path: root/9agents/test/e2e.sh

#!/usr/bin/env bash
# End-to-end suite for 9agents: the read-only, fresh-from-disk 9P view of
# every harness's state.
#
# Usage: bash 9agents/test/e2e.sh <9agents> [<9ns>]   (zig build 9agents-itest)
#
# Part A always runs, entirely on fixtures: a fake home with fixture roots
# pinned by --root, a scratch XDG_RUNTIME_DIR registry, plan9port's 9p as
# the client. It proves: the posted name is listed, the roots walk and
# read, a transcript comes back byte-identical (cmp), credentials-shaped
# files are unreachable, a file appended after the daemon started is
# visible at once, writes answer EPERM, and --unix/--fd listen forms work.
#
# Part B (the money shot) runs against the REAL roots and the REAL
# registry only when the `agents` name is free, and only reads: the posted
# name in /run/user/<uid>/9p, 9p walks of the live ~/.claude, a real
# transcript byte-identical through the tree, the 9ns --mntgen mount of
# /mnt/9p/agents, and the live ~/.claude/.credentials.json unreachable.
# It is skipped (not failed) when a live daemon already owns the name.
#
# Exit 0 on success (or when the machine cannot run a part), 1 on failure.
set -u

H9=$(realpath "${1:?path to 9agents}")
HARNESS_TMP_XDG="${XDG_RUNTIME_DIR:-}"
[ $# -ge 2 ] && [ -n "$2" ] && NS=$(realpath "$2")
P9P=/usr/lib/plan9/bin/9p
TMP=$(mktemp -d "${TMPDIR:-/tmp}/9agents.XXXXXX")
PIDS=()
FAILED=0
PASSED=0

cleanup() {
    for p in "${PIDS[@]:-}"; do [ -n "$p" ] && kill "$p" 2>/dev/null; done
    rm -rf "$TMP"
    # part B posts into the *real* registry under a scratch name; a crash
    # between the post and the unpost would otherwise leave a socket there.
    [ -n "${REAL_SOCKET:-}" ] && rm -f "$REAL_SOCKET"
    return 0
}
trap cleanup EXIT

pass() { PASSED=$((PASSED + 1)); echo "ok   - $1"; }
fail() { FAILED=$((FAILED + 1)); echo "FAIL - $1"; shift; [ $# -gt 0 ] && printf '       %s\n' "$@"; }
expect_eq() { # name expected actual
    if [ "$2" = "$3" ]; then pass "$1"; else fail "$1" "expected: $(printf %q "$2")" "actual:   $(printf %q "$3")"; fi
}
expect_contains() { # name needle haystack
    case "$3" in *"$2"*) pass "$1" ;; *) fail "$1" "missing: $(printf %q "$2")" "in: $(printf %q "$3")" ;; esac
}
expect_missing() { # name needle haystack
    case "$3" in *"$2"*) fail "$1" "found: $(printf %q "$2")" "in: $(printf %q "$3")" ;; *) pass "$1" ;; esac
}

if [ ! -x "$P9P" ]; then
    echo "SKIP: plan9port 9p not at $P9P"; exit 0
fi
if ! unshare -Urm true 2>/dev/null; then
    echo "SKIP: unprivileged user namespaces unavailable"; exit 0
fi

start_daemon() { # args... -> sets DAEMON_PID, logs to $TMP/daemon.log
    "$H9" "$@" >"$TMP/daemon.log" 2>&1 &
    DAEMON_PID=$!
    PIDS+=("$DAEMON_PID")
}
wait_posted() { # socket path
    for _ in $(seq 1 100); do [ -S "$1" ] && return 0; sleep 0.05; done
    return 1
}
p9() { "$P9P" -a "unix!$1" "${@:2}"; }

# ============================================================================
echo "# part A: fixture roots, scratch registry"
# ============================================================================
export XDG_RUNTIME_DIR="$TMP"
REG="$TMP/9p"
FX="$TMP/home"
mkdir -p "$FX"

