super-agents 0.1.0__tar.gz

super_agents-0.1.0/.gitignore

@@ -0,0 +1,13 @@
+node_modules/
+dist/
+.DS_Store
+*.log
+__pycache__/
+*.py[cod]
+.pytest_cache/
+.ruff_cache/
+.venv/
+
+# BEGIN multi-managed: generated-files
+# These files are generated by multi.
+# END multi-managed: generated-files

super_agents-0.1.0/.reports/super-agent-instructions-read-diagnostic.md

@@ -0,0 +1,107 @@
+# Super Agent Instructions Read Diagnostic
+
+Date: 2026-05-22
+
+## Summary
+
+New Super Agents were not consistently reading `/Users/gabemontague/.openbase/codex_home/super-agent-instructions.md` because this MCP server did not read that file. The Python MCP implementation only forwarded `developerInstructions` when the MCP caller explicitly supplied that field.
+
+The likely cause of the observed split result is that one launch path injected the contents of `super-agent-instructions.md` as explicit developer instructions, while another path used the direct `super_agents_start` / `super_agents_start_turn` tools without doing so. Direct MCP tool calls currently inherit the normal Codex home/project instruction discovery, such as `/Users/gabemontague/.openbase/codex_home/AGENTS.md`, but not `super-agent-instructions.md`.
+
+Fix implemented: `src/super_agents/mcp_server.py` now loads default Super Agent developer instructions at MCP input-cleaning time and applies them to new threads, direct follow-up turns, and queued turns when callers omit `developerInstructions`.
+
+## Evidence
+
+- Before the fix, `src/super_agents/mcp_server.py` defined `super_agents_start`, `super_agents_start_turn`, and `super_agents_queue_turn` with an optional `developerInstructions` input field, but no default instruction-file loading.
+- `src/super_agents/app_server_client.py` sends `developerInstructions` to `thread/start` only if `input_data.get("developerInstructions")` is truthy.
+- `src/super_agents/app_server_client.py` always builds `turn/start` with `collaborationMode.settings.developer_instructions`, but the value is only taken from the tool input. If omitted, it is `None`.
+- `resume_thread()` calls `thread/resume` with `threadId`, `approvalPolicy`, `sandbox`, and shell config only. It does not read or reapply any Super Agent instruction file.
+- `queue_turn_by_label()` stores the cleaned turn input and `_drain_queue()` later calls `start_turn()` with that stored input. If the queued call omitted `developerInstructions`, queued execution also omits the Super Agent instructions.
+- Repository search found no references to `super-agent-instructions.md` or `CODEX_SUPER_AGENT_INSTRUCTIONS_PATH` in `src/`, `tests/`, `legacy/`, or `README.md`.
+- `default_reasoning_effort()` reads `SUPER_AGENTS_DEFAULT_CONFIG_PATH`, `LIVEKIT_DISPATCHER_CONFIG_PATH`, or `~/.openbase/codex_home/dispatcher-config.json`, but only for reasoning effort. It does not load instructions.
+- `login_shell_config_override()` passes shell environment policy to Codex app-server. It sets environment variables such as `PATH`, `SHELL`, `HOME`, `USER`, and `LOGNAME`; it does not set instruction content or instruction paths.
+
+Current local files also show why the random-animal test is a useful detector:
+
+- `/Users/gabemontague/.openbase/codex_home/super-agent-instructions.md` contains `The random animal is "raccoon".`
+- `/Users/gabemontague/.openbase/codex_home/AGENTS.md` contains `The random animal is "monkey".`
+
+A Super Agent answering `raccoon` saw the Super Agent instruction file through some path outside this MCP's default direct tool behavior. A Super Agent answering `unknown` probably did not receive either the Super Agent file or a clear relevant instruction in its active context. A direct `super_agents_start` thread should not be expected to know `raccoon` unless the caller passed that file content as `developerInstructions`.
+
+## Flow Coverage
+
+`super_agents_start`
+
+- Tool path: `mcp_server.py` -> `clean_thread_input()` -> `CodexAppServerClient.start_thread()`.
+- Sends `thread/start` with `cwd`, `approvalPolicy`, `sandbox`, and shell config.
+- Adds `developerInstructions` only when provided by the caller.
+- Does not read `CODEX_SUPER_AGENT_INSTRUCTIONS_PATH`.
+- Does not read `~/.openbase/codex_home/super-agent-instructions.md`.
+- Does not read dispatcher config for instructions.
+
+`super_agents_start_turn`
+
+- Tool path: `mcp_server.py` -> `clean_start_turn_by_name_input()` -> `CodexAppServerClient.start_turn_by_label()` -> `start_turn()`.
+- Resolves an existing named thread and sends `turn/start`.
+- Includes `collaborationMode.settings.developer_instructions` as the caller-provided value or `None`.
+- Does not load default Super Agent instructions for follow-up turns.
+
+`super_agents_queue_turn`
+
+- Tool path: `mcp_server.py` -> `clean_queue_turn_input()` -> `queue_turn_by_label()` -> `_drain_queue()` -> `start_turn()`.
+- Queued turns preserve only the provided turn input.
+- If `developerInstructions` was absent at queue time, the eventual turn starts with `developer_instructions: None`.
+
+`super_agents_resume`
+
+- Tool path: `resume_by_label()` -> `resume_thread()`.
+- Resumes an existing app-server thread.
+- Does not pass new developer instructions.
+- Existing named threads created before an instruction-file change can retain whatever instruction context was stored when the thread was created or previously run. The MCP state file does not persist developer instructions, so the MCP cannot update old named threads automatically from `super-agent-instructions.md`.
+
+## Named Thread Reuse
+
+`super_agents_start_turn` resolves the latest matching named thread, defaulting to `latest_any` inside `start_turn_by_label()`. That means a "fresh" test can accidentally reuse an existing named thread if the name is reused. In that case, the result may reflect old thread context rather than the current instruction file.
+
+Even for truly new names, direct `super_agents_start` only passes default Codex app-server thread configuration. The app-server may load normal Codex home/project instructions, but this repo does not tell it to load `super-agent-instructions.md`.
+
+## Recommended Fix
+
+Add a single default-instruction loader in this MCP server and apply it consistently before app-server requests are built. This fix has now been implemented.
+
+Recommended behavior:
+
+1. Resolve the instruction path from `CODEX_SUPER_AGENT_INSTRUCTIONS_PATH` if set.
+2. Otherwise use `$CODEX_HOME/super-agent-instructions.md` if `CODEX_HOME` is set.
+3. Otherwise use `~/.openbase/codex_home/super-agent-instructions.md`.
+4. Read the file at call time, not process startup time, so changes are picked up by new threads and new turns without restarting the MCP process.
+5. For `super_agents_start`, use the caller-provided `developerInstructions` if supplied; otherwise use the loaded default.
+6. For `super_agents_start_turn` and `super_agents_queue_turn`, also default missing `developerInstructions` to the loaded file. Preserve explicit `developerInstructions: null` as an escape hatch if callers need to suppress the default.
+7. For `super_agents_resume`, either keep resume as a pure resume operation and document that it does not update existing thread instructions, or add an explicit follow-up-turn path that reapplies defaults. Do not assume resume alone can mutate prior thread context.
+
+The implementation is in `mcp_server.py` input cleaning, before `CodexAppServerClient` receives the sanitized payload. That keeps instruction policy at the MCP boundary and avoids changing the lower-level app-server client semantics.
+
+## Fix Applied
+
+- Added `default_super_agent_instructions_path()` with this precedence:
+  1. `CODEX_SUPER_AGENT_INSTRUCTIONS_PATH`
+  2. `$CODEX_HOME/super-agent-instructions.md`
+  3. `~/.openbase/codex_home/super-agent-instructions.md`
+- Added `default_super_agent_instructions()` to read the file at call time, so new threads and turns can pick up file edits without a Python process restart.
+- Updated `clean_thread_input()` so `super_agents_start` defaults missing `developerInstructions` to the loaded file.
+- Updated `clean_turn_input()` so `super_agents_start_turn` and `super_agents_queue_turn` default missing `developerInstructions` to the loaded file.
+- Preserved explicit caller-provided developer instructions as the highest-priority value.
+- Preserved explicit `developerInstructions: null` as a turn-flow escape hatch to suppress the default.
+
+## Verification Performed
+
+- Searched the repo for `super-agent-instructions`, `CODEX_SUPER_AGENT_INSTRUCTIONS_PATH`, `developerInstructions`, `developer_instructions`, dispatcher config usage, and Super Agents tool entry points.
+- Read the Python MCP tool definitions and app-server client request builders.
+- Compared the current Codex home `super-agent-instructions.md`, `AGENTS.md`, and dispatcher config.
+- Added focused tests for configured instruction path loading, `$CODEX_HOME` fallback loading, explicit override behavior, and explicit-null suppression for turn flows.
+- Ran focused Python tests with cache disabled: `PYTHONDONTWRITEBYTECODE=1 pytest -q -p no:cacheprovider tests/test_app_server_client.py tests/test_mcp_server.py`.
+- Result: `27 passed`.
+
+## Reload Note
+
+The MCP process that exposes these tools must be reloaded before already-running clients see the code change. No service restart was performed as part of this task.

