4a25).

2. release-cold-boot-walk (PRODUCTION since v1.41)

Status: active; v1.1 amendment landed 2026-05-18 absorbing per-persona Studio clones + Step 1 collapse + verify-aggregate discipline. Type: playbook. Authored by: Vela V47 (Mike-V47 walk; cycle-comparable per dogfood discipline). Executed by: Vela per release; Mike-conditional gate. Lives at: .tropo/playbooks/release-cold-boot-walk.playbook.md (UID 6f3d2a18). Persona substrate at: .tropo/personas/cold-boot/ (CAPSULE a7c9d4e2 + engineer/operator/enterprise persona files; Studios may amend or add). Tests: user-experience first-encounter — three persona sub-agents walk fresh Studios, create first agents, ask questions, score the experience.

Dogfooded twice: v1.41 (baseline avg 7.33/10) + v1.42 (confirmatory 7.33/10; zero net movement signaling v1.42 was correctly out-of-scope for UX movement). Both walks surfaced two real Stream-E regressions (channels/ + system/ folders missing) that vela-test-plan didn't catch — substantive ship-defect-surfacing earning the harness's keep.

3. release-tropo-work-walk (PROPOSED; extension of #2 first per Mike-V47 directive 2026-05-18)

Status: proposed; not authored. Path: start by extending the existing personas with Tropo Work scenario tails (project + tasks + decision + board for each persona); after 2-3 cycles of clean dogfood signal, split into a separate release-tropo-work-walk.playbook if scope earns it. Tests: user-experience usage-pattern — exercises projects, tasks, decisions, notes, boards, pipelines, collections (the eight Tropo Work primitives per L1 §4) at depth.

Mike-V47 directive 2026-05-18: "earn the abstraction strict — start with scenario tail extension; the dogfood signal tells us whether to split into a separate playbook."

4. Feature-specific test cases (PROPOSED; Mike-V47 pinned 2026-05-18)

Status: proposed; substrate folder + capsule shape not yet authored. Path: .tropo/test-cases/v<X.Y.Z>-<feature-slug>.md (per Mike's pin); ships with every release; persistent per-feature. Tests: specific features still work mechanically — every release adds test cases for the cycle's substantive feature work; any Studio can run any test case at any time.

Mike-V47 verbatim 2026-05-18: "It would be cool if when we did a release, there were some test cases included in the .tropo/ structure that could test specific features."

The fourth primitive composes with #1-3 — different shape (mechanical verification vs persona walks); different cadence (per-feature persistent vs per-cycle); together they form the complete envelope.

§Why This Is First-Class (Mike-V47 Calibration 2026-05-18)

V47's initial design-walk pulled back from subsystem-class elevation per a strict reading of "earn the abstraction": "Release test harness is a TOOLING surface that serves Tropo Governance, not a peer subsystem." Mike overrode that read.

The calibration: when substrate is load-bearing AND compounds across every release AND has multi-component composability, elevation to subsystem-class is the correct posture — not over-investment. The test harness clears all three:

  • Load-bearing: the harness IS the release ship-gate. Without it, releases ship on substrate-verification alone (vela-test-plan) — which v1.41 + v1.42 dogfoods empirically proved is insufficient (cold-boot walk caught two P0 Stream-E regressions vela-test-plan missed).
  • Compounds across releases: cycle-comparable signal accumulates. The harness gets MORE valuable across releases, not less. Each cycle's baseline informs the next.
  • Multi-component: four primitives at minimum (vela-test-plan + cold-boot-walk + tropo-work-walk + test-cases); each at its canonical type-location; needs a coherent discovery surface.

That clears the substrate-subsystem-class bar. Tropo Work earned it; Tropo Governance earned it; the Tropo Test Harness earns it.

§How Components Compose (Architecture)

                    Tropo Test Harness (this hub — 952f3aa3)
                              │
                              │ indexes
                              ▼
       ┌──────────────────────┼──────────────────────┐
       │                      │                      │
   FORM-CHECK            UX FIRST-TOUCH          UX DEPTH
       │                      │                      │
       ▼                      ▼                      ▼
 vela-test-plan       release-cold-boot-walk   release-tropo-work-walk
 (cycle-specific      (cycle-comparable;       (cycle-comparable;
  per release)         3 personas;              extends personas
                       9 binary gates;          with scenario tails
                       1-10 score;              first; may split to
                       gap inventory)           own playbook later)

                              │
                       MECHANICAL REGRESSION
                              │
                              ▼
                  Feature-specific test cases
                  (persistent per-feature;
                   ship with every release;
                   any Studio runs anytime)

Each primitive answers a question the others can't. Substrate-test verifies FORM. Cold-boot-walk verifies STRANGER-ENCOUNTER UX. Tropo-work-walk verifies USAGE-PATTERN UX. Test-cases verify FEATURE-LEVEL REGRESSION. Together: complete release-quality envelope.

