registry / @dev-sergeev/pi-gigachat / 0.4.0

@dev-sergeev/pi-gigachat@0.4.0

GLM 5.2 and Qwen 3.5 provider for pi-coding-agent via a GigaChat-compatible API

AI Security Review

scanned 4d ago · by lpm-firewall-ai

LPM flags this version as an AI-agent control-surface risk. Installation changes the shared Pi agent's default provider and model without an opt-in check. This modifies a foreign agent control surface beyond the package's own extension files.

Static reason

One or more suspicious static signals were detected.

Trigger

npm installation with postinstall hooks enabled.

Impact

Existing agent defaults are overridden, directing subsequent default model selection to the package's GigaChat provider.

Mechanism

The hook reads the agent settings, overwrites defaultProvider and defaultModel, then replaces the settings file and writes a marker.

Policy narrative

The automatic postinstall hook locates Pi's shared agent directory and replaces its saved provider and model defaults. It performs this mutation regardless of existing selections and without requesting consent. