Skip to content

docs: README leads with measured OpenEMR times and the 30 of 30 claims result - #1338

Merged
abrichr merged 8 commits into
mainfrom
claude/agent-claims-readback
Oct 20, 2026
Merged

abrichr merged 8 commits into
mainfrom
claude/agent-claims-readback

Conversation

@abrichr

@abrichr abrichr commented Oct 19, 2026 •

Copy link
Copy Markdown
Member

The README now opens with what was measured.

OpenEMR referral, median wall time from a fresh browser through sign-in, the entry and the read-back. The figures come from SCOREBOARD.md#openemr-agent-screen-program-and-api-step, added in #1365.

Who does the entry Time Correct Model cost
Agent alone (Claude Sonnet 5.5 computer use) 36.5 s 62 of 63 about $0.08
OpenAdapt program, screen steps 22.5 s 18 of 18 $0
OpenAdapt program, API step 3.3 s 18 of 18 $0

The labels say exactly how each row ran:

  • The program rows ran on 19 Oct on one copy of the app, interleaved, without the approval gate and without resumable runs.
  • The API step was admitted by the benchmark harness, not by a person.
  • The agent rows are the 14–15 Oct study, on its own copy of the app.

The table uses wall time because it's the same clock for all three rows; task time isn't.

Claims versus read-back (MockMed, Amendment 3): the agent reported all 30 bad saves as done. Reading the records back caught all 30, with no false alarms on 10 clean saves. The wording notes that the test app shows a save as done before its server answers.

Also:

  • CAPABILITIES.md gets a measured-results row. It says the approval-gated production profile still runs the screen steps, and that work to change this is in progress.
  • llms-full.txt is regenerated.

Also: public copy lists only shipped work. README, CAPABILITIES.md, llms.txt and the agent README no longer say "in review", "being merged", "not yet released" or "being built", and the "What is not done yet" section is now "Limits", covering only what the shipped product doesn't do. Rows for merged PRs read "On main"; rows for still-open PRs (#1291, #1298) are removed until they merge. tests/test_shipped_only_copy.py fails if status words come back.

Merge after #1365, which adds the scoreboard section these link to.

🤖 Generated with Claude Code

abrichr and others added 3 commits October 19, 2026 09:53
@abrichr abrichr changed the title docs: an agent's claims versus the read-back (30 of 30 bad saves) docs: README leads with measured OpenEMR times and the 30 of 30 claims result Oct 19, 2026
abrichr and others added 5 commits October 19, 2026 14:02
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…edit gap

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Removes 'in review', 'being merged', 'not yet released', 'being built' and the
'What is not done yet' section from README, CAPABILITIES, llms.txt and the agent
README. Rows for merged PRs read 'On main'; rows for unmerged ones are removed.

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

abrichr commented Oct 19, 2026

Copy link
Copy Markdown
Member Author

Included in batch #1385, which runs CI once for several reviewed PRs. This PR is marked merged when #1385 merges. Please don't push here unless asked; a new commit leaves it out of the batch.

@abrichr
abrichr merged commit 99374a6 into main Oct 20, 2026
24 of 39 checks passed
abrichr added a commit that referenced this pull request Oct 20, 2026
@abrichr
abrichr deleted the claude/agent-claims-readback branch October 20, 2026 02:22
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.

1 participant