# The fixture home: five roots, a real-shaped claude project with a
# transcript, credentials-shaped files at several depths, codex sessions,
# an omp agent dir, hermes logs and dsh profiles.
mkfile() { mkdir -p "$(dirname "$1")"; printf '%s' "$2" > "$1"; }
mkfile "$FX/claude/projects/-tmp-proj/session-abc.jsonl" '{"type":"user","message":"first line"}
{"type":"assistant","message":"second line"}
'
mkfile "$FX/claude/projects/-tmp-proj/.credentials.json" 'STAY OUT'
mkfile "$FX/claude/projects/-tmp-proj/auth.json" 'STAY OUT'
mkfile "$FX/claude/projects/-tmp-proj/token.bin" 'STAY OUT'
mkfile "$FX/claude/history.jsonl" '{"display":"claude prompt one"}
{"display":"claude prompt two"}
'
mkdir -p "$FX/claude/skills/revu"
mkfile "$FX/claude/skills/revu/SKILL.md" '# revu skill'
mkfile "$FX/codex/sessions/2026/09/21/rollout-x.jsonl" '{"session":"codex one"}
'
mkfile "$FX/codex/session_index.jsonl" '{"id":"x"}
'
mkfile "$FX/codex/history.jsonl" '{"text":"codex prompt"}
'
mkfile "$FX/omp/agent/history.db" 'fake-history-db'
mkfile "$FX/omp/agent/config.yml" 'STAY OUT'
mkfile "$FX/hermes/auth.json" 'STAY OUT'
mkfile "$FX/hermes/logs/app.log" 'hermes log line
'
mkfile "$FX/hermes/state.db" 'fake-hermes-state'
mkfile "$FX/dsh/profiles/p1.yaml" 'name: p1'
mkfile "$FX/dsh/.credentials.yaml" 'STAY OUT'

start_daemon --root "claude=$FX/claude" --root "codex=$FX/codex" \
    --root "omp=$FX/omp" --root "hermes=$FX/hermes" --root "dsh=$FX/dsh" \
    --name agents
wait_posted "$REG/agents" || { fail "the daemon posts as agents" "$(cat "$TMP/daemon.log")"; exit 1; }

# 1. The posted name is listed.
expect_contains "posted name listed in the registry" agents "$(ls "$REG")"

# 2. The roots walk; a real transcript reads back byte-identical.
expect_contains "ls / shows the roots" claude "$(p9 "$REG/agents" ls /)"
expect_contains "ls / shows codex" codex "$(p9 "$REG/agents" ls /)"
expect_contains "ls / shows skills" skills "$(p9 "$REG/agents" ls /)"
p9 "$REG/agents" read /claude/projects/-tmp-proj/session-abc.jsonl > "$TMP/via-9p.jsonl" 2>"$TMP/read.err" \
    || fail "transcript read through the tree" "$(cat "$TMP/read.err")"
cmp -s "$TMP/via-9p.jsonl" "$FX/claude/projects/-tmp-proj/session-abc.jsonl" \
    && pass "transcript byte-identical through the tree (cmp)" \
    || fail "transcript byte-identical through the tree (cmp)" "differs"
expect_eq "history file reads" "$(cat "$FX/claude/history.jsonl")" "$(p9 "$REG/agents" read /claude/history)"
expect_eq "skills union mirrors the harness tree" "$(cat "$FX/claude/skills/revu/SKILL.md")" \
    "$(p9 "$REG/agents" read /skills/claude/revu/SKILL.md)"

# 3. A credentials-shaped file is unreachable anywhere in the tree.
PROJ_LS=$(p9 "$REG/agents" ls /claude/projects/-tmp-proj)
for bad in .credentials.json auth.json token.bin; do
    expect_missing "credentials-shaped $bad not listed" "$bad" "$PROJ_LS"
    p9 "$REG/agents" read "/claude/projects/-tmp-proj/$bad" >/dev/null 2>&1 \
        && fail "credentials-shaped $bad unreachable" "read succeeded" \
        || pass "credentials-shaped $bad unreachable"
done
expect_missing "hermes auth.json not listed" auth.json "$(p9 "$REG/agents" ls /hermes)"
expect_missing "omp config.yml not listed" config.yml "$(p9 "$REG/agents" ls /omp)"
expect_missing "dsh .credentials.yaml not listed" credentials.yaml "$(p9 "$REG/agents" ls /dsh)"
sleep 0.3
expect_contains "hermes logs still served" logs "$(p9 "$REG/agents" ls /hermes)"

# 3b. A file at the very top of a mirror root: its relative path is the
# bare name, the one case a join onto an empty directory path gets wrong.
expect_contains "a file at the top of a mirror root is listed" state.db "$(p9 "$REG/agents" ls /hermes)"
expect_eq "a file at the top of a mirror root reads back" "$(cat "$FX/hermes/state.db")" \
    "$(p9 "$REG/agents" read /hermes/state.db)"

# 3c. A directory bigger than the listing caps fails loudly. A short
# listing is indistinguishable from a small directory, so the daemon must
# never answer one: the read errors and the client sees it.
mkdir -p "$FX/dsh/wide"
seq 1 1100 | while read -r i; do : > "$FX/dsh/wide/f$(printf %05d "$i")"; done
if p9 "$REG/agents" ls /dsh/wide > "$TMP/wide.out" 2>"$TMP/wide.err"; then
    fail "an over-cap directory fails instead of truncating" "listed $(wc -l < "$TMP/wide.out") entries"
else
    pass "an over-cap directory fails instead of truncating ($(head -c 80 "$TMP/wide.err"))"
fi
rm -rf "$FX/dsh/wide"

# 4. Live visibility: a file appended after the daemon started.
printf '{"type":"assistant","message":"third line"}\n' >> "$FX/claude/projects/-tmp-proj/session-abc.jsonl"
expect_eq "append after start is visible immediately" \
    "$(cat "$FX/claude/projects/-tmp-proj/session-abc.jsonl")" \
    "$(p9 "$REG/agents" read /claude/projects/-tmp-proj/session-abc.jsonl)"
mkdir -p "$FX/claude/projects/-tmp-proj2"
mkfile "$FX/claude/projects/-tmp-proj2/session-new.jsonl" '{"new":true}
'
expect_contains "new session dir appears at once" -tmp-proj2 "$(p9 "$REG/agents" ls /claude/projects)"

# 5. The facts and the write refusal.
DAEMON_PID_TEXT=$(p9 "$REG/agents" read /pid)
expect_eq "pid fact answers the daemon's pid" "$DAEMON_PID" "$DAEMON_PID_TEXT"
[ -n "$(p9 "$REG/agents" read /uptime)" ] && pass "uptime fact answers" || fail "uptime fact answers" "empty"
printf 'x' | p9 "$REG/agents" write /pid >/dev/null 2>&1 \
    && fail "write answers EPERM" "write succeeded" \
    || pass "write answers EPERM"

# 6. The --unix listen form beside the post.
"$H9" --unix "$TMP/plain.sock" --no-post --root "claude=$FX/claude" >"$TMP/d2.log" 2>&1 &
PIDS+=($!)
for _ in $(seq 1 100); do [ -S "$TMP/plain.sock" ] && break; sleep 0.05; done
expect_eq "--unix listen form serves" "$(cat "$FX/claude/history.jsonl")" "$(p9 "$TMP/plain.sock" read /claude/history)"

# 7. The --fd listen form: one 9P session over a connected stream fd
# (the 9ns --spawn / socket-activation shape), driven through a
# socketpair: the daemon sees EOF when both ends close and exits.
if command -v python3 >/dev/null; then
    python3 - "$H9" "$FX" <<'EOF'
import os, socket, subprocess, sys
h9, fx = sys.argv[1], sys.argv[2]
a, b = socket.socketpair()
pid = os.fork()
if pid == 0:
    a.close()
    os.dup2(b.fileno(), 7)
    os.execv(h9, [h9, "--fd", "7", "--root", f"claude={fx}/claude"])
b.close()
a.close()
os.waitpid(pid, 0)
EOF
    pass "--fd listen form runs a session and exits on hangup"
else
    echo "skip - --fd form: python3 not available"
fi

kill "$DAEMON_PID" 2>/dev/null
wait "$DAEMON_PID" 2>/dev/null
sleep 0.2
[ -S "$REG/agents" ] && fail "SIGTERM unposts the name" "socket still there" || pass "SIGTERM unposts the name"

# ============================================================================
echo "# part C: /active, the derived view"
# ============================================================================
# A fake agent: a process whose argv[0] basename is a harness name, with a
# session record Claude Code's own shape. The proc root is the real /proc
# (the fixture seam is unit-tested); nothing real is ever killed, because
# the only process this part touches is the sleep it started itself.
ACT="$TMP/act"
mkdir -p "$ACT/bin" "$ACT/home/.claude/sessions" "$ACT/home/.claude/projects" "$ACT/rt/9p/zmx"
cp /bin/sleep "$ACT/bin/claude"
# A zmx that records how it was called and posts the name, as the real one
# does; a move must never be tested against the user's live sessions.
cat > "$ACT/bin/zmx" <<ZEOF
#!/bin/sh
echo "\$@" > "$ACT/zmx-argv"
: > "$ACT/rt/9p/zmx/\$2"
exit 0
ZEOF
chmod +x "$ACT/bin/zmx"

