@topy-ai/maggie 0.7.57 → 0.7.59

package/README-zh-TW.md

@@ -29,7 +29,9 @@ npx @topy-ai/maggie init --scope project --agent all
 npx @topy-ai/maggie doctor --scope project --project .
 ```
 
-目前 npm 版本為 `0.7.57`：新增 `maggie security public-check`，對你**無法控制**的站點做被動、等同瀏覽器的檢查（安全 headers、TLS、首頁洩漏的版本與其已知 CVE、SPF/DMARC、security.txt），讓你能通知站主；只限真人互動執行，絕不探測、提交或掃描，主動掃描（web-scan、zap-baseline）與主機檢查仍須先驗證域名。
+目前 npm 版本為 `0.7.58`：新增 `maggie security fix-plan`（由發現整理出的修復 runbook：檔案權限、防火牆、套件/apt 更新、入侵防護、Web server 設定、EOL；產生一份你自己在 host lock 下、備份後執行的 shell script，Maggie 從不代為套用），並把 lynis tarball 以 sha256 pin 住，讓 `host-audit` 能透過 SSH 做深度主機稽核。
+
+0.7.57 新增 `maggie security public-check`，對你**無法控制**的站點做被動、等同瀏覽器的檢查（安全 headers、TLS、首頁洩漏的版本與其已知 CVE、SPF/DMARC、security.txt），讓你能通知站主；只限真人互動執行，絕不探測、提交或掃描，主動掃描（web-scan、zap-baseline）與主機檢查仍須先驗證域名。
 
 0.7.56 新增 Maggie Security（`maggie-security` skill 與 `maggie security` 指令，共 46 個指令、29 個 MCP 工具），涵蓋套件、runtime、資料庫、Web server、Linux 發行版與 WordPress 的漏洞公告與 end-of-life 日期，以及 secrets、程式碼、網站與主機檢查、報告、CI gate 與每日監控。
 
@@ -141,7 +143,7 @@ structure drift 或重複 treatment product 時 fail closed。
 deployment、memory、feedback 和 MaggieDash。內容先 draft/review，外部寫入、
 publish 與 production deployment 需要明確確認。
 
-目前 npm 版本為 `0.7.57`，包含 public-check、Maggie Security、batch 31 排程 agent routines、報告收件匣與無人值守 agent 限制、PHP runtime 盤點與主機鎖、pull 網站改寫、query 點子與社群佇列檢查，batch 30 MaggieDash 先裝核心、release privacy gate、feedback 備援、社群連結與圖片檢查、大量頁面產生與 app 檢查，batch 29 預約規則、部署、app QA、助理與社群功能，以及 batch 28 產品 app、assistant tools、mobile、product release、skills catalog 與 bridge host 功能、batch 26、27 feedback 修正與 idempotent `maggie update`與 Booking email-template 合約/CLI validator 同步。本版完成 2026-09-24 booking feedback batch：Calendar grid/free-slot entry 與 host modal focus-restoration evidence。本版新增 Starter marketplace provisioning plan；本版完成 feedback batches 22、23：加入 declarative browser interaction evidence、design markup/style/content-key gate、compiler source anchoring、entity double-escape protection、`update --help` read-only、noindex hypothetical indexability、可過期 baseline acknowledgement、DOM-only link extraction regression，以及 MaggieDash 0.2.10 的 content override、短期 browser session、CDN-safe public-session cache contracts。authentication 和 isolated dev server 仍由 host 負責。`maggie-booking` workflow、傳統 email/password、Stripe、Resend、public booking flow 及既有安全邊界維持不變；provider-bound video generation 與 transactional delivery 仍記錄在 issue #39，前一批延後功能仍在 issue #38，不會直接成為 active memory。0.7.33 記錄並審查一筆更正回饋；確認 0.7.32 已正確處理 batch index，沒有修改已驗證的 shared behavior。0.7.32 加入 feedback 一基索引支援、MaggieDash panel 的 source/freshness/error evidence、Consent Mode 與實際 write/no-op reconciliation、SEO baseline recapture、sitemap origin rebasing、content-diff scope、migration ledger reconciliation、deployment credential preflight、release provenance/runtime preflight、icon release gate、opt-in blog auto-publish gate，以及 least-privilege VPS deployer。也修復 top-level `maggie feedback` dispatcher parity。0.7.31 加入 update 差異摘要、atomic installer copy、subset install manifest merge，以及 inventory drift 時 doctor non-zero。0.7.30 補上 feedback tracking issue closeout gate 與 batch 14/15 audit ledger。0.7.29 修正 npm package landing page 使用英文 README；0.7.28 加入 versioned Gemini model policy、明確 fallback 與 provenance、bundled-first CLI
