Investigate in your agent.

Turn this advisory into a focused investigation. Ask your agent to check your project’s versions, explain the evidence, and suggest a compatible fix.

Your agent needs remote MCP support and access to https://cve.new/mcp.

Read the setup guide

Skip to content
cve.new Create a CVE report
Check your stack
CVE-2026-56838high7.8

PraisonAI recipe.run_stream skips dangerous-tool policy enforcement

Impact

PraisonAI versions 4.5.87 to 4.6.59 have a weakness where the recipe.run_stream() function skips policy enforcement for dangerous tools. This allows a recipe to execute commands without proper authorization.

Summarized by @cf/meta/llama-3.3-70b-instruct-fp8-fast. Verify details against the original sources.
Applies when

As a result, a recipe that honestly declares execute_command in TEMPLATE.yaml requires.tools is denied by recipe.run(), but reaches the execution engine through recipe.run_stream() with allow_dangerous_tools=False.

Affected packages

This source has no supported package-version data for matching. You can read the advisory here, but this record alone cannot power a dependency check.

Affected products

Vendor version evidence. Product records are separate from automatic dependency matching.

MervinPraison · GitHub_M

PraisonAI

1 range
GitHub_M version evidence

Compare with this publisher’s ranges. This does not inspect your installation.

affected

>= 4.5.87, < 4.6.59

unaffected

unknown

Other versions: unknown

Original source description

PraisonAI is a multi-agent teams system. From version 4.5.87 to before version 4.6.59, PraisonAI recipe execution blocks default-denied dangerous tools unless the caller explicitly passes allow_dangerous_tools=True. The normal recipe.run() path enforces this with _check_tool_policy(). The streaming path, recipe.run_stream(), loads the same recipe, checks dependencies, and then calls _execute_recipe() without running the dangerous-tool policy check. As a result, a recipe that honestly declares execute_command in TEMPLATE.yaml requires.tools is denied by recipe.run(), but reaches the execution engine through recipe.run_stream() with allow_dangerous_tools=False. This issue has been patched in version 4.6.59.

Inside the record

Evidence available

Package identity · not supplied
Version evidence
Remediation guidance · not supplied
Weakness classification
Additional publishers

Describes the evidence in this view, not whether the product is safe.

Original records

CVE-2026-56838

View the latest unmodified CVE JSON from the official source. It may be newer than this imported page.

cve.new JSON →

Underlying weaknesses

The source’s classification of what caused the vulnerability.

CWE-78 ↗

CWE-78: Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')

GitHub_M

Related →
CWE-693 ↗

CWE-693: Protection Mechanism Failure

GitHub_M

Related →
CWE-863 ↗

CWE-863: Incorrect Authorization

GitHub_M

Related →
Contributing organizations · 2
GitHub_M

Assigning organization

2026-10-23
CISA-ADP

Additional data publisher

2026-10-29

What changed

Follow changes
No substantive changes have been recorded here yet. History starts when cve.new observes a change after tracking is enabled; earlier edits are not reconstructed.

Large field values are marked as truncated previews. Original source snapshots remain retained.

References

1 sources
github.com/MervinPraison/PraisonAI/security/advisories/GHSA-v847-hxxw-3pxg
exploitGitHub_M · CISA-ADP