"$ACT/bin/claude" 600 &
AGENT=$!
PIDS+=("$AGENT")
sleep 0.3
# Field 22 of /proc/<pid>/stat, counted from the last ')' — the executable
# name can hold spaces and parentheses, so the line is never just split.
AGENT_START=$(sed 's/.*) //' "/proc/$AGENT/stat" | awk '{print $20}')
AGENT_CWD=$(readlink "/proc/$AGENT/cwd")
printf '{"pid":%d,"sessionId":"sess-e2e","cwd":"%s","name":"e2e-agent","status":"idle","procStart":"%s"}\n' \
    "$AGENT" "$AGENT_CWD" "$AGENT_START" > "$ACT/home/.claude/sessions/$AGENT.json"

act_roots() {
    echo --root "claude=$ACT/home/.claude" --root "codex=$ACT/home/.codex" \
         --root "omp=$ACT/home/.omp" --root "hermes=$ACT/home/.hermes" --root "dsh=$ACT/home/.dsh"
}

# First: the read-only default. No --allow-move, so zmx cannot be written.
XDG_RUNTIME_DIR="$ACT/rt" "$H9" --no-post --unix "$ACT/ro.sock" $(act_roots) >"$ACT/ro.log" 2>&1 &
PIDS+=("$!")
wait_posted "$ACT/ro.sock" || { fail "read-only daemon listens" "$(cat "$ACT/ro.log")"; }
expect_contains "the agent lists under its harness" "$AGENT" "$(p9 "$ACT/ro.sock" ls /active/claude)"
expect_eq "its session comes from the harness's own record" "sess-e2e" "$(p9 "$ACT/ro.sock" read "/active/claude/$AGENT/session")"
expect_eq "the route taken is named" "registry" "$(p9 "$ACT/ro.sock" read "/active/claude/$AGENT/via")"
expect_eq "the harness's own name is served" "e2e-agent" "$(p9 "$ACT/ro.sock" read "/active/claude/$AGENT/name")"
expect_eq "the harness's own status is served" "idle" "$(p9 "$ACT/ro.sock" read "/active/claude/$AGENT/status")"
if echo "nope" | p9 "$ACT/ro.sock" write "/active/claude/$AGENT/zmx" 2>/dev/null; then
    fail "without --allow-move the tree stays read-only" "the write was accepted"
else
    pass "without --allow-move the tree stays read-only"
fi
expect_eq "a refused move leaves the agent running" "yes" "$([ -d "/proc/$AGENT" ] && echo yes)"

# Now a daemon that allows moves.
XDG_RUNTIME_DIR="$ACT/rt" "$H9" --no-post --unix "$ACT/rw.sock" $(act_roots) \
    --allow-move --zmx "$ACT/bin/zmx" >"$ACT/rw.log" 2>&1 &
PIDS+=("$!")
wait_posted "$ACT/rw.sock" || { fail "move-allowing daemon listens" "$(cat "$ACT/rw.log")"; }

# Every refusal must come before anything is destroyed.
for bad in "has space" "../escape" "semi;colon"; do
    echo "$bad" | p9 "$ACT/rw.sock" write "/active/claude/$AGENT/zmx" 2>/dev/null
    if [ $? -eq 0 ]; then fail "an illegal zmx name is refused ($bad)"; else pass "an illegal zmx name is refused ($bad)"; fi
done
: > "$ACT/rt/9p/zmx/occupied"
echo "occupied" | p9 "$ACT/rw.sock" write "/active/claude/$AGENT/zmx" 2>/dev/null \
    && fail "a name a live session holds is refused" || pass "a name a live session holds is refused"
expect_eq "no refusal killed the agent" "yes" "$([ -d "/proc/$AGENT" ] && echo yes)"