+目前 npm 版本為 `0.7.59`，包含 fix-plan、public-check、Maggie Security、batch 31 排程 agent routines、報告收件匣與無人值守 agent 限制、PHP runtime 盤點與主機鎖、pull 網站改寫、query 點子與社群佇列檢查，batch 30 MaggieDash 先裝核心、release privacy gate、feedback 備援、社群連結與圖片檢查、大量頁面產生與 app 檢查，batch 29 預約規則、部署、app QA、助理與社群功能，以及 batch 28 產品 app、assistant tools、mobile、product release、skills catalog 與 bridge host 功能、batch 26、27 feedback 修正與 idempotent `maggie update`與 Booking email-template 合約/CLI validator 同步。本版完成 2026-09-24 booking feedback batch：Calendar grid/free-slot entry 與 host modal focus-restoration evidence。本版新增 Starter marketplace provisioning plan；本版完成 feedback batches 22、23：加入 declarative browser interaction evidence、design markup/style/content-key gate、compiler source anchoring、entity double-escape protection、`update --help` read-only、noindex hypothetical indexability、可過期 baseline acknowledgement、DOM-only link extraction regression，以及 MaggieDash 0.2.10 的 content override、短期 browser session、CDN-safe public-session cache contracts。authentication 和 isolated dev server 仍由 host 負責。`maggie-booking` workflow、傳統 email/password、Stripe、Resend、public booking flow 及既有安全邊界維持不變；provider-bound video generation 與 transactional delivery 仍記錄在 issue #39，前一批延後功能仍在 issue #38，不會直接成為 active memory。0.7.33 記錄並審查一筆更正回饋；確認 0.7.32 已正確處理 batch index，沒有修改已驗證的 shared behavior。0.7.32 加入 feedback 一基索引支援、MaggieDash panel 的 source/freshness/error evidence、Consent Mode 與實際 write/no-op reconciliation、SEO baseline recapture、sitemap origin rebasing、content-diff scope、migration ledger reconciliation、deployment credential preflight、release provenance/runtime preflight、icon release gate、opt-in blog auto-publish gate，以及 least-privilege VPS deployer。也修復 top-level `maggie feedback` dispatcher parity。0.7.31 加入 update 差異摘要、atomic installer copy、subset install manifest merge，以及 inventory drift 時 doctor non-zero。0.7.30 補上 feedback tracking issue closeout gate 與 batch 14/15 audit ledger。0.7.29 修正 npm package landing page 使用英文 README；0.7.28 加入 versioned Gemini model policy、明確 fallback 與 provenance、bundled-first CLI
 dispatch、`maggie --version`、feedback batch review 聚合與重複偵測、marketplace enrichment evidence、booking
 worker/resolver evidence，以及 read-only migration preflight。它也包含 host-owned mobile app surface contract、signed-in
 camera-state QA、直接 Astro route resolution、correlated feedback batch，以及

package/README.md

@@ -89,7 +89,19 @@ Maggie keeps the existing project foundation and asks for decisions before
 shared routes, analytics, or publishing boundaries change. The current
 package ships 23 installable skills and a local-first MaggieDash foundation.
 
-This package is `0.7.57`. It adds `maggie security public-check`: passive,
+This package is `0.7.59`. It makes `maggie security host-audit` (the deep
+lynis audit over SSH) robust on hosts where a storage test can stall: lynis runs
+in its own process group under a watchdog, and the bundled profile skips the ZFS
+storage/filesystem tests that hang, so the audit completes instead of timing out.
+
+The 0.7.58 release added `maggie security fix-plan` — a
+reviewed remediation runbook built from the open findings (file permissions,
+firewall, package/apt updates, intrusion prevention, web-server config, end of
+life). It writes a shell script you run yourself, under a host lock, after a
+backup; Maggie never applies it. This release also pins the `lynis` tarball
+(sha256) so `host-audit` can run the deep host audit over SSH.
+
+The 0.7.57 release added `maggie security public-check`: passive,
 browser-equivalent checks of sites you do **not** control (security headers,
 TLS, versions disclosed on the home page and their known CVEs, SPF/DMARC,
 `security.txt`), so you can tell an owner about a visible problem. It is
@@ -97,7 +109,7 @@ interactive-only and never probes, submits or scans; the active scanners
 (`web-scan`, `zap-baseline`) and host checks stay verified-only.
