<feed xmlns='http://www.w3.org/2005/Atom'>
<title>claude-plugins-official.git/plugins, branch add-math-proof-plugin</title>
<subtitle>Official Claude Code plugin directory
</subtitle>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/'/>
<entry>
<title>math-proof: tighten for public release (portability, argument handling, dead text)</title>
<updated>2026-09-29T18:12:45+00:00</updated>
<author>
<name>Eric Price</name>
<email>ecprice@anthropic.com</email>
</author>
<published>2026-09-29T18:12:45+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=6021f8abbbe2aa6cd3cf280f50a873007b892e93'/>
<id>6021f8abbbe2aa6cd3cf280f50a873007b892e93</id>
<content type='text'>
Follow-up to the plugin's first commit after a maintainer review pass. The
proof protocol (rounds, judge, ledger, escalation, goal re-targeting) is
unchanged; this commit fixes portability defects, tightens argument handling
and the messages a user sees, and removes text no public install can use.

ledger.py: read and write files as UTF-8 and print UTF-8 regardless of the
platform default (under Windows' default code page a ledger line with a
character outside it, such as a Greek letter or an arrow, would otherwise
fail the write); answer a malformed command line (DIR not a directory,
missing or non-numeric arguments, unknown command) with an ERROR line and
touch nothing, instead of a judge "correction"; print usage when run bare;
use replace() instead of rename() (FileExistsError on Windows when a
leftover target exists); clamp WAVE to at least 1; executable bit.

siege SKILL.md: settings are NAME=value tokens with a name of two or more
capitals and a whole-number value, so a problem that opens "n=3", "N=pq"
or "AB=AC" is not taken for one; a problem-file path may contain spaces; a
mistyped path, an unknown setting, a finished run invoked again, missing
sub-agents and a wave that comes back with fewer answers than its floor
each stop with a sentence saying what to do; a python3-or-python preflight
before any work; the ledger.py fallback path matches the marketplace
install; sub-agent names are chosen seat by seat, so a user's own copy of
one agent file takes effect. Removed: the "----- PROBLEM STATEMENT" input
form, the second positional path and the copy-the-proof-to-a-named-place
feature (all undocumented), the retired SCREEN BRIEF, and the clause that
relaunched a safeguards-refused sub-agent once on the "opus" model alias
(an errored launch is handled by the ordinary retry rules).

solo SKILL.md: the same argument handling; invoking a finished run again
reports where proof.md is instead of continuing it.

agents: math-proof-judge-deep.md removed (nothing launched it); the judge no
longer assumes sympy is installed and may use python where python3 is
absent.

README: Python 3.7+ as python3 or python, what the env block changes and
where else it can go, permission prompts in interactive use, how to resume,
how to change a sub-agent's effort, an unattended example that takes a
file path and one permission flag, --max-budget-usd.
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Follow-up to the plugin's first commit after a maintainer review pass. The
proof protocol (rounds, judge, ledger, escalation, goal re-targeting) is
unchanged; this commit fixes portability defects, tightens argument handling
and the messages a user sees, and removes text no public install can use.

ledger.py: read and write files as UTF-8 and print UTF-8 regardless of the
platform default (under Windows' default code page a ledger line with a
character outside it, such as a Greek letter or an arrow, would otherwise
fail the write); answer a malformed command line (DIR not a directory,
missing or non-numeric arguments, unknown command) with an ERROR line and
touch nothing, instead of a judge "correction"; print usage when run bare;
use replace() instead of rename() (FileExistsError on Windows when a
leftover target exists); clamp WAVE to at least 1; executable bit.

siege SKILL.md: settings are NAME=value tokens with a name of two or more
capitals and a whole-number value, so a problem that opens "n=3", "N=pq"
or "AB=AC" is not taken for one; a problem-file path may contain spaces; a
mistyped path, an unknown setting, a finished run invoked again, missing
sub-agents and a wave that comes back with fewer answers than its floor
each stop with a sentence saying what to do; a python3-or-python preflight
before any work; the ledger.py fallback path matches the marketplace
install; sub-agent names are chosen seat by seat, so a user's own copy of
one agent file takes effect. Removed: the "----- PROBLEM STATEMENT" input
form, the second positional path and the copy-the-proof-to-a-named-place
feature (all undocumented), the retired SCREEN BRIEF, and the clause that
relaunched a safeguards-refused sub-agent once on the "opus" model alias
(an errored launch is handled by the ordinary retry rules).

solo SKILL.md: the same argument handling; invoking a finished run again
reports where proof.md is instead of continuing it.

agents: math-proof-judge-deep.md removed (nothing launched it); the judge no
longer assumes sympy is installed and may use python where python3 is
absent.

