GHSA-943m-6wx2-rc2j CVE-2026-47418 PYSEC-2026-2932
praisonai-platform: Project endpoints accept any project_id without workspace ownership check, cross-workspace read/update/delete IDOR
- Severity
- 8.1 / 10
- High · CVSS 3.1
- Exploit chance (EPSS)
- 0.41%
- Chance of exploitation in the next 30 days. Higher than 35.54% of scored vulnerabilities.
- Known exploited
- No
- Not in CISA's catalog
- Weakness
- CWE-639 Authorization Bypass Through User-Controlled Key
- Fix
0.1.4- Upgrade praisonai-platform to this version or later
Description
Summary
Type: Insecure Direct Object Reference. The project CRUD endpoints (GET / PATCH / DELETE /workspaces/{workspace_id}/projects/{project_id} and GET .../{project_id}/stats) gate access on require_workspace_member(workspace_id) only, then resolve project_id through ProjectService.get(project_id) / update(project_id, ...) / delete(project_id) / get_stats(project_id). None of these calls thread workspace_id through to constrain the lookup. A user who is a member of any workspace W1 can read, modify, delete, or read stats for projects that belong to a different workspace W2.
File: src/praisonai-platform/praisonai_platform/services/project_service.py, lines 47-108; route handlers at src/praisonai-platform/praisonai_platform/api/routes/projects.py, lines 51-108.
Root cause: identical to the agent and issue IDORs in this codebase. The route accepts workspace_id from URL, uses it solely for the membership gate, then calls ProjectService.get(project_id) which is session.get(Project, project_id) — a primary-key-only lookup with no workspace_id predicate. update and delete call self.get(project_id) first, inheriting the gap. get_stats likewise has no workspace check.
Affected Code
File 1: src/praisonai-platform/praisonai_platform/services/project_service.py, lines 47-108.
class ProjectService:
...
async def get(self, project_id: str) -> Optional[Project]:
"""Get project by ID."""
return await self._session.get(Project, project_id) # <-- BUG: no workspace_id predicate
async def update(
self,
project_id: str,
...
) -> Optional[Project]:
project = await self.get(project_id) # <-- inherits the gap
...
async def delete(self, project_id: str) -> bool:
project = await self.get(project_id) # <-- inherits the gap
...
async def get_stats(self, project_id: str) -> dict:
... # <-- also no workspace check; returns issue counts for any project
File 2: src/praisonai-platform/praisonai_platform/api/routes/projects.py, lines 51-108.
@router.get("/{project_id}", response_model=ProjectResponse)
async def get_project(
workspace_id: str,
project_id: str,
user: AuthIdentity = Depends(require_workspace_member),
session: AsyncSession = Depends(get_db),
):
svc = ProjectService(session)
project = await svc.get(project_id) # <-- workspace_id never threaded through
if project is None:
raise HTTPException(status_code=404, detail="Project not found")
return ProjectResponse.model_validate(project)
@router.patch("/{project_id}", response_model=ProjectResponse)
async def update_project(...):
svc = ProjectService(session)
project = await svc.update(project_id, title=body.title, ...) # <-- writes to any project in the DB
@router.delete("/{project_id}", ...)
async def delete_project(...):
deleted = await svc.delete(project_id) # <-- deletes any project in the DB
@router.get("/{project_id}/stats")
async def project_stats(...):
return await svc.get_stats(project_id) # <-- returns stats for any project in the DB
Why it's wrong: workspace_id from the route is treated as a UI hint (gates "are you in some workspace W?") rather than an authoritative predicate (should also gate "is the project you are addressing actually inside W?"). The MemberService in this same codebase uses a composite (workspace_id, user_id) key and demonstrates the safe pattern; the project service simply did not apply it.
Exploit Chain
- Attacker registers a workspace
W_attacker(where they are a member) and harvests a target project UUIDP_T. Project IDs leak through the activity feed (act_svc.logrecordsentity_id), issue records (every issue carriesproject_id), webhook payloads, error messages, exported issue dumps, or operator screenshots. State: attacker holdsP_T. - Attacker authenticates and sends
GET /workspaces/W_attacker/projects/P_T.require_workspace_member(W_attacker, attacker)passes. State: control flow entersget_projectwithworkspace_id=W_attacker, project_id=P_T. ProjectService.get(P_T)runssession.get(Project, "P_T"), which isSELECT * FROM projects WHERE id = 'P_T' LIMIT 1with noworkspace_idfilter. The row is returned:title,description(often the project's confidential roadmap),status,lead_type,lead_id,icon,created_at,workspace_id(the foreign workspace's UUID is itself disclosed). State: response body is the JSON-serialised foreign project.- Attacker repeats with
PATCH /workspaces/W_attacker/projects/P_Tand{"title": "<reset>", "description": "<wiped>", "status": "archived"}.update_projectcallssvc.update(P_T, ...)and mutates the foreign row. State: target project is silently re-titled, re-described, and archived. - Attacker calls
DELETE /workspaces/W_attacker/projects/P_Tto delete the foreign project entirely. State: target project is gone (every issue still referencing it now has a danglingproject_id). - Attacker calls
GET /workspaces/W_attacker/projects/P_T/statsto read aggregate issue counts (open/closed/in-progress) for the foreign project — useful for competitive intelligence even when full-issue read is not possible. - Final state: any attacker with one workspace-member token can enumerate, exfiltrate, rewrite, and delete every project in the multi-tenant deployment given the project UUIDs.
Security Impact
Severity: sec-high. CVSS: network attack, low complexity, low privileges, no user interaction, scope unchanged, high confidentiality (project content + cross-workspace metadata via the leaked workspace_id field), high integrity (arbitrary writes / deletes), no availability claim (issue rows survive parent-project deletion).
