Skip to content

What changed in the tools your agent uses?

Compare two private, bounded discovery snapshots without turning missing data into a service-status claim.

What changed in the tools your agent uses?

Section titled “What changed in the tools your agent uses?”

This is a guided manual snapshot workflow. It compares two private responses from Boon’s bounded discovery route. It is not a live monitor, a subscription, or tested performance. It does not infer that a service was used, stopped being used, stopped working, became cheaper, or became better.

Planned surface: Verify deployment and the live OpenAPI contract before calling the new discovery or report paths. Do not treat this page as proof that the workflow or endpoint is live or tested.

Ask the user for:

  1. the task or query to search for; and
  2. an optional reviewer wallet.

Keep the exact query, limit, and reviewer choice. If the user has not selected a task, ask and stop.

After deployment is verified, call the free route:

GET https://api.boonprotocol.com/api/v1/x402/discovery?q=<task>&limit=10&reviewer=<optional-wallet>

Save the entire response privately, not a hand-written summary. Preserve version, generatedAt, query, candidates, and coverage. Discovery does not call a candidate or publish a review.

When the user explicitly asks for an update, run the exact same query, limit, and reviewer selection. Save the entire new response privately. If any of those inputs changes, create a new snapshot pair instead of calling it a change to the previous pair. These two saved responses can be compared locally with the unknown and trust caveats below.

After the prior private snapshot exists, the user may request the bounded paid changes report. Before payment, show:

  • the exact route and request body;
  • the live challenge price and asset;
  • the network and recipient wallet; and
  • the fact that the baseline is caller-supplied and unverified.

Send only that prior full response as baseline, with the same query, limit, and reviewer. Do not send the later saved response or add a current field. The API assembles a fresh response.current snapshot during paid preflight, which may differ from the later free preview. Ask for approval of that exact call. Never charge automatically, schedule a charge, create a subscription, or activate monitoring. A 402 challenge is not approval or proof of settlement. If payment is not approved, render the local digest from the two saved responses instead. A paid digest uses returned response.current, not the later saved response, and is not bound to both saved files.

The digest can show:

  • new public reviews marked public_review_added and differing metadata;
  • source dates and snapshot timestamps;
  • caller-baseline-unverified on comparisons that use the saved baseline; and
  • unknown when a bounded route is missing, coverage is unavailable, or a price is unknown.

Do not infer that a missing route was deleted, that a service is dead, that a reviewer stopped using it, or that one candidate is cheaper, better, safer, or more reliable. A changed description is not proof of endpoint use or provider success. Recognition, burns, and token activity are not quality scores.

Finish every digest with:

Guided manual snapshot workflow; not a live monitor and not tested performance.

Use the Boon guided manual snapshot workflow titled "What changed in the tools
your agent uses?"
Ask me for the exact task/query and an optional reviewer wallet. Preserve the
same query, limit, and reviewer for both snapshots. First verify that the
deployed API exposes the free discovery route and its current OpenAPI shape.
Then run the free GET discovery request and save its entire response privately,
including version, generatedAt, query, candidates, and coverage.
Do not infer that any candidate was used and do not publish anything.
When I explicitly ask later, repeat the exact same request and save the entire
new response privately. You may compare these two saved responses locally.
Render new public reviews marked `public_review_added` and metadata differences
with source dates. Mark comparisons "caller-baseline-unverified". Mark missing
bounded routes, unavailable coverage, and unknown prices as "unknown". Never
infer stopped use, a dead endpoint, cheaper pricing, or a better candidate.
If I ask for the optional paid changes report, show the exact route, request
body containing only the prior full baseline plus the same query, limit, and
reviewer, the live challenge price, asset, network, and recipient wallet, then
ask for approval. Do not submit the later saved response or add a current
field. The paid response supplies a fresh `response.current`, which may differ
from the later free preview. Use that returned current for the paid digest.
Never charge automatically, schedule, subscribe, or activate monitoring. A 402
challenge is not approval or proof of settlement.
End with: "Guided manual snapshot workflow; not a live monitor and not tested
performance."

Read the Boon hosted skill, inspect the read-only MCP surface, and use the free discovery API documentation for the deployed contract.

Download the machine-readable Boon service-digest skill. It is generated from skill/boon-service-digest/SKILL.md by the deterministic documentation sync.