CVE-2026-57148

PraisonAI is a multi-agent teams system. Prior to 0.1.6, praisonai_platform/services/auth_service.py falls back to the public dev-secret-change-me HS256 signing key when PLATFORM_JWT_SECRET is unset, while the startup and token-issuance guards are disabled because PLATFORM_ENV also defaults to dev. An unauthenticated attacker can sign a JWT containing an attacker-chosen sub value, and AuthService._verify_token() accepts it as an authenticated identity, enabling user or workspace-owner impersonation when a target identifier is known. This vulnerability is fixed in praisonai-platform 0.1.6.

PraisonAI is a multi-agent teams system. Prior to 0.1.6, praisonai_platform/services/auth_service.py falls back to the public dev-secret-change-me HS256 signing key when PLATFORM_JWT_SECRET is unset, while the startup and token-issuance guards are disabled because PLATFORM_ENV also defaults to dev. An unauthenticated attacker can sign a JWT containing an attacker-chosen sub value, and AuthService._verify_token() accepts it as an authenticated identity, enabling user or workspace-owner impersonation when a target identifier is known. This vulnerability is fixed in praisonai-platform 0.1.6.

• Affected: praisonai-platform (PyPI) <= 0.1.4 — including 0.1.4, the version GHSA-3qg8-5g3r-79v5 declares as the patch; main HEAD 8acf77c531e624c46d3d61dcae37e9942e90972c is also affected. File src/praisonai-platform/praisonai_platform/services/auth_service.py

• CWE: CWE-1188 (Insecure Default Initialization) + CWE-798 (Use of Hard-coded Credentials) -> CWE-287 (Improper Authentication)

Overview

GHSA-3qg8-5g3r-79v5 (Critical) reported that praisonai-platform's JWT signing secret defaulted to the hardcoded literal "dev-secret-change-me", and that the production guard meant to prevent this was default-open (it only fired when PLATFORM_ENV != "dev", but PLATFORM_ENV defaults to "dev"). That advisory declares the issue patched in >= 0.1.4. It is not. The shipped praisonai-platform==0.1.4 (and current main) still resolves the signing key to "dev-secret-change-me" in any deployment that does not explicitly set PLATFORM_JWT_SECRET, because the 0.1.4 change merely duplicated the same default-open guard into a second function instead of failing closed. An unauthenticated attacker reads the literal from the public source, forges a JWT with an arbitrary sub, and is authenticated as that user — including a workspace owner.

Impact

Any deployment that runs praisonai-platform 0.1.4 without explicitly exporting a strong PLATFORM_JWT_SECRET signs and verifies session JWTs with the publicly known key "dev-secret-change-me". The package's documented entry point — python -m praisonai_platform --host 0.0.0.0 --port 8000 (equivalently uvicorn praisonai_platform.api.app:app --host 0.0.0.0) — sets neither PLATFORM_JWT_SECRET nor PLATFORM_ENV, so this is the default state, not an edge case. A repository-wide search finds both variables only at the two guard sites and in test fixtures; no shipped Dockerfile, compose file, or deployment doc sets either.

Consequences:

• Complete authentication bypass (unauthenticated). Knowing only the public default secret read from source, an attacker mints HS256({"sub": <user id>, "email": …, "exp": <future>}, "dev-secret-change-me"). The platform's own verifier accepts it and returns an authenticated identity for the attacker-chosen sub — no account and no prior access required. This is the headline defect: the identical break GHSA-3qg8 was scored 9.8 for.

• Workspace-owner takeover (when a target owner's id is known). Forging the sub of a workspace owner satisfies require_workspace_member / require_workspace_owner and the owner-gated routes, yielding owner-level read/update/delete of every resource in that workspace plus member/role management. uuid4 user ids are unguessable, so impersonating a specific owner additionally requires learning that owner's id — which any co-member can read directly from GET /{workspace_id}/members (returns List[MemberResponse], each carrying user_id and role, to any holder of require_workspace_member), and which also surfaces in logs and referrals. The end state matches the three Critical advisories of the 0.1.4 wave (this one, plus GHSA-c2m8-4gcg-v22g 9.6 and GHSA-h8q5-cp56-rr65).

• Resource destruction / lock-out (A:H). Owner impersonation reaches DELETE /workspaces/{workspace_id} (gated by require_workspace_owner), which deletes the entire workspace and every contained resource, and DELETE /{workspace_id}/members/{user_id}, which evicts legitimate members — irrecoverable denial of the workspace to its rightful users.