 
 The 0.7.56 release added Maggie Security, a new
-`maggie-security` skill and `maggie security` command family (47 commands, 30
+`maggie-security` skill and `maggie security` command family (48 commands, 30
 MCP tools, 7 contracts under `contracts/maggie-security/`).
 
 - **Advisories for the whole stack.** Packages (npm, pnpm, Yarn, Bun,
@@ -844,8 +856,8 @@ artifact schemas.
 Recommended upgrade sequence for the current release:
 
 ```bash
-npx @topy-ai/maggie@0.7.57 update --scope project --project . --force
-npx @topy-ai/maggie@0.7.57 cleanup --scope project --project .
+npx @topy-ai/maggie@0.7.59 update --scope project --project . --force
+npx @topy-ai/maggie@0.7.59 cleanup --scope project --project .
 ```
 
 Maintainers should pass npm credentials through the repository helper, never

package/bundled-skills/maggie-security/SKILL.md

@@ -50,6 +50,7 @@ Follow [Maggie Memory Hook](../../references/memory-hook.md) at invocation and c
 | "are we being attacked / were we hacked?" | `maggie security logs-traffic --file access.log`, `maggie security logs-sigma --file access.log`, `maggie security logs-ioc --file access.log`; servers: `maggie security logs-remote --ssh web1=user@host` | `security_logs_check` |
 | "is my server hardened?" | `maggie security host-quick --ssh web1=user@host` (or `--print-collector` for the person to run), `maggie security host-nginx-config --ssh web1=user@host`, deep: `maggie security host-audit --ssh web1=user@host` | `security_host_check` |
 | "protect my server" (CrowdSec / fail2ban / Cloudflare) | `maggie security protect-plan --host web1 [--engine crowdsec\|fail2ban]` → the person reviews and runs it under the host lock; afterwards `maggie security protect-plan verify --plan FILE --ssh web1=user@host` | `security_protect_plan` |
+| "give me the fixes to run" | after the checks: `maggie security fix-plan --script fix.sh --output RUNBOOK.md` — a reviewed runbook (file permissions, firewall, apt updates, intrusion prevention, nginx, EOL) built from the findings; the person reviews it and runs it themselves under the host lock. Maggie never applies it. | — |
 | "monitor uptime / tell me when the site is down" | `maggie security uptime-init` (verified domains) → `maggie security uptime-deploy --hostname status.example.com` (plan) → `maggie security uptime-status --gatus URL`; jobs report with `maggie security uptime-push` | `security_uptime_status` |
 | "how do we compare to OWASP?" / "threat model" | `maggie security checklist --level 1`, `maggie security threat-model --include-findings` | `security_checklist_get`, `security_threat_model` |
 | "is this npm package safe to add?" | `maggie security preinstall PKG[@VERSION]` before installing (W5); never install on `block` | `security_package_vet` |

package/bundled-skills/maggie-security/references/commands.md

@@ -122,4 +122,5 @@ Formats: nginx/Apache combined and vcombined, Caddy JSON, Cloudflare Logpush JSO
 | Command | What it does |
 |---|---|
 | `maggie security report --kind weekly\|daily_ops\|incident\|gate [--audience owner\|technical] [--save]` | Markdown + JSON report from the latest runs; `--save` writes the `report_save` payload |
+| `maggie security fix-plan [--priority p0..p4] [--script fix.sh] [--output RUNBOOK.md]` | a reviewed remediation runbook from the open findings: grouped by category (file permissions, firewall, package/apt updates, intrusion prevention, web-server config, EOL), each step the commands the findings themselves carry (never invented). Writes a shell script you run yourself, under `maggie deployment host-lock`, after a backup and with a second SSH session open. Maggie never applies it. |
 | `maggie security mcp [--tools p0\|all] [--print-config claude\|codex]` | local stdio MCP server |

package/bundled-tools/clis/maggie_security_host.py

@@ -266,18 +266,38 @@ def lynis_findings(ctx: core.Context, data: dict, host: str) -> list[dict]:
 
 
 LYNIS_PROFILE = """# Maggie lynis profile: no upload, skip tests that do not apply to web servers
+# and skip the storage/filesystem enumeration that hangs on ZFS hosts
+# (the remote watchdog below is the backstop if anything else stalls).