# The move itself.
if echo "e2e-moved" | p9 "$ACT/rw.sock" write "/active/claude/$AGENT/zmx" 2>"$ACT/move.err"; then
    pass "a legal name moves the agent"
else
    fail "a legal name moves the agent" "$(cat "$ACT/move.err")"
fi
sleep 0.4
expect_eq "the agent it replaced is gone" "gone" "$([ -d "/proc/$AGENT" ] || echo gone)"
# The command is fixed by the harness: the client supplied only the name.
expect_eq "zmx ran the harness with its session resumed" \
    "run e2e-moved -d claude --resume sess-e2e" "$(cat "$ACT/zmx-argv")"
expect_missing "no client byte reached the command" "e2e-moved -d claude --resume sess-e2e;" "$(cat "$ACT/zmx-argv")"

# ============================================================================
echo "# part B: the real roots, the real registry (read-only)"
# ============================================================================
export XDG_RUNTIME_DIR="${HARNESS_TMP_XDG:-/run/user/$(id -u)}"
REAL_REG="$XDG_RUNTIME_DIR/9p"
# A scratch name, not `agents`: the whole point of part B is to read the real
# roots through the real registry, and that has nothing to do with which name
# the tree is posted under. Taking `agents` would mean this suite could only
# run while the machine's own 9agents.service was stopped — so it would either
# be skipped on any machine that actually uses the daemon, or fight it.
REAL_NAME="agents-itest-$$"
REAL_SOCKET="$REAL_REG/$REAL_NAME"
if [ -e "$REAL_SOCKET" ]; then
    # $$ collided with a leftover socket from a crashed run of this suite.
    echo "SKIP: scratch name $REAL_NAME is already taken"
else
    HOME_DIR=$(getent passwd "$(id -u)" | cut -d: -f6)
    start_daemon --name "$REAL_NAME"
    if wait_posted "$REAL_SOCKET"; then
        expect_contains "posted name listed in the real registry" "$REAL_NAME" "$(ls "$REAL_REG")"
        expect_contains "ls / shows the claude root" claude "$(p9 "$REAL_SOCKET" ls /)"
        # A real transcript, byte-identical through the tree.
        REAL_T=$(ls "$HOME_DIR/.claude/projects"/*/*.jsonl 2>/dev/null | head -n 1 || true)
        if [ -n "$REAL_T" ]; then
            REL="${REAL_T#"$HOME_DIR/.claude/projects/"}"
            p9 "$REAL_SOCKET" read "/claude/projects/$REL" > "$TMP/real-via-9p" 2>/dev/null \
                && cmp -s "$TMP/real-via-9p" "$REAL_T" \
                && pass "real transcript byte-identical through the tree" \
                || fail "real transcript byte-identical through the tree" "$REAL_T"
        else
            echo "skip - no real claude transcript found"
        fi
        # The credentials at ~/.claude are unreachable: no mount serves the
        # root dir, and the exclusion rule holds everywhere else.
        p9 "$REAL_SOCKET" read /claude/.credentials.json >/dev/null 2>&1 \
            && fail "live .credentials.json unreachable" "read succeeded" \
            || pass "live .credentials.json unreachable"
        CLAUDE_LS=$(p9 "$REAL_SOCKET" ls /claude)
        expect_missing "no credentials leaked into /claude" credentials "$CLAUDE_LS"
        # The money shot: the mntgen mount every interactive fish sees.
        if [ -n "$NS" ]; then
            OUT=$(timeout 60 "$NS" --mntgen -- sh -c "ls /mnt/9p/$REAL_NAME && head -c 200 /mnt/9p/$REAL_NAME/claude/history" 2>"$TMP/mntgen.err")
            RC=$?
            if [ $RC -eq 0 ]; then
                expect_contains "mntgen mount lists the agents tree" claude "$OUT"
                expect_contains "mntgen mount reads claude/history" claude "$OUT"
            else
                fail "mntgen money shot (exit $RC)" "$(cat "$TMP/mntgen.err")"
            fi
        else
            echo "skip - mntgen money shot: 9ns not provided"
        fi
        kill "$DAEMON_PID" 2>/dev/null
        wait "$DAEMON_PID" 2>/dev/null
        sleep 0.2
        [ -S "$REAL_SOCKET" ] && fail "stop unposts the real name" "socket still there" || pass "stop unposts the real name"
    else
        fail "daemon posts into the real registry" "$(cat "$TMP/daemon.log")"
    fi
fi

echo "# $PASSED passed, $FAILED failed"
[ "$FAILED" -eq 0 ]