# Tara service discovery · delivery receipt · 8 September 2026

The [24-slide guide](../../decks/tara-service-discovery.html) and
[PDF](../../decks/tara-service-discovery.pdf) teach recommendation-backed search,
saved-item reconciliation and launch resolution with native conditional pools,
seven branches, an owner exchange, three record specimens and worked comparisons.
[Authoring](../../authoring/v1-tara-service-discovery.md) records the source,
sequence, visual decisions and unresolved application-consumer boundaries.
Baseline: `6830fb7b036c002a6453589593ac825645049538`; primary pin:
`1ceae52c7896a59ac74a4429d410f3218f0b91c6`.

The library has 68 guides, 746 unique slides, 769 guide-slide instances and
413 cited paths. All 722 prior slides, scripts, narration entries and MP3 bytes
remain exact. Twenty-four tracks add 1,897.670 seconds (31.6 minutes).
The source mechanisms taught here do not close a whole Tara unit or full V1.

[Environment](environment.json) records the installed browser, runtime and speech
package versions.

## Evidence

- [26 source cases](source-cases.json) execute twenty complete pinned modules,
  including dedicated BFF routes through Fastify with declared data, middleware
  and fixed-clock seams. Token verification and database operations are outside
  that fixture. [22-path continuity](source-continuity.json) also binds the
  inspected Tara service transport branch and local web-search cross-reference.
- [174 final browser checks](browser/report.json) retain 96 captures, 72 embedded
  axe runs, six enlarged-text cases and three rejected removals: the favorite
  pool, partial-view preservation and explicit query override. No fit findings
  or unexpected browser errors remain.
- [Focused center review](center.json) passes fourteen checks, keeps all 67 prior
  guide HTML files and four shared UI assets exact, retains 48 embedded captures,
  checks both audio boundaries and exports all 24 PDF pages. Browser audio output
  is disabled. [PDF visible text/order](pdf-visual-text.json) and
  [optical review](optical-review.json) are separately retained.
- The settled presentation suite passes 118 cases. Scoped narration decode/pause,
  freshness, coverage accounting, ESLint, Ruff, formatting and local-link checks
  pass. [Speech reuse](speech-reuse.json) binds the prior scripts, entries and
  bytes; it does not claim a fresh human listening session.

## Attempts and limits

The first draft passed 48 desktop/narrow fit cases. All 24 desktop slides were
inspected in contact sheets, with native-size and narrow samples recorded in the
optical review. Comparison headings describe their columns, peer emphasis is
null, and the BFF lane receives a semantic identity-ownership emphasis.

Initial source attempts found a missing direct Fastify symlink and cross-VM
array prototypes. The fixture selects the installed package with an explicit
environment override and serializes comparisons at its declared boundary;
pinned source buffers still compare exactly. ESLint found an unused assignment,
which was removed. Retained failed logs are not passing evidence.
[Log preservation](log-preservation.json) retains exact compressed originals for
two console logs whose readable copies trim trailing whitespace. No product
runtime implementation was changed by this presentation increment.

Run `python3 docs/presentations/presentation-center/verification/v1-tara-discovery-2026-09-08/verify-receipt.py`
from the repository root. Later library edits invalidate this frozen receipt;
use the delivery revision for historical checks. The
[full V1 program](../../authoring/v1-completion.md) remains active. Application
storage, retries, actual destination content, service/database readback and
playback require their own call traces. Human comprehension/listening, full
application accessibility and live member/clinical acceptance were not performed.
