Skip to content

docs: correct two stale claims found by a drift check - #94

Merged
imran-siddique merged 1 commit into
mainfrom
docs/status-drift-sweep
Aug 10, 2026
Merged

docs: correct two stale claims found by a drift check#94
imran-siddique merged 1 commit into
mainfrom
docs/status-drift-sweep

Conversation

@imran-siddique

Copy link
Copy Markdown
Member

Prompted by the cMCP reload item, where a drifted status file described an older state than the code and hid a live production defect. These two are documentation only, but the same check found them.

1. ROADMAP names a module that does not exist

Two places referenced the A2A wire binding as ca2a_runtime.transport.a2a:

$ ls src/ca2a_runtime/transport/
__init__.py  a2a_adapter.py  client.py  constants.py  server.py  wire.py

It is transport.a2a_adapter. A reader following the reference finds nothing.

2. profile.md P-6 reads as a claim about the protocol

It said, unqualified:

The appraisal is one-directional: the caller appraised the callee, not the reverse

That was accurate when written. Since #90 the protocol is mutual — a callee issues a challenge and appraises the caller's offer before opening the sealed payload. What is still one-directional is the recorded hardware runs, which is a different statement.

The note now separates the two and repeats that mutual is not simultaneous (the caller still commits a sealed payload before the callee has appraised it), matching the wording already in LIMITATIONS.md.

What the sweep did not change

Checked and found accurate, so left alone:

  • LIMITATIONS.md — already corrected in feat(ca2a): appraise the caller before opening its payload #90 with the protocol/hardware split.
  • cMCP STATUS.md — the rows are specific and hedged; the one bad row was the reload item, fixed in #492.
  • cMCP session-policy docs on the sensitivity vocabulary — updated correctly by #489 when it landed, no drift.

Separately, ca2a#47's checkboxes for the live transport are satisfied by work that has since merged. I have not ticked them here, because the remaining items on that tracker are hardware claims and I would rather not tick anything on an attestation tracker without the run to point at. Flagging it rather than editing it.

🤖 Generated with Claude Code

ROADMAP named the A2A wire binding module `ca2a_runtime.transport.a2a`
in two places. There is no such module; it is `transport.a2a_adapter`.
A reader following the reference finds nothing.

profile.md P-6 said "the appraisal is one-directional: the caller
appraised the callee, not the reverse" with no qualifier, which now reads
as a statement about the protocol. The protocol is mutual as of
2026-08-09; the recorded hardware runs are the thing that is still
one-directional. The note now separates those and repeats that mutual is
not simultaneous, matching the wording already in LIMITATIONS.md.

Prompted by the reload item in cmcp, where a drifted status file described
an older state than the code and hid a live defect. These two are only
documentation, but the same check found them.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@imran-siddique
imran-siddique merged commit 029bc6e into main Aug 10, 2026
12 checks passed
@imran-siddique
imran-siddique deleted the docs/status-drift-sweep branch August 10, 2026 18:47
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