super_agents-0.1.0/.weekend-commit-rewriter.git/FETCH_HEAD

@@ -0,0 +1,4 @@
+c8e296a6c0e7c3ac5f3a1170c1dc4b09cdaa4363		branch 'main' of /Users/gabemontague/Projects/super-agents
+c8e296a6c0e7c3ac5f3a1170c1dc4b09cdaa4363		remote-tracking branch 'origin/HEAD' of /Users/gabemontague/Projects/super-agents
+c8e296a6c0e7c3ac5f3a1170c1dc4b09cdaa4363		remote-tracking branch 'origin/main' of /Users/gabemontague/Projects/super-agents
+2edeb4de0949988a4b1bd50156a1fbfb733aad14	not-for-merge	tag 'v0.1.0' of /Users/gabemontague/Projects/super-agents

super_agents-0.1.0/.weekend-commit-rewriter.git/HEAD

@@ -0,0 +1 @@
+ref: refs/heads/main

super_agents-0.1.0/.weekend-commit-rewriter.git/config

@@ -0,0 +1,12 @@
+[core]
+	repositoryformatversion = 0
+	filemode = true
+	bare = true
+	ignorecase = true
+	precomposeunicode = true
+[remote "source"]
+	url = /Users/gabemontague/Projects/super-agents
+	fetch = +refs/heads/*:refs/remotes/source/*
+[remote "origin"]
+	url = git@github.com:montaguegabe/super-agents.git
+	fetch = +refs/heads/*:refs/remotes/origin/*

super_agents-0.1.0/.weekend-commit-rewriter.git/description

@@ -0,0 +1 @@
+Unnamed repository; edit this file 'description' to name the repository.

super_agents-0.1.0/.weekend-commit-rewriter.git/hooks/applypatch-msg.sample

@@ -0,0 +1,15 @@
+#!/bin/sh
+#
+# An example hook script to check the commit log message taken by
+# applypatch from an e-mail message.
+#
+# The hook should exit with non-zero status after issuing an
+# appropriate message if it wants to stop the commit.  The hook is
+# allowed to edit the commit message file.
+#
+# To enable this hook, rename this file to "applypatch-msg".
+
+. git-sh-setup
+commitmsg="$(git rev-parse --git-path hooks/commit-msg)"
+test -x "$commitmsg" && exec "$commitmsg" ${1+"$@"}
+:

super_agents-0.1.0/.weekend-commit-rewriter.git/hooks/commit-msg.sample

@@ -0,0 +1,24 @@
+#!/bin/sh
+#
+# An example hook script to check the commit log message.
+# Called by "git commit" with one argument, the name of the file
+# that has the commit message.  The hook should exit with non-zero
+# status after issuing an appropriate message if it wants to stop the
+# commit.  The hook is allowed to edit the commit message file.
+#
+# To enable this hook, rename this file to "commit-msg".
+
+# Uncomment the below to add a Signed-off-by line to the message.
+# Doing this in a hook is a bad idea in general, but the prepare-commit-msg
+# hook is more suited to it.
+#
+# SOB=$(git var GIT_AUTHOR_IDENT | sed -n 's/^\(.*>\).*$/Signed-off-by: \1/p')
+# grep -qs "^$SOB" "$1" || echo "$SOB" >> "$1"
+
+# This example catches duplicate Signed-off-by lines.
+
+test "" = "$(grep '^Signed-off-by: ' "$1" |
+	 sort | uniq -c | sed -e '/^[ 	]*1[ 	]/d')" || {
+	echo >&2 Duplicate Signed-off-by lines.
+	exit 1
+}

super_agents-0.1.0/.weekend-commit-rewriter.git/hooks/fsmonitor-watchman.sample

@@ -0,0 +1,174 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+use IPC::Open2;
+
+# An example hook script to integrate Watchman
+# (https://facebook.github.io/watchman/) with git to speed up detecting
+# new and modified files.
+#
+# The hook is passed a version (currently 2) and last update token
+# formatted as a string and outputs to stdout a new update token and
+# all files that have been modified since the update token. Paths must
+# be relative to the root of the working tree and separated by a single NUL.
+#
+# To enable this hook, rename this file to "query-watchman" and set
+# 'git config core.fsmonitor .git/hooks/query-watchman'
+#
+my ($version, $last_update_token) = @ARGV;
+
+# Uncomment for debugging
+# print STDERR "$0 $version $last_update_token\n";
+
+# Check the hook interface version
+if ($version ne 2) {
+	die "Unsupported query-fsmonitor hook version '$version'.\n" .
+	    "Falling back to scanning...\n";
+}
+
+my $git_work_tree = get_working_dir();
+
+my $retry = 1;
+
+my $json_pkg;
+eval {
+	require JSON::XS;
+	$json_pkg = "JSON::XS";
+	1;
+} or do {
+	require JSON::PP;
+	$json_pkg = "JSON::PP";
+};
+
+launch_watchman();
+
+sub launch_watchman {
+	my $o = watchman_query();
+	if (is_work_tree_watched($o)) {
+		output_result($o->{clock}, @{$o->{files}});
+	}
+}
+
+sub output_result {
+	my ($clockid, @files) = @_;
+
+	# Uncomment for debugging watchman output
+	# open (my $fh, ">", ".git/watchman-output.out");
+	# binmode $fh, ":utf8";
+	# print $fh "$clockid\n@files\n";
+	# close $fh;
+
+	binmode STDOUT, ":utf8";
+	print $clockid;
+	print "\0";
+	local $, = "\0";
+	print @files;
+}
+
+sub watchman_clock {
+	my $response = qx/watchman clock "$git_work_tree"/;
+	die "Failed to get clock id on '$git_work_tree'.\n" .
+		"Falling back to scanning...\n" if $? != 0;
+
+	return $json_pkg->new->utf8->decode($response);
+}
+
+sub watchman_query {
+	my $pid = open2(\*CHLD_OUT, \*CHLD_IN, 'watchman -j --no-pretty')
+	or die "open2() failed: $!\n" .
+	"Falling back to scanning...\n";
+
+	# In the query expression below we're asking for names of files that
+	# changed since $last_update_token but not from the .git folder.
+	#
+	# To accomplish this, we're using the "since" generator to use the
+	# recency index to select candidate nodes and "fields" to limit the
+	# output to file names only. Then we're using the "expression" term to
+	# further constrain the results.
+	my $last_update_line = "";
+	if (substr($last_update_token, 0, 1) eq "c") {
+		$last_update_token = "\"$last_update_token\"";
+		$last_update_line = qq[\n"since": $last_update_token,];
+	}
+	my $query = <<"	END";
+		["query", "$git_work_tree", {$last_update_line
+			"fields": ["name"],
+			"expression": ["not", ["dirname", ".git"]]
+		}]
+	END
+
+	# Uncomment for debugging the watchman query
+	# open (my $fh, ">", ".git/watchman-query.json");
+	# print $fh $query;
+	# close $fh;
+
+	print CHLD_IN $query;
+	close CHLD_IN;
+	my $response = do {local $/; <CHLD_OUT>};
+
+	# Uncomment for debugging the watch response
+	# open ($fh, ">", ".git/watchman-response.json");
+	# print $fh $response;
+	# close $fh;
+
+	die "Watchman: command returned no output.\n" .
+	"Falling back to scanning...\n" if $response eq "";
+	die "Watchman: command returned invalid output: $response\n" .
+	"Falling back to scanning...\n" unless $response =~ /^\{/;
+
+	return $json_pkg->new->utf8->decode($response);
+}
+
+sub is_work_tree_watched {
+	my ($output) = @_;
+	my $error = $output->{error};
+	if ($retry > 0 and $error and $error =~ m/unable to resolve root .* directory (.*) is not watched/) {
+		$retry--;
+		my $response = qx/watchman watch "$git_work_tree"/;
+		die "Failed to make watchman watch '$git_work_tree'.\n" .
+		    "Falling back to scanning...\n" if $? != 0;
+		$output = $json_pkg->new->utf8->decode($response);
+		$error = $output->{error};
+		die "Watchman: $error.\n" .
+		"Falling back to scanning...\n" if $error;
+
+		# Uncomment for debugging watchman output
+		# open (my $fh, ">", ".git/watchman-output.out");
+		# close $fh;
+
+		# Watchman will always return all files on the first query so
+		# return the fast "everything is dirty" flag to git and do the
+		# Watchman query just to get it over with now so we won't pay
+		# the cost in git to look up each individual file.
+		my $o = watchman_clock();
+		$error = $output->{error};
+
+		die "Watchman: $error.\n" .
+		"Falling back to scanning...\n" if $error;
+
+		output_result($o->{clock}, ("/"));
+		$last_update_token = $o->{clock};
+
+		eval { launch_watchman() };
+		return 0;
+	}
+
+	die "Watchman: $error.\n" .
+	"Falling back to scanning...\n" if $error;
+
+	return 1;
+}
+
+sub get_working_dir {
+	my $working_dir;
+	if ($^O =~ 'msys' || $^O =~ 'cygwin') {
+		$working_dir = Win32::GetCwd();
+		$working_dir =~ tr/\\/\//;
+	} else {
+		require Cwd;
+		$working_dir = Cwd::cwd();
+	}
+
+	return $working_dir;
+}

super_agents-0.1.0/.weekend-commit-rewriter.git/hooks/post-update.sample

@@ -0,0 +1,8 @@
+#!/bin/sh
+#
+# An example hook script to prepare a packed repository for use over
+# dumb transports.
+#
+# To enable this hook, rename this file to "post-update".
+
+exec git update-server-info

super_agents-0.1.0/.weekend-commit-rewriter.git/hooks/pre-applypatch.sample

@@ -0,0 +1,14 @@
+#!/bin/sh
+#
+# An example hook script to verify what is about to be committed
+# by applypatch from an e-mail message.
+#
+# The hook should exit with non-zero status after issuing an
+# appropriate message if it wants to stop the commit.
+#
+# To enable this hook, rename this file to "pre-applypatch".
+
+. git-sh-setup
+precommit="$(git rev-parse --git-path hooks/pre-commit)"
+test -x "$precommit" && exec "$precommit" ${1+"$@"}
+:

super_agents-0.1.0/.weekend-commit-rewriter.git/hooks/pre-commit.sample

@@ -0,0 +1,49 @@
+#!/bin/sh
+#
+# An example hook script to verify what is about to be committed.
+# Called by "git commit" with no arguments.  The hook should
+# exit with non-zero status after issuing an appropriate message if
+# it wants to stop the commit.
+#
+# To enable this hook, rename this file to "pre-commit".
+
+if git rev-parse --verify HEAD >/dev/null 2>&1
+then
+	against=HEAD
+else
+	# Initial commit: diff against an empty tree object
+	against=$(git hash-object -t tree /dev/null)
+fi
+
+# If you want to allow non-ASCII filenames set this variable to true.
+allownonascii=$(git config --type=bool hooks.allownonascii)
+
+# Redirect output to stderr.
+exec 1>&2
+
+# Cross platform projects tend to avoid non-ASCII filenames; prevent
+# them from being added to the repository. We exploit the fact that the
+# printable range starts at the space character and ends with tilde.
+if [ "$allownonascii" != "true" ] &&
+	# Note that the use of brackets around a tr range is ok here, (it's
+	# even required, for portability to Solaris 10's /usr/bin/tr), since
+	# the square bracket bytes happen to fall in the designated range.
+	test $(git diff-index --cached --name-only --diff-filter=A -z $against |
+	  LC_ALL=C tr -d '[ -~]\0' | wc -c) != 0
+then
+	cat <<\EOF
+Error: Attempt to add a non-ASCII file name.
+
+This can cause problems if you want to work with people on other platforms.
+
+To be portable it is advisable to rename the file.
+
+If you know what you are doing you can disable this check using:
+
+  git config hooks.allownonascii true
+EOF
+	exit 1
+fi
+
+# If there are whitespace errors, print the offending file names and fail.
+exec git diff-index --check --cached $against --