• Affected population: every default (no PLATFORM_JWT_SECRET) deployment of 0.1.4 — the version users upgrade to specifically because GHSA-3qg8 told them 0.1.4 is fixed.

PR:N / AC:L apply to the authentication-bypass primitive: minting a valid session for a known sub needs no account, only the public secret. Targeted takeover of a specific owner additionally requires that owner's user id (readable by any co-member from the member-list response above, or recoverable from logs / prior exposure); this conditions…

PraisonAI is a multi-agent teams system. Prior to 0.1.6, praisonai_platform/services/auth_service.py falls back to the public dev-secret-change-me HS256 signing key when PLATFORM_JWT_SECRET is unset, while the startup and token-issuance guards are disabled because PLATFORM_ENV also defaults to dev. An unauthenticated attacker can sign a JWT containing an attacker-chosen sub value, and AuthService._verify_token() accepts it as an authenticated identity, enabling user or workspace-owner impersonation when a target identifier is known. This vulnerability is fixed in praisonai-platform 0.1.6.

PraisonAI is a multi-agent teams system. Prior to 0.1.6, praisonai_platform/services/auth_service.py falls back to the public dev-secret-change-me HS256 signing key when PLATFORM_JWT_SECRET is unset, while the startup and token-issuance guards are disabled because PLATFORM_ENV also defaults to dev. An unauthenticated attacker can sign a JWT containing an attacker-chosen sub value, and AuthService._verify_token() accepts it as an authenticated identity, enabling user or workspace-owner impersonation when a target identifier is known. This vulnerability is fixed in praisonai-platform 0.1.6.

• Affected: praisonai-platform (PyPI) <= 0.1.4 — including 0.1.4, the version GHSA-3qg8-5g3r-79v5 declares as the patch; main HEAD 8acf77c531e624c46d3d61dcae37e9942e90972c is also affected. File src/praisonai-platform/praisonai_platform/services/auth_service.py

• CWE: CWE-1188 (Insecure Default Initialization) + CWE-798 (Use of Hard-coded Credentials) -> CWE-287 (Improper Authentication)

Overview

GHSA-3qg8-5g3r-79v5 (Critical) reported that praisonai-platform's JWT signing secret defaulted to the hardcoded literal "dev-secret-change-me", and that the production guard meant to prevent this was default-open (it only fired when PLATFORM_ENV != "dev", but PLATFORM_ENV defaults to "dev"). That advisory declares the issue patched in >= 0.1.4. It is not. The shipped praisonai-platform==0.1.4 (and current main) still resolves the signing key to "dev-secret-change-me" in any deployment that does not explicitly set PLATFORM_JWT_SECRET, because the 0.1.4 change merely duplicated the same default-open guard into a second function instead of failing closed. An unauthenticated attacker reads the literal from the public source, forges a JWT with an arbitrary sub, and is authenticated as that user — including a workspace owner.

Impact

Any deployment that runs praisonai-platform 0.1.4 without explicitly exporting a strong PLATFORM_JWT_SECRET signs and verifies session JWTs with the publicly known key "dev-secret-change-me". The package's documented entry point — python -m praisonai_platform --host 0.0.0.0 --port 8000 (equivalently uvicorn praisonai_platform.api.app:app --host 0.0.0.0) — sets neither PLATFORM_JWT_SECRET nor PLATFORM_ENV, so this is the default state, not an edge case. A repository-wide search finds both variables only at the two guard sites and in test fixtures; no shipped Dockerfile, compose file, or deployment doc sets either.

Consequences:

• Complete authentication bypass (unauthenticated). Knowing only the public default secret read from source, an attacker mints HS256({"sub": <user id>, "email": …, "exp": <future>}, "dev-secret-change-me"). The platform's own verifier accepts it and returns an authenticated identity for the attacker-chosen sub — no account and no prior access required. This is the headline defect: the identical break GHSA-3qg8 was scored 9.8 for.

• Workspace-owner takeover (when a target owner's id is known). Forging the sub of a workspace owner satisfies require_workspace_member / require_workspace_owner and the owner-gated routes, yielding owner-level read/update/delete of every resource in that workspace plus member/role management. uuid4 user ids are unguessable, so impersonating a specific owner additionally requires learning that owner's id — which any co-member can read directly from GET /{workspace_id}/members (returns List[MemberResponse], each carrying user_id and role, to any holder of require_workspace_member), and which also surfaces in logs and referrals. The end state matches the three Critical advisories of the 0.1.4 wave (this one, plus GHSA-c2m8-4gcg-v22g 9.6 and GHSA-h8q5-cp56-rr65).