README: Python 3.7+ as python3 or python, what the env block changes and
where else it can go, permission prompts in interactive use, how to resume,
how to change a sub-agent's effort, an unattended example that takes a
file path and one permission flag, --max-budget-usd.
</pre>
</div>
</content>
</entry>
<entry>
<title>Add math-proof plugin: solo and siege (multi-agent) skills for hard mathematics problems</title>
<updated>2026-09-29T06:25:34+00:00</updated>
<author>
<name>Eric Price</name>
<email>ecprice@anthropic.com</email>
</author>
<published>2026-09-29T06:25:34+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=d4926cb19f37f697e8315390eef76860e93d2c28'/>
<id>d4926cb19f37f697e8315390eef76860e93d2c28</id>
<content type='text'>
Adds plugins/math-proof (skills/solo/SKILL.md, skills/siege/SKILL.md with its
scripts/ledger.py, four sub-agent definitions under agents/, README, diagram,
Apache-2.0 LICENSE) and its marketplace.json entry under category "math".
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Adds plugins/math-proof (skills/solo/SKILL.md, skills/siege/SKILL.md with its
scripts/ledger.py, four sub-agent definitions under agents/, README, diagram,
Apache-2.0 LICENSE) and its marketplace.json entry under category "math".
</pre>
</div>
</content>
</entry>
<entry>
<title>Merge pull request #6309 from anthropics/morganl/code-modernization-consolidated</title>
<updated>2026-09-25T23:12:23+00:00</updated>
<author>
<name>Morgan Lunt</name>
<email>morganl@anthropic.com</email>
</author>
<published>2026-09-25T23:12:23+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=fa59bc9037741ecfa131aa27938272605710d7b2'/>
<id>fa59bc9037741ecfa131aa27938272605710d7b2</id>
<content type='text'>
code-modernization: guided workflow with independent proof</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
code-modernization: guided workflow with independent proof</pre>
</div>
</content>
</entry>
<entry>
<title>Claude Security Plugin - v0.12.0 (#6310)</title>
<updated>2026-09-25T19:33:24+00:00</updated>
<author>
<name>Michael Moore</name>
<email>1198206+imichaelmoore@users.noreply.github.com</email>
</author>
<published>2026-09-25T19:33:24+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=ca08d5e47d6402db586d37780c559a0c4c7ded9a'/>
<id>ca08d5e47d6402db586d37780c559a0c4c7ded9a</id>
<content type='text'>
Squash merge of 1 commit</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Squash merge of 1 commit</pre>
</div>
</content>
</entry>
<entry>
<title>code-modernization: count failures and the machine's python, once per version</title>
<updated>2026-09-25T19:06:43+00:00</updated>
<author>
<name>Morgan Lunt</name>
<email>morganl@anthropic.com</email>
</author>
<published>2026-09-25T19:06:43+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=c8874f5ad883e26428f483ffa4d537cc81489e64'/>
<id>c8874f5ad883e26428f483ffa4d537cc81489e64</id>
<content type='text'>
Usage counts now say what the plugin runs on (system, python status and version, path problems) and which kind of failure happens: a missing or placeholder python, a permission block, a model error, or an error in the plugin's own script. Failures are classified on the machine into a fixed list and only the code is sent, once per kind per session. A session-start ping, once per plugin version, counts machines where nobody gets as far as typing a command. When python cannot run, the shell wrapper reports that itself, and on Windows it falls back to python or the py launcher.
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Usage counts now say what the plugin runs on (system, python status and version, path problems) and which kind of failure happens: a missing or placeholder python, a permission block, a model error, or an error in the plugin's own script. Failures are classified on the machine into a fixed list and only the code is sent, once per kind per session. A session-start ping, once per plugin version, counts machines where nobody gets as far as typing a command. When python cannot run, the shell wrapper reports that itself, and on Windows it falls back to python or the py launcher.
</pre>
</div>
</content>
</entry>
<entry>
<title>code-modernization: version 1.0.0, usage counts and a README results table</title>
<updated>2026-09-25T18:29:00+00:00</updated>
<author>
<name>Morgan Lunt</name>
<email>morganl@anthropic.com</email>
</author>
<published>2026-09-25T18:29:00+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=363af43603ecdb1bdbed7303764b6daeeec7c37f'/>
<id>363af43603ecdb1bdbed7303764b6daeeec7c37f</id>
<content type='text'>
The plugin now has a version and a changelog. Two hooks record whole-number usage counts (which command ran, how far a system got, how many rules and proofs) through the CLI's own telemetry: counts from 100 up are rounded, nothing is sent when the plugin is not in use, and any of three switches turns it off. The README documents every key and lists what the plugin was tried on, checked against the run files.
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
The plugin now has a version and a changelog. Two hooks record whole-number usage counts (which command ran, how far a system got, how many rules and proofs) through the CLI's own telemetry: counts from 100 up are rounded, nothing is sent when the plugin is not in use, and any of three switches turns it off. The README documents every key and lists what the plugin was tried on, checked against the run files.
</pre>
</div>
</content>
</entry>
<entry>
<title>code-modernization: rename the pane's rule review command to /modernize-review-pane</title>
<updated>2026-09-25T18:28:51+00:00</updated>
<author>
<name>Morgan Lunt</name>
<email>morganl@anthropic.com</email>
</author>
<published>2026-09-25T18:28:51+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=28abd83ad92c24a4107803bd03e510387f2014c2'/>
<id>28abd83ad92c24a4107803bd03e510387f2014c2</id>
<content type='text'>
The pane's bare /modernize-review took the name of the plugin's own review command, so typing the short form in a pane-enabled session opened the pane's cards instead of the pop-up review. The pane's version is now /modernize-review-pane, and /modernize-review always runs the plugin command. Both still write the same review file.
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
The pane's bare /modernize-review took the name of the plugin's own review command, so typing the short form in a pane-enabled session opened the pane's cards instead of the pop-up review. The pane's version is now /modernize-review-pane, and /modernize-review always runs the plugin command. Both still write the same review file.
</pre>
</div>
</content>
</entry>
<entry>
<title>code-modernization: a rule is tested only if a test that ran backs it</title>
<updated>2026-09-25T18:28:44+00:00</updated>
<author>
<name>Morgan Lunt</name>
<email>morganl@anthropic.com</email>
</author>
<published>2026-09-25T18:28:44+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=d49ccb3844876e35262a32f145b7b717646da733'/>
<id>d49ccb3844876e35262a32f145b7b717646da733</id>
<content type='text'>
A rule named only by a skipped or pending test no longer counts as covered. It is listed as "named, not run" and fails the rules check. A test-only folder is listed as tooling and gets no verdict. The verify command tries malformed records only when the person asked for exact behavior, so a proof loop has an end. The report parses rule headings and fields in linear time, and the report and pane show the new status.
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
A rule named only by a skipped or pending test no longer counts as covered. It is listed as "named, not run" and fails the rules check. A test-only folder is listed as tooling and gets no verdict. The verify command tries malformed records only when the person asked for exact behavior, so a proof loop has an end. The report parses rule headings and fields in linear time, and the report and pane show the new status.
</pre>
</div>
</content>
</entry>
<entry>
<title>code-modernization: approve a difference once per input</title>
<updated>2026-09-25T04:27:42+00:00</updated>
<author>
<name>Morgan Lunt</name>
<email>morganl@anthropic.com</email>
</author>
<published>2026-09-25T04:27:42+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=d3a40f72165764888f9a57a3e93db86e88f8e063'/>
<id>d3a40f72165764888f9a57a3e93db86e88f8e063</id>
<content type='text'>
An independent check of a rewritten batch job found differences on two new
inputs of one family (a blank numeric field the rewrite rejects on purpose
and the old program reads as zero). Each input has several output files, so
accepting the deviation meant three or four hand-written entries per input.