super_agents-0.1.0/.weekend-commit-rewriter.git/hooks/pre-merge-commit.sample

@@ -0,0 +1,13 @@
+#!/bin/sh
+#
+# An example hook script to verify what is about to be committed.
+# Called by "git merge" with no arguments.  The hook should
+# exit with non-zero status after issuing an appropriate message to
+# stderr if it wants to stop the merge commit.
+#
+# To enable this hook, rename this file to "pre-merge-commit".
+
+. git-sh-setup
+test -x "$GIT_DIR/hooks/pre-commit" &&
+        exec "$GIT_DIR/hooks/pre-commit"
+:

super_agents-0.1.0/.weekend-commit-rewriter.git/hooks/pre-push.sample

@@ -0,0 +1,53 @@
+#!/bin/sh
+
+# An example hook script to verify what is about to be pushed.  Called by "git
+# push" after it has checked the remote status, but before anything has been
+# pushed.  If this script exits with a non-zero status nothing will be pushed.
+#
+# This hook is called with the following parameters:
+#
+# $1 -- Name of the remote to which the push is being done
+# $2 -- URL to which the push is being done
+#
+# If pushing without using a named remote those arguments will be equal.
+#
+# Information about the commits which are being pushed is supplied as lines to
+# the standard input in the form:
+#
+#   <local ref> <local oid> <remote ref> <remote oid>
+#
+# This sample shows how to prevent push of commits where the log message starts
+# with "WIP" (work in progress).
+
+remote="$1"
+url="$2"
+
+zero=$(git hash-object --stdin </dev/null | tr '[0-9a-f]' '0')
+
+while read local_ref local_oid remote_ref remote_oid
+do
+	if test "$local_oid" = "$zero"
+	then
+		# Handle delete
+		:
+	else
+		if test "$remote_oid" = "$zero"
+		then
+			# New branch, examine all commits
+			range="$local_oid"
+		else
+			# Update to existing branch, examine new commits
+			range="$remote_oid..$local_oid"
+		fi
+
+		# Check for WIP commit
+		commit=$(git rev-list -n 1 --grep '^WIP' "$range")
+		if test -n "$commit"
+		then
+			echo >&2 "Found WIP commit in $local_ref, not pushing"
+			exit 1
+		fi
+	fi
+done
+
+exit 0