• Resource destruction / lock-out (A:H). Owner impersonation reaches DELETE /workspaces/{workspace_id} (gated by require_workspace_owner), which deletes the entire workspace and every contained resource, and DELETE /{workspace_id}/members/{user_id}, which evicts legitimate members — irrecoverable denial of the workspace to its rightful users.

• Affected population: every default (no PLATFORM_JWT_SECRET) deployment of 0.1.4 — the version users upgrade to specifically because GHSA-3qg8 told them 0.1.4 is fixed.

PR:N / AC:L apply to the authentication-bypass primitive: minting a valid session for a known sub needs no account, only the public secret. Targeted takeover of a specific owner additionally requires that owner's user id (readable by any co-member from the member-list response above, or recoverable from logs / prior exposure); this conditions…

• Affected: praisonai-platform (PyPI) <= 0.1.4 — including 0.1.4, the version GHSA-3qg8-5g3r-79v5 declares as the patch; main HEAD 8acf77c531e624c46d3d61dcae37e9942e90972c is also affected. File src/praisonai-platform/praisonai_platform/services/auth_service.py

• CWE: CWE-1188 (Insecure Default Initialization) + CWE-798 (Use of Hard-coded Credentials) -> CWE-287 (Improper Authentication)

Overview

GHSA-3qg8-5g3r-79v5 (Critical) reported that praisonai-platform's JWT signing secret defaulted to the hardcoded literal "dev-secret-change-me", and that the production guard meant to prevent this was default-open (it only fired when PLATFORM_ENV != "dev", but PLATFORM_ENV defaults to "dev"). That advisory declares the issue patched in >= 0.1.4. It is not. The shipped praisonai-platform==0.1.4 (and current main) still resolves the signing key to "dev-secret-change-me" in any deployment that does not explicitly set PLATFORM_JWT_SECRET, because the 0.1.4 change merely duplicated the same default-open guard into a second function instead of failing closed. An unauthenticated attacker reads the literal from the public source, forges a JWT with an arbitrary sub, and is authenticated as that user — including a workspace owner.

Impact

Any deployment that runs praisonai-platform 0.1.4 without explicitly exporting a strong PLATFORM_JWT_SECRET signs and verifies session JWTs with the publicly known key "dev-secret-change-me". The package's documented entry point — python -m praisonai_platform --host 0.0.0.0 --port 8000 (equivalently uvicorn praisonai_platform.api.app:app --host 0.0.0.0) — sets neither PLATFORM_JWT_SECRET nor PLATFORM_ENV, so this is the default state, not an edge case. A repository-wide search finds both variables only at the two guard sites and in test fixtures; no shipped Dockerfile, compose file, or deployment doc sets either.

Consequences:

• Complete authentication bypass (unauthenticated). Knowing only the public default secret read from source, an attacker mints HS256({"sub": <user id>, "email": …, "exp": <future>}, "dev-secret-change-me"). The platform's own verifier accepts it and returns an authenticated identity for the attacker-chosen sub — no account and no prior access required. This is the headline defect: the identical break GHSA-3qg8 was scored 9.8 for.

• Workspace-owner takeover (when a target owner's id is known). Forging the sub of a workspace owner satisfies require_workspace_member / require_workspace_owner and the owner-gated routes, yielding owner-level read/update/delete of every resource in that workspace plus member/role management. uuid4 user ids are unguessable, so impersonating a specific owner additionally requires learning that owner's id — which any co-member can read directly from GET /{workspace_id}/members (returns List[MemberResponse], each carrying user_id and role, to any holder of require_workspace_member), and which also surfaces in logs and referrals. The end state matches the three Critical advisories of the 0.1.4 wave (this one, plus GHSA-c2m8-4gcg-v22g 9.6 and GHSA-h8q5-cp56-rr65).

• Resource destruction / lock-out (A:H). Owner impersonation reaches DELETE /workspaces/{workspace_id} (gated by require_workspace_owner), which deletes the entire workspace and every contained resource, and DELETE /{workspace_id}/members/{user_id}, which evicts legitimate members — irrecoverable denial of the workspace to its rightful users.

• Affected population: every default (no PLATFORM_JWT_SECRET) deployment of 0.1.4 — the version users upgrade to specifically because GHSA-3qg8 told them 0.1.4 is fixed.

PR:N / AC:L apply to the authentication-bypass primitive: minting a valid session for a known sub needs no account, only the public secret. Targeted takeover of a specific owner additionally requires that owner's user id (readable by any co-member from the member-list response above, or recoverable from logs / prior exposure); this conditions…