Attacker capability: read, edit, archive, delete, and stats-fingerprint any project in the multi-tenant deployment given the project UUID. Beyond plain content disclosure, the response also includes workspace_id, allowing the attacker to map the deployment's workspace topology (which workspaces exist, which projects each owns).
Preconditions: praisonai-platform is deployed multi-tenant; the attacker has any membership token; the target project's UUID is known or guessable.
Differential: source-inspection-verified end-to-end. The asymmetry between ProjectService.get(project_id) (no workspace check) and MemberService.get(workspace_id, user_id) (composite key check) confirms the gap. With the suggested fix below, ProjectService.get(workspace_id, project_id) returns None for foreign-workspace projects and the route handler returns 404.
Suggested Fix
Same shape as the companion agent and issue advisories. Make the resource-lookup query include the workspace predicate; treat foreign-workspace rows as 404.
--- a/src/praisonai-platform/praisonai_platform/services/project_service.py
+++ b/src/praisonai-platform/praisonai_platform/services/project_service.py
@@ -45,9 +45,12 @@ class ProjectService:
await self._session.flush()
return project
- async def get(self, project_id: str) -> Optional[Project]:
- """Get project by ID."""
- return await self._session.get(Project, project_id)
+ async def get(self, workspace_id: str, project_id: str) -> Optional[Project]:
+ """Get project by ID, scoped to a workspace."""
+ stmt = select(Project).where(
+ Project.id == project_id, Project.workspace_id == workspace_id
+ )
+ return (await self._session.execute(stmt)).scalar_one_or_none()
async def update(
self,
+ workspace_id: str,
project_id: str,
...
) -> Optional[Project]:
- project = await self.get(project_id)
+ project = await self.get(workspace_id, project_id)
- async def delete(self, project_id: str) -> bool:
+ async def delete(self, workspace_id: str, project_id: str) -> bool:
- project = await self.get(project_id)
+ project = await self.get(workspace_id, project_id)
- async def get_stats(self, project_id: str) -> dict:
+ async def get_stats(self, workspace_id: str, project_id: str) -> dict:
+ # Also constrain the underlying issue counts query by workspace_id.
Update the route handlers in routes/projects.py to thread workspace_id through every call. The same single-key-lookup pattern is filed separately for AgentService, IssueService, CommentService, and LabelService.
Affected versions
PyPI
| Package | Vulnerable | Fixed in |
|---|---|---|
| praisonai-platform | < 0.1.4 |
0.1.4 |
How to fix
Move to the fixed version or later with your package manager, update the lockfile, and run your tests. If the package only arrives through another dependency, upgrade that dependency first. If none of its releases pulls a fixed version yet, force the version with your package manager's override.
praisonai-platform PyPI → 0.1.4
- pip
pip install --upgrade "praisonai-platform>=0.1.4"- poetry
poetry add "praisonai-platform>=0.1.4"- uv
uv add "praisonai-platform>=0.1.4"
Severity breakdown
CVSS 3.1 8.1
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:N
- Attack vector
- Network
- Attack complexity
- Low
- Privileges required
- Low
- User interaction
- None
- Scope
- Unchanged
- Confidentiality
- High
- Integrity
- High
- Availability
- None
Weakness
CWE-639: Authorization Bypass Through User-Controlled Key
The system's authorization functionality does not prevent one user from gaining access to another user's data or record by modifying the key value identifying the data.
How to prevent it
- Architecture and Design: For each and every data access, ensure that the user has sufficient privilege to access the record that is being requested.
- Architecture and Design, Implementation: Make sure that the key that is used in the lookup of a specific user's record is not controllable externally by the user or that any tampering can be detected.
- Architecture and Design: Use encryption in order to make it more difficult to guess other legitimate values of the key or associate a digital signature with the key so that the server can verify that there has been no tampering.
References
Advisories
Packages
Other links
Timeline
- Published
- Last updated by the source
- Latest EPSS score
Related advisories
| Advisory | Severity | Why it's related |
|---|---|---|
| GHSA-xxgv-vgvj-qvxh PraisonAI: Platform members can rewrite shared labels and owner issue labels without owner/admin authorization | High | Also affects praisonai-platform |
| GHSA-c78w-2q4r-68r7 PraisonAI: Platform member PATCH routes allow owner resource rewrites and project lead reassignment delete bypass | High | Also affects praisonai-platform |
| GHSA-mxmx-rh57-jx58 PraisonAI: Platform members can delete owner issue dependencies through member-owned related issues | High | Also affects praisonai-platform |
| GHSA-2fjj-qqg8-fg7x praisonai-platform: Authorization Bypass Through User-Controlled Key | Medium | Also affects praisonai-platform |
| GHSA-cwj8-7gp2-ggcw praisonai-platform: default JWT signing secret 'dev-secret-change-me' enables token forgery | Critical | Also affects praisonai-platform |
| GHSA-f38v-77qj-h4jq praisonai-platform 0.1.4 still boots on the hardcoded JWT secret dev-secret-change-me (default-open production guard) | Critical | Also affects praisonai-platform |
| GHSA-rh39-9c67-59mh PraisonAI: Missing ownership check on DELETE endpoints allows members to delete others' content in Platform API | High | Also affects praisonai-platform |
| GHSA-7p8g-6c6g-h9w7 praisonai-platform: Agent endpoints accept any agent_id without workspace ownership check, cross-workspace read/update/delete IDOR | High | Also affects praisonai-platform |
MergeHarbor checks every repository's dependencies against this database, ranks what matters by exploitability and reachability, routes it to the owning team and opens the fix pull request.
Check your repos with MergeHarbor