Commit 08e4949
fix(pm-dispatch): give the decision-facets marker a sanitizer-safe spelling (#14264)
The 四棱 block's machine-findable marker was prescribed in exactly one
spelling — a raw HTML comment — and that spelling is not readable through
the channel the seats verify and extract with. Measured 2026-09-01 on a
live specimen, both channels on the same comment in the same minute: the
comment's bytes ARE in GitHub's storage (the page's embedded JSON payload
returns the marker line intact), and the MCP read path returns that line
with its text removed and the newline kept. A seat performing the
prescribed read-back therefore sees absence and cannot tell it from a
strip — which is what happened, and produced a repair comment for a
marker that had never been lost.
The remedy is the one the sibling os-dev-report marker has carried all
along and that was never carried over: two equivalent spellings, one of
them ordinary visible text, plus the clause that absence of the HTML form
is never evidence of absence of the thing. The plain-text form is the
same literal token, so literal-text grep extraction is unchanged.
Line-neutral by construction: the prescription is rewritten in place and
the file stays at 54 of 54, headroom 0, measured with the gate's own
wrapLine rather than counted by hand. The now-obsolete 转义拼写写入
qualifier is dropped — the measurement shows escaping was never what
decided survival.
Claude-Session: https://claude.ai/code/session_01Whev4BkZ4BRcgiXYo4muWP
Co-authored-by: Claude <noreply@anthropic.com>1 parent c5a9a43 commit 08e4949
1 file changed
Lines changed: 7 additions & 7 deletions
Lines changed: 7 additions & 7 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
38 | 38 | | |
39 | 39 | | |
40 | 40 | | |
41 | | - | |
42 | | - | |
43 | | - | |
44 | | - | |
45 | | - | |
46 | | - | |
47 | | - | |
| 41 | + | |
| 42 | + | |
| 43 | + | |
| 44 | + | |
| 45 | + | |
| 46 | + | |
| 47 | + | |
48 | 48 | | |
49 | 49 | | |
50 | 50 | | |
| |||
0 commit comments