super_agents-0.1.0/.weekend-commit-rewriter.git/hooks/pre-rebase.sample

@@ -0,0 +1,169 @@
+#!/bin/sh
+#
+# Copyright (c) 2006, 2008 Junio C Hamano
+#
+# The "pre-rebase" hook is run just before "git rebase" starts doing
+# its job, and can prevent the command from running by exiting with
+# non-zero status.
+#
+# The hook is called with the following parameters:
+#
+# $1 -- the upstream the series was forked from.
+# $2 -- the branch being rebased (or empty when rebasing the current branch).
+#
+# This sample shows how to prevent topic branches that are already
+# merged to 'next' branch from getting rebased, because allowing it
+# would result in rebasing already published history.
+
+publish=next
+basebranch="$1"
+if test "$#" = 2
+then
+	topic="refs/heads/$2"
+else
+	topic=`git symbolic-ref HEAD` ||
+	exit 0 ;# we do not interrupt rebasing detached HEAD
+fi
+
+case "$topic" in
+refs/heads/??/*)
+	;;
+*)
+	exit 0 ;# we do not interrupt others.
+	;;
+esac
+
+# Now we are dealing with a topic branch being rebased
+# on top of master.  Is it OK to rebase it?
+
+# Does the topic really exist?
+git show-ref -q "$topic" || {
+	echo >&2 "No such branch $topic"
+	exit 1
+}
+
+# Is topic fully merged to master?
+not_in_master=`git rev-list --pretty=oneline ^master "$topic"`
+if test -z "$not_in_master"
+then
+	echo >&2 "$topic is fully merged to master; better remove it."
+	exit 1 ;# we could allow it, but there is no point.
+fi
+
+# Is topic ever merged to next?  If so you should not be rebasing it.
+only_next_1=`git rev-list ^master "^$topic" ${publish} | sort`
+only_next_2=`git rev-list ^master           ${publish} | sort`
+if test "$only_next_1" = "$only_next_2"
+then
+	not_in_topic=`git rev-list "^$topic" master`
+	if test -z "$not_in_topic"
+	then
+		echo >&2 "$topic is already up to date with master"
+		exit 1 ;# we could allow it, but there is no point.
+	else
+		exit 0
+	fi
+else
+	not_in_next=`git rev-list --pretty=oneline ^${publish} "$topic"`
+	/usr/bin/perl -e '
+		my $topic = $ARGV[0];
+		my $msg = "* $topic has commits already merged to public branch:\n";
+		my (%not_in_next) = map {
+			/^([0-9a-f]+) /;
+			($1 => 1);
+		} split(/\n/, $ARGV[1]);
+		for my $elem (map {
+				/^([0-9a-f]+) (.*)$/;
+				[$1 => $2];
+			} split(/\n/, $ARGV[2])) {
+			if (!exists $not_in_next{$elem->[0]}) {
+				if ($msg) {
+					print STDERR $msg;
+					undef $msg;
+				}
+				print STDERR " $elem->[1]\n";
+			}
+		}
+	' "$topic" "$not_in_next" "$not_in_master"
+	exit 1
+fi
+
+<<\DOC_END
+
+This sample hook safeguards topic branches that have been
+published from being rewound.
+
+The workflow assumed here is:
+
+ * Once a topic branch forks from "master", "master" is never
+   merged into it again (either directly or indirectly).
+
+ * Once a topic branch is fully cooked and merged into "master",
+   it is deleted.  If you need to build on top of it to correct
+   earlier mistakes, a new topic branch is created by forking at
+   the tip of the "master".  This is not strictly necessary, but
+   it makes it easier to keep your history simple.
+
+ * Whenever you need to test or publish your changes to topic
+   branches, merge them into "next" branch.
+
+The script, being an example, hardcodes the publish branch name
+to be "next", but it is trivial to make it configurable via
+$GIT_DIR/config mechanism.
+
+With this workflow, you would want to know:
+
+(1) ... if a topic branch has ever been merged to "next".  Young
+    topic branches can have stupid mistakes you would rather
+    clean up before publishing, and things that have not been
+    merged into other branches can be easily rebased without
+    affecting other people.  But once it is published, you would
+    not want to rewind it.
+
+(2) ... if a topic branch has been fully merged to "master".
+    Then you can delete it.  More importantly, you should not
+    build on top of it -- other people may already want to
+    change things related to the topic as patches against your
+    "master", so if you need further changes, it is better to
+    fork the topic (perhaps with the same name) afresh from the
+    tip of "master".
+
+Let's look at this example:
+
+		   o---o---o---o---o---o---o---o---o---o "next"
+		  /       /           /           /
+		 /   a---a---b A     /           /
+		/   /               /           /
+	       /   /   c---c---c---c B         /
+	      /   /   /             \         /
+	     /   /   /   b---b C     \       /
+	    /   /   /   /             \     /
+    ---o---o---o---o---o---o---o---o---o---o---o "master"
+
+
+A, B and C are topic branches.
+
+ * A has one fix since it was merged up to "next".
+
+ * B has finished.  It has been fully merged up to "master" and "next",
+   and is ready to be deleted.
+
+ * C has not merged to "next" at all.
+
+We would want to allow C to be rebased, refuse A, and encourage
+B to be deleted.
+
+To compute (1):
+
+	git rev-list ^master ^topic next
+	git rev-list ^master        next
+
+	if these match, topic has not merged in next at all.
+
+To compute (2):
+
+	git rev-list master..topic
+
+	if this is empty, it is fully merged to "master".
+
+DOC_END