 upload=no
 skip-test=PRNT-2307
 skip-test=PRNT-2308
 skip-test=STRG-1840
 skip-test=STRG-1842
+skip-test=STRG-1846
+skip-test=STRG-1902
 skip-test=NAME-4402
 skip-test=NAME-4404
 skip-test=KRB-5108
 skip-test=LDAP-2219
-skip-test=STRG-1902
+skip-test=FILE-6310
+skip-test=FILE-6311
+skip-test=FILE-6329
+skip-test=FILE-6330
+skip-test=FILE-6332
+skip-test=FILE-6336
+skip-test=FILE-6354
+skip-test=FILE-6362
+skip-test=FILE-6368
+skip-test=FILE-6374
+skip-test=FILE-6376
 """
 
+# A ZFS host can wedge a lynis storage test for a very long time, and killing the
+# ssh client leaves the stuck test running and holding the channel. So we run
+# lynis in its own process group under a remote watchdog that SIGKILLs the whole
+# group; the command always returns and flushes, finished or not.
+LYNIS_WATCHDOG_SECONDS = 600
+
 
 def host_audit(ctx: core.Context):
     args = ctx.args
@@ -329,10 +349,21 @@ def remote_lynis(ctx: core.Context, alias: str, target: str) -> dict:
         subprocess.run(["ssh", "-o", "BatchMode=yes", target, f"umask 077 && mkdir -p {remote_dir}"], check=True, timeout=60)
         subprocess.run(["scp", "-q", "-o", "BatchMode=yes", str(tarball), f"{target}:{remote_dir}/lynis.tar.gz"], check=True, timeout=300)
         profile = shlex.quote(LYNIS_PROFILE)
-        script = (f"cd {remote_dir} && tar xzf lynis.tar.gz && printf %s {profile} > maggie.prf && cd lynis && "
-                  f"$( [ $(id -u) != 0 ] && sudo -n true 2>/dev/null && echo sudo -n ) ./lynis audit system --quick --no-colors --no-log "
-                  f"--profile {remote_dir}/maggie.prf --report-file {remote_dir}/report.dat --logfile /dev/null >/dev/null 2>&1; cat {remote_dir}/report.dat")
-        proc = subprocess.run(["ssh", "-o", "BatchMode=yes", target, script], capture_output=True, text=True, timeout=1800, check=False)
+        # Run lynis with sudo when available (a full audit needs root). Run it in
+        # its own session (setsid) under a watchdog that SIGKILLs the whole process
+        # group if it overruns, so a stuck storage test cannot hang the session or
+        # leave an orphan holding the ssh channel. Read report.dat back with the
+        # same sudo (a root run writes it 0600, so a plain cat reads nothing).
+        inner = (f"setsid ./lynis audit system --quick --no-colors --no-log "
+                 f"--profile {remote_dir}/maggie.prf --report-file {remote_dir}/report.dat --logfile /dev/null >/dev/null 2>&1 & "
+                 f"LP=$!; ( sleep {LYNIS_WATCHDOG_SECONDS}; kill -9 -$LP 2>/dev/null ) & WD=$!; "
+                 f"wait $LP 2>/dev/null; kill $WD 2>/dev/null")
+        script = (f"SUDO=$( [ $(id -u) != 0 ] && sudo -n true 2>/dev/null && echo 'sudo -n' ); "
+                  f"cd {remote_dir} && tar xzf lynis.tar.gz && printf %s {profile} > maggie.prf && cd lynis && "
+                  f"$SUDO sh -c {shlex.quote(inner)} </dev/null >/dev/null 2>&1; "
+                  f"$SUDO cat {remote_dir}/report.dat 2>/dev/null")
+        proc = subprocess.run(["ssh", "-o", "BatchMode=yes", target, script], input="", capture_output=True, text=True,
+                              timeout=LYNIS_WATCHDOG_SECONDS + 180, check=False)
         report = proc.stdout
     finally:
         subprocess.run(["ssh", "-o", "BatchMode=yes", target, f"rm -rf {remote_dir}"], timeout=60, check=False)
@@ -385,7 +416,8 @@ def nginx_findings(ctx: core.Context, nginx: str, php: str, key_perms: str, host
     findings = []
     seen = set()
 
-    def add(fid: str, title: str, severity: str, summary: str, snippet: str | None = None, where: str = "", asvs: list[str] | None = None) -> None:
+    def add(fid: str, title: str, severity: str, summary: str, snippet: str | None = None, where: str = "", asvs: list[str] | None = None,
+            steps: list[str] | None = None) -> None:
         key = (fid, where)
         if key in seen:
             return
@@ -393,7 +425,7 @@ def nginx_findings(ctx: core.Context, nginx: str, php: str, key_perms: str, host
         findings.append(core.make_finding(kind="misconfiguration", command="host-nginx-config", primary_id=fid, site=ctx.site,
                                           title=title + (f" ({where})" if where else ""), severity=severity, severity_source="maggie",
                                           confidence="likely", target={"host": host, "path": where or None}, asvs_ids=asvs or [], rule_id=fid,
-                                          remediation={"summary": summary, "steps": ["sudo nginx -t && sudo systemctl reload nginx"],
+                                          remediation={"summary": summary, "steps": steps or ["sudo nginx -t && sudo systemctl reload nginx"],
                                                        "snippets": [{"kind": "nginx", "text": snippet}] if snippet else [],
                                                        "references": ["https://nginx.org/en/docs/"], "cheat_sheet": None}))
 
@@ -471,7 +503,7 @@ def nginx_findings(ctx: core.Context, nginx: str, php: str, key_perms: str, host
         parts = line.split(None, 2)
         if len(parts) == 3 and (int(parts[0][-1]) > 0 or int(parts[0][-2]) > 4 or parts[1] != "root"):
             add("ngx-tls-key-perms", f"TLS private key is readable by others ({parts[0]} {parts[1]})", "high", "chmod 600 the key and chown root.",
-                where=parts[2], asvs=["V13.3.1"])
+                where=parts[2], asvs=["V13.3.1"], steps=[f"sudo chmod 600 {shlex.quote(parts[2])} && sudo chown root:root {shlex.quote(parts[2])}"])
     return findings
 
 

package/bundled-tools/clis/maggie_security_report.py

@@ -153,6 +153,9 @@ def why(finding: dict) -> str:
         return "Anyone who sees the file can use this key."
     if kev:
         return "Attackers are already using this flaw (on the KEV list)."
+    if kind in ("misconfiguration", "header", "tls", "exposure", "attack-signal"):
+        sev = {"critical": "A serious", "high": "A serious", "medium": "A moderate", "low": "A minor", "info": "A minor"}[finding["severity"]]
+        return f"{sev} configuration weakness an attacker could use."
     severity = {"critical": "A serious", "high": "A serious", "medium": "A moderate", "low": "A minor", "info": "A minor"}[finding["severity"]]
     return f"{severity} known flaw (CVE) affects the version you run."
 
@@ -184,8 +187,10 @@ def group_actions(findings: list[dict]) -> list[dict]:
         items = core.sort_findings(groups[key])
         lead = items[0]
         if key[0] == "single" or len(items) == 1:
+            la = lead.get("affected") or {}
             actions.append({"priority": lead["priority"], "title": action_title(lead), "why": why(lead), "how": how(lead),
-                            "ids": [lead["id"]], "count": 1, "lead": lead})
+                            "ids": [lead["id"]], "count": 1, "lead": lead,
+                            "package": la.get("name"), "ecosystem": la.get("ecosystem"), "fix_version": (lead.get("fix") or {}).get("recommended")})
             continue
         affected = lead["affected"]
         eco = affected.get("ecosystem")
@@ -205,7 +210,8 @@ def group_actions(findings: list[dict]) -> list[dict]:
                   + ("; attackers are using at least one (on the KEV list)." if exploited else "."))
         step_from = next((f for f in items if (f.get("fix") or {}).get("recommended") == target_version), lead)
         actions.append({"priority": lead["priority"], "title": title, "why": reason, "how": how(step_from),
-                        "ids": [f["id"] for f in items], "count": len(items), "lead": lead})
+                        "ids": [f["id"] for f in items], "count": len(items), "lead": lead,
+                        "package": affected.get("name"), "ecosystem": eco, "fix_version": target_version})
     return sorted(actions, key=lambda a: (core.PRIORITIES.index(a["priority"]), -a["count"]))
 
 
@@ -219,19 +225,25 @@ def explain_jargon(text: str, used: set[str]) -> str:
     return re.sub(r"\b(" + "|".join(JARGON) + r")\b", sub, text)
 
 
-def build(ctx: core.Context) -> dict:
-    args = ctx.args
-    since = args.since
-    if args.kind == "gate" and since in (None, "last-report"):
-        since = "6h"  # the gate reads this deploy's runs only
+def resolve_since(ctx: core.Context, since: str | None, kind: str = "weekly") -> str:
+    if kind == "gate" and since in (None, "last-report"):
+        return "6h"  # the gate reads this deploy's runs only
     if since in (None, "last-report"):
         last = last_report_time(ctx.state)
         if last:
             hours = max(1, int((core.utcnow() - core.parse_time(last)).total_seconds() // 3600) + 1)
-            since = f"{hours}h"
-        else:
-            since = "7d"
-    runs = latest_runs(ctx.state, since, args.runs)
+            return f"{hours}h"
+        return "7d"
+    return since
+
+
+def collect_findings(ctx: core.Context, since: str, run_ids: list[str] | None) -> tuple[list[dict], list[dict], set[str]]:
+    """The current findings for a site: the latest run per command merged with the ledger.
+
+    Returns (findings with their derived status, the runs used, the set of fixed ids).
+    Shared by `report` and `fix-plan` so both see exactly the same state.
+    """
+    runs = latest_runs(ctx.state, since, run_ids)
     if not runs:
         raise core.UsageError(f"no security runs in the last {since}; run e.g. `maggie security vuln-watch` first")
     ledger = core.Ledger(ctx.state.path("ledger.jsonl"))
@@ -254,7 +266,13 @@ def build(ctx: core.Context) -> dict:
         first_seen = (state or {}).get("first_seen") or finding.get("first_seen")
         finding = dict(finding, status="new" if first_seen and core.parse_time(first_seen) >= floor and status in ("new", "open") else status)
         findings.append(finding)
-    findings = core.sort_findings(findings)
+    return core.sort_findings(findings), runs, fixed_ids
+
+
+def build(ctx: core.Context) -> dict:
+    args = ctx.args
+    since = resolve_since(ctx, args.since, args.kind)
+    findings, runs, fixed_ids = collect_findings(ctx, since, args.runs)
     actionable = [f for f in findings if f["status"] in ("new", "open")]
     grouped = group_actions([f for f in actionable if f["priority"] in ("p0", "p1")])
     p0 = sum(1 for a in grouped if a["priority"] == "p0")
@@ -271,7 +289,7 @@ def build(ctx: core.Context) -> dict:
 
     built = set(maggie_security.area_commands())
     missing = [c for c in EXPECTED.get(site_type, EXPECTED["unknown"]) if c not in ran and c in built]
-    period = {"from": core.iso(floor), "to": core.iso()}
+    period = {"from": core.iso(core.utcnow() - core.parse_duration(since)), "to": core.iso()}
     actions = [f for f in actionable if f["priority"] in ("p0", "p1")]
     report_id = f"{core.iso()[:10]}-{args.kind}-{core.sha256_bytes(json.dumps(sorted(f['id'] for f in findings)).encode())[:6]}"
     data = {
@@ -490,6 +508,160 @@ def incident_cmd(ctx: core.Context) -> dict:
     return data
 
 
+# --------------------------------------------------------------------------- fix-plan (a reviewed remediation runbook)
+
+# A fix-plan turns the open findings into an ordered runbook of the fixes the
+# findings themselves already carry. It never connects to a host and never runs
+# anything — the operator reviews it and runs it under a host lock. It only
+# emits remediation text the checks authored; it does not invent commands.
+
+FIX_CATEGORIES = [
+    ("secrets-permissions", "Secrets and file permissions", "Quick, high-value fixes — do these first."),
+    ("firewall", "Firewall", "Close everything except the ports you actually serve."),
+    ("packages", "Package updates", "Apply in a maintenance window; reboot if the kernel or libc changed."),
+    ("intrusion-prevention", "Intrusion prevention", "Use `maggie security protect-plan` for a reviewed fail2ban/CrowdSec setup."),
+    ("web-server", "Web server configuration", "Edit the server/vhost files, then `sudo nginx -t && sudo systemctl reload nginx`."),
+    ("end-of-life", "End-of-life upgrades", "Plan a move to a supported release line."),
+    ("other", "Other", "Review and apply as appropriate."),
+]
+
+
+def fix_category(finding: dict) -> str:
+    kind = finding.get("kind")
+    rule = (finding.get("source") or {}).get("rule_id") or ""
+    command = (finding.get("source") or {}).get("command") or ""
+    title = finding.get("title", "").lower()
+    if kind == "secret" or "readable by others" in title or "world-readable" in title or "permission" in title:
+        return "secrets-permissions"
+    if "firewall" in title or rule.startswith("hq-firewall"):
+        return "firewall"
+    if "fail2ban" in title or "crowdsec" in title or "intrusion" in title:
+        return "intrusion-prevention"
+    if kind == "vulnerability":
+        return "packages"
+    if kind == "eol":
+        return "end-of-life"
+    if command in ("host-nginx-config", "headers", "tls") or kind in ("header", "tls"):
+        return "web-server"
+    return "other"
+
+
+def fix_commands(action: dict) -> tuple[list[str], list[tuple[str, str]], list[str]]:
+    """(shell commands, (kind, snippet) paste-ins, TODO notes) faithfully from the lead finding."""
+    lead = action["lead"]
+    rem = lead.get("remediation") or {}
+    steps = [s for s in (rem.get("steps") or []) if not s.startswith(("Listed by CISA", "Check you can log in"))]
+    snippets = [(s.get("kind") or "text", s.get("text") or "") for s in (rem.get("snippets") or []) if s.get("text")]
+    commands = [s for s in steps if re.match(r"^\s*(sudo |apt|chmod|chown|ufw|systemctl|nginx |visudo|sed |install )", s)]
+    todos = [] if (commands or snippets) else [rem.get("summary") or action["title"]]
+    return commands, snippets, todos
+
+
+def fix_plan(ctx: core.Context) -> dict:
+    args = ctx.args
+    since = resolve_since(ctx, args.since)
+    findings, _runs, _fixed = collect_findings(ctx, since, args.runs)
+    actionable = [f for f in findings if f["status"] in ("new", "open")]
+    if args.priority:
+        floor = core.PRIORITIES.index(args.priority)
+        actionable = [f for f in actionable if core.PRIORITIES.index(f["priority"]) <= floor]
+    actions = group_actions(actionable)
+    buckets: dict[str, list[dict]] = {key: [] for key, _, _ in FIX_CATEGORIES}
+    for action in actions:
+        buckets[fix_category(action["lead"])].append(action)
+
+    host = next((f.get("target", {}).get("host") for f in actionable if (f.get("target") or {}).get("host")), None)
+    lock_target = host or "<user@host>"
+    sh: list[str] = [
+        "#!/usr/bin/env bash",
+        f"# Remediation runbook for {ctx.site}" + (f" ({host})" if host else "") + f" — generated {core.iso()[:10]} by maggie security fix-plan.",
+        "#",
+        "# REVIEW EVERY LINE BEFORE RUNNING. Maggie did not and will not run any of this.",
+        "# Run it yourself, as the operator, after:",
+        "#   1. taking a snapshot/backup of the server,",
+        "#   2. opening a SECOND SSH session (so a bad sshd/firewall change cannot lock you out),",
+        f"#   3. taking the change lock: maggie deployment host-lock acquire --host {lock_target} --purpose fix-plan",
+        "# Package upgrades belong in a maintenance window. Test nginx with `nginx -t` before every reload.",
+        "set -euo pipefail",
+        "",
+    ]
+    md: list[str] = [f"# Remediation runbook — {ctx.site}" + (f" ({host})" if host else ""), "",
+                     "Generated by `maggie security fix-plan`. **Review before running — Maggie never runs these.** "
+                     "Take a backup, keep a second SSH session open, and work under "
+                     f"`maggie deployment host-lock acquire --host {lock_target} --purpose fix-plan`.", ""]
+    apt_packages: list[str] = []
+    total = 0
+    for key, label, note in FIX_CATEGORIES:
+        items = buckets.get(key) or []
+        if not items:
+            continue
+        sh += [f"# =================== {label} ===================", f"# {note}", ""]
+        md += [f"## {label}", f"_{note}_", ""]
+        for action in items:
+            total += 1
+            sh.append(f"# [{action['priority']}] {action['title']} — {action['why']}")
+            md.append(f"- **{action['title']}** — {action['why']}")
+            if key == "packages":
+                # Build the fix command ourselves; never copy the finding's step text
+                # (it reads "apt-get install --only-upgrade X  (fixed in Y)" — not valid shell).