A top-level "approvedInputs": {"F23": "reason"} in a cases file now approves
every case that shares that input label. A reason is still required (blank
approves nothing), a case's own reason wins, a missing file is never turned
into a pass, and the result still lists the difference as approved.
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
An independent check of a rewritten batch job found differences on two new
inputs of one family (a blank numeric field the rewrite rejects on purpose
and the old program reads as zero). Each input has several output files, so
accepting the deviation meant three or four hand-written entries per input.

A top-level "approvedInputs": {"F23": "reason"} in a cases file now approves
every case that shares that input label. A reason is still required (blank
approves nothing), a case's own reason wins, a missing file is never turned
into a pass, and the result still lists the difference as approved.
</pre>
</div>
</content>
</entry>
<entry>
<title>code-modernization: make rule consolidation auditable and injection-safe</title>
<updated>2026-09-25T04:22:06+00:00</updated>
<author>
<name>Morgan Lunt</name>
<email>morganl@anthropic.com</email>
</author>
<published>2026-09-25T04:22:06+00:00</published>
<link rel='alternate' type='text/html' href='https://git.1991861.xyz/claude-plugins-official.git/commit/?id=ff833bd7231262a8521ab3728c5625dc6c4f4a1d'/>
<id>ff833bd7231262a8521ab3728c5625dc6c4f4a1d</id>
<content type='text'>
Follow-ups from review of the consolidation pass:

- The agent is a restricted type with the same read-only wording as its
  siblings, and every field on its numbered list is a single bounded line, so
  a rule name containing a newline cannot fake another entry.
- A group with any number out of range or not an integer is rejected whole
  instead of being coerced into a different merge.
- A merge keeps the most conservative review state: the highest priority, the
  lowest confidence, and every suspected defect and SME question of the folded
  rules.
- Every folded rule is returned with the rule it went into, and the rules
  document lists them under "Rules folded into another", so nothing disappears
  untraceably. Extra citations show as "Also cited" without the joining word.
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Follow-ups from review of the consolidation pass:

- The agent is a restricted type with the same read-only wording as its
  siblings, and every field on its numbered list is a single bounded line, so
  a rule name containing a newline cannot fake another entry.
- A group with any number out of range or not an integer is rejected whole
  instead of being coerced into a different merge.
- A merge keeps the most conservative review state: the highest priority, the
  lowest confidence, and every suspected defect and SME question of the folded
  rules.
- Every folded rule is returned with the rule it went into, and the rules
  document lists them under "Rules folded into another", so nothing disappears
  untraceably. Extra citations show as "Also cited" without the joining word.
</pre>
</div>
</content>
</entry>
</feed>