§Where Each Component Lives (Canonical Locations)

Component Canonical location Discovery via
Hub (this entry) vault/files/952f3aa3.md Tropo Subsystems graph walk; member_of aae9a37b
release-cold-boot-walk.playbook .tropo/playbooks/release-cold-boot-walk.playbook.md (UID 6f3d2a18) .tropo/playbooks/ directory + playbook catalog
Cold-boot persona substrate .tropo/personas/cold-boot/ (CAPSULE a7c9d4e2; engineer + operator + enterprise) .tropo/personas/cold-boot/CAPSULE.md
Per-cycle vela-test-plans vault/files/<uid>.md with note_subtype: vela-test-plan grep note_subtype: vela-test-plan in vault index
Per-cycle run folders argo-os/playbook-runs/release-cold-boot-walk-v<X.Y.Z>-<date>/ folder naming convention; gitignored
Per-cycle aggregate reports /Users/mike/dev/tropo-releases/v<X.Y.Z>/cold-boot-walk-report.md release artifact convention; outside platform repo
Cycle-finding briefs vault/files/<uid>.md with type: design-brief routed to dev-pipeline 01-inbox grep member_of: 0a1a36fe
(Future) Tropo-work-walk substrate Persona scenario tails first; later potentially .tropo/playbooks/release-tropo-work-walk.playbook.md TBD when shape proven
(Future) Feature-specific test cases .tropo/test-cases/v<X.Y.Z>-<feature>.md TBD when authored

No substrate moves to accommodate this hub. Each component stays at its type-canonical location. The hub indexes; the type-discipline governs each piece.

§Rubric + Scoring Discipline (cold-boot-walk + future tropo-work-walk)

Per release-cold-boot-walk.playbook v1.1:

  • 3 binary gates per persona — Greeted clean / First agent created / Create-then-use closed
  • 1 holistic 1-10 score with one-paragraph persona-lens rationale
  • Gap inventory per the sa.cold-boot naive-session discipline (every off-target read, every ambiguity, every friction point logged)
  • Aggregate verdict: SHIP-CLEAR (all gates pass + avg ≥ 8.0) / MIXED (gates pass + avg 6.0-7.9) / SHIP-BLOCKED (any gate fails OR avg < 6.0 OR cycle-blocking findings)
  • Cycle-comparable signal: same harness + same personas + different substrate cycle = score-delta + gap-class-delta inform next-cycle scope priorities

Personas: three production (engineer / operator / enterprise); Studios may amend or add. Cross-sleeve coverage (Opus + Sonnet + Haiku) is a future variant (target_sleeve: field in persona substrate).

§Provenance + Cycle History

  • v1.40 cycle (2026-05-17): harness concept first walked (Mike-V47 brainstorm); release-cold-boot-walk.playbook v0.1 authored
  • v1.41 cycle (2026-05-18): harness v1.0 LOCKED; first-cycle dogfood — verdict MIXED; aggregate avg 7.33/10; two P0 ship-defects surfaced (channels/ + system/ folders missing — earning the harness's keep on first cycle); brief 5f9d9e3e filed for v1.43 absorption
  • v1.42 cycle (2026-05-18): second-cycle dogfood — verdict MIXED; aggregate avg 7.33/10 (zero net movement; cycle thesis empirically validated — v1.42 was infrastructure-class); path migration to /Users/mike/dev/tropo-releases/ outside platform repo (Mike-V47 directive); harness v1.0 → v1.1 amendment landed (per-persona Studio clones at dispatch + Step 1 collapse + verify-aggregate discipline added per V47 authoring-discipline failure A72 caught)
  • 2026-05-18 (this hub): subsystem-class elevation per Mike-V47 directive; first-class Tropo subsystem; member_of tropo-subsystems

§Composition with Other Subsystems

  • Tropo Governance (8dd772a0) — governs the harness via doctrine pins (7b69bd19 Quality-over-speed; 839a65f9 Deletion Discipline; 0aefe71d canonical doctrine entry). The harness IS a governance instrument; Governance gives it teeth.
  • Tropo Playbooks (76bab75f) — release-cold-boot-walk.playbook member_of; the playbook substrate's type-discipline governs the orchestration layer.
  • Tropo Work (2d083137) — the subsystem the future tropo-work-walk exercises; harness verifies that Tropo Work primitives work in stranger-encountered Studios.
  • Tropo Agents (99ed55fd) — the personas + the sa.cold-boot session-agent contract live here; harness exercises the agent-creation surface.

§What's Next (Roadmap)

v1.43 cycle (Argus A72 lane + Vela harness-amendment):

  • Argus absorbs brief 5f9d9e3e — P0 channels/ + system/ ship-skeletons; P1 substrate-coherence drift cleanup
  • Vela harness v1.1 amendment already landed (per-persona clones + Step 1 collapse + verify-aggregate)
  • v1.43 release-cold-boot-walk against the fixed substrate should produce first positive cycle-comparable delta (target avg ≥ 8.0)

v1.44+ cycles (proposed):

  • Extend personas with Tropo Work scenario tails (project + tasks + decision + board)
  • After 2-3 cycles of clean signal, split tropo-work-walk into its own playbook if scope earns it
  • Feature-specific test cases substrate (Mike's pinned idea) — author .tropo/test-cases/ folder + capsule shape; per-feature mechanical verification

v2.0.0 public ship horizon:

  • Cross-sleeve persona coverage (Opus + Sonnet + Haiku) for stranger-encounter signal robustness
  • Test-cases discipline locked + Studios can amend per their domain
  • The harness as it ships v2.0.0 is the substrate every external Studio inherits

Tropo Test Harness | UID 952f3aa3 | Authored 2026-05-18 by Vela V47 per Mike-V47 subsystem-class elevation | First-class Tropo subsystem "Four primitives. Four release-step gates. Cycle-comparable signal compounds across every release. The harness IS the ship-gate; the substrate IS the substrate; together they earn v2.0.0 public ship."