+                name = action.get("package")
+                osfam = (action.get("ecosystem") or "").split(":")[0]
+                if name and osfam in ("Ubuntu", "Debian") and action.get("fix_version"):
+                    apt_packages.append(name)
+                    sh.append(f"#   {name} -> {action['fix_version']} (in the apt batch above)")
+                else:
+                    sh.append(f"#   TODO (manual): {action['how']}")
+                    md.append(f"  - {action['how']}")
+                sh.append("")
+                continue
+            commands, snippets, todos = fix_commands(action)
+            for cmd in commands:
+                sh.append(cmd)
+            for kind, text in snippets:
+                sh.append(f"#   {kind} config — paste into the right block:")
+                sh += [f"#     {line}" for line in text.splitlines()]
+                md.append(f"  - `{kind}`: `{text}`")
+            for todo in todos:
+                sh.append(f"#   TODO (manual): {todo}")
+                md.append(f"  - {todo}")
+            if commands:
+                md.append("  - run: " + "; ".join(f"`{c}`" for c in commands))
+            sh.append("")
+        md.append("")
+    if apt_packages:
+        pkgs = " ".join(sorted(set(apt_packages)))
+        insert = ["# Apply available Ubuntu/Debian security updates for the affected packages:",
+                  "sudo apt-get update",
+                  f"sudo apt-get install --only-upgrade {pkgs}",
+                  "# Or apply all pending security updates at once:",
+                  "# sudo unattended-upgrade -d    # or: sudo apt-get upgrade", ""]
+        anchor = sh.index("# =================== Package updates ===================") + 3
+        sh[anchor:anchor] = insert
+        md.append("### apt batch")
+        md.append(f"```\nsudo apt-get update && sudo apt-get install --only-upgrade {pkgs}\n```\n")
+    runbook = "\n".join(md).rstrip() + "\n"
+    script = "\n".join(sh).rstrip() + "\n"
+    if args.script:
+        path = Path(args.script)
+        path.parent.mkdir(parents=True, exist_ok=True)
+        path.write_text(script, encoding="utf-8")
+        path.chmod(0o755)
+    if getattr(args, "output", None):
+        out = Path(args.output)
+        out.parent.mkdir(parents=True, exist_ok=True)
+        out.write_text(runbook, encoding="utf-8")
+    return {"schema": "maggie-security-fix-plan.v1", "site": ctx.site, "host": host, "generated_at": core.iso(),
+            "actions": total, "apt_packages": sorted(set(apt_packages)),
+            "script_path": str(args.script) if args.script else None,
+            "note": "Review before running. Maggie never applies these changes.", "markdown": runbook}
+
+
+def _fix(p: argparse.ArgumentParser) -> None:
+    p.add_argument("--priority", choices=["p0", "p1", "p2", "p3", "p4"], help="only include findings at or above this priority")
+    p.add_argument("--script", help="write the runnable shell runbook to this path (chmod 755)")
+    p.add_argument("--runs", action="append", help="use these run ids instead of the latest per command")
+
+
 def _incident(p: argparse.ArgumentParser) -> None:
     p.add_argument("action", choices=["start", "log", "close", "show", "list"])
     p.add_argument("--id")
@@ -503,4 +675,6 @@ def _incident(p: argparse.ArgumentParser) -> None:
 COMMANDS = [core.Command("report", "one per-site report from the latest runs and the ledger (Markdown + JSON, report_save payload)", [],
                          configure, run, findings=False),
             core.Command("incident", "timeline for a CVE response or other incident (local records; used by report --kind incident)", [], _incident,
-                         incident_cmd, findings=False)]
+                         incident_cmd, findings=False),
+            core.Command("fix-plan", "a reviewed remediation runbook from the open findings (Markdown + a shell script); never applied by Maggie",
+                         [], _fix, fix_plan, findings=False, routine_safe=False)]

package/bundled-tools/clis/security_data/tools.lock.json

@@ -202,7 +202,13 @@
       "forbidden_flags": [
         "--upload"
       ],
-      "notes": "GPL: run from the upstream tarball on the host; never vendored."
+      "notes": "GPL: downloaded on use, sha256-verified, uploaded to a private /tmp dir on the host and removed after; never vendored into the package.",
+      "assets": {
+        "any": {
+          "url": "https://downloads.cisofy.com/lynis/lynis-3.1.5.tar.gz",
+          "sha256": "8d2c6652ba60116a82514522b666ca77293f4bfc69f1e581028769f7ebb52ba4"
+        }
+      }
     },
     "goaccess": {
       "version": "1.9.4",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@topy-ai/maggie",
-  "version": "0.7.57",
+  "version": "0.7.59",
   "description": "Install and manage Maggie Skills for AI coding agents",
   "license": "MIT",
   "type": "module",