aeoess-aps: appraisal.status is a constant none; tested_against 0.9.0 - #139
Merged
imran-siddique merged 1 commit intoAug 27, 2026
Conversation
|
🔴 Contributor Check: HIGH
Automated check by AgenTrust Contributor Check. |
github-actions
Bot
added
the
needs-review:HIGH
Two changes, no others. 1. appraisal.status becomes the constant "none". The agentrust-trace-adapters convention that landed on integrations main in commit e1aa231 (2026-08-08) sets it that way for a record assembled from evidence another system produced: "Nobody appraised the evidence. Transcribing is not appraising", and "A vendor's bare ALLOW/DENY result is still a policy decision, not an appraisal of the evidence behind that decision." An APS verdict is exactly such a policy decision. Our mapping was reviewed as defensible on 2026-08-03. A clearer adapter convention landed on 2026-08-09 that separates policy decisions from evidence appraisal, and the exporter aligns to that convention here. VERDICT_TO_APPRAISAL is replaced by KNOWN_VERDICTS, a frozenset, so an unknown verdict is still refused rather than transcribed. The verdict remains in the record where it belongs, as policy.enforcement_mode and policy.version. The README field table and the tests are updated to match, and a new test pins the status as a constant so a future edit cannot quietly reintroduce a verdict-to-status mapping. 2. integration.yaml tested_against agentrust-trace 0.5.1 -> 0.9.0. Nothing here touches revocation. That work waits on agentrust-io#89 and the mapping issue. Verified against agentrust-trace 0.9.0 / agentrust-trace-tests 0.5.1 / agent-passport-system 2.11.0: 27 passed; emit_record prints appraisal.status none; trace-tests verify --level 0 gives PASS, 8 checks, 0 skipped, 1 UNVERIFIED, which is the documented expected result. "none" is in the 0.9.0 schema enum for appraisal.status. Refs: agentrust-io#86, agentrust-io#89 Signed-off-by: Tymofii Pidlisnyi <171286556+aeoess@users.noreply.github.com> (cherry picked from commit f4a525a)
aeoess
force-pushed
the
exporter/appraisal-none-tested-against-0.9.0
branch
from
August 26, 2026 18:20
f4a525a to
38a1b1b
Compare
imran-siddique
approved these changes
Aug 27, 2026
imran-siddique
left a comment
imran-siddique
left a comment
Member
There was a problem hiding this comment.
Correct, and in the right direction: this removes a claim rather than adding one.
The verdict-to-appraisal mapping was the defensible-but-wrong kind. permit becoming affirming reads as "the evidence was appraised and held up", when what actually happened is that a policy engine returned a decision. none is the honest value, and the verdict is not lost: it stays in policy.enforcement_mode and policy.version, which is where a consumer should look for it.
Checked against the bar:
- Runnable against a released package:
agentrust-trace 0.9.0, andtested_againstmoved with it. - The unknown-verdict refusal is preserved.
KNOWN_VERDICTSkeeps doing that job afterVERDICT_TO_APPRAISALstopped being the set it was checked against, which is the part that would have been easy to drop. - README, module docstring and the three verdict tests all moved together.
test_every_known_verdict_maps_to_a_valid_ear_statuswas adapted rather than deleted. - CI green. The
gatefailure is the maintainer-approval check, which this review satisfies.
Thanks for citing the convention commit and the date the earlier mapping was reviewed. That is what made this reviewable without reconstructing the history.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Aligns the exporter with the adapter convention that landed in
e1aa231(2026-08-08): a record assembled from evidence another system produced carriesappraisal.status: none, because transcribing a policy decision is not appraising the evidence behind it.Changes, all in
integrations/aeoess-aps/:aps_trace.py:VERDICT_TO_APPRAISALremoved;KNOWN_VERDICTSkeeps the refusal of unknown verdicts;appraisal.statusis the constantnone. The verdict is still carried throughpolicy.enforcement_modeandpolicy.version.README.md: field table and the appraisal note updated to match.integration.yaml:tested_against: agentrust-trace: "0.9.0".tests/test_mapping.py: asserts the constant and the unchanged refusal paths.pytest: 27 passed.trace-tests verify --level 0: PASS, 8 checks, 0 skipped, 1 UNVERIFIED (TR-SIG-005, the unsigned graded artifact, unchanged from the 2026-08-04 baseline).Nothing about revocation is in this change; that follows the mapping issue.