You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Nov 4, 2026. It is now read-only.
Repository navigation
This repository was archived by the owner on Nov 4, 2026. It is now read-only.
finding(process): the session-URL PR footer does NOT survive an edit — a PATCH normalises it down to the bare form, the opposite of what the rule promises #6483
Measured while updating PR #6472 (objectui#6051). Filed unassigned, observational.
The rule as written
The dev-agent operating text tells agents to sign PR bodies with the session
URL form, on the stated ground that it survives edits while the bare form does
not:
_Generated by [Claude Code](https://claude.ai/code)_ <- "stripped on every edit"
_Generated by [Claude Code](https://claude.ai/code/session_<id>)_ <- "survives both paths"
What was measured
Three writes against objectstack-ai/objectui PR #6472, each read back through GET /repos/objectstack-ai/objectui/pulls/6472:
#
write
body ended with
read back as
1
POST /pulls (create)
session form
session form
2
PATCH /pulls/6472 (edit)
session form
bare form
3
PATCH /pulls/6472 (edit)
session form
bare form
Write 3 was run specifically to confirm 2 was not a one-off. It is deterministic.
Both halves of the rule are wrong for the EDIT path
The session form does not survive an edit — it is normalised DOWN to the
bare form. That is the opposite of what the rule promises, and it is the exact
spelling the rule tells agents to use because it survives.
The bare form is not stripped by a subsequent edit — write 3 read back the
bare form intact, so the surviving footer after an edit is the bare one.
Why it matters
Silent loss of attribution, in the direction nobody checks. An agent follows the
rule, sees the session URL after POST, and never looks again; the first edit —
a review fixup, a rebase note, a PM-requested change — quietly drops the session
id. The footer stays present, so nothing looks broken; only the link back to
the session that produced the PR is gone. On #6472 the edit that lost it was a PM
ruling being written into the body, i.e. exactly the kind of edit the rule exists
to survive.
The create path is unaffected. Comments are a separate channel and were not
re-measured here (the report comment on objectui#6051 kept its footer, but a
platform signature was also appended to it, which is its own behaviour).
What a fix would look like
Correct the operating text to state what the edit path actually does, and say
what an agent should do about it — e.g. treat the session link as create-only and
put the session URL somewhere an edit does not normalise (a line of body prose, or
a comment) when the attribution needs to survive edits. No code change in this
repo.
Related: objectui#6051 / PR #6472 (where it was measured).
Activity
docs: cite landing commits where pending changesets and source commen…
docs(citations): re-point the 30 dead objectstack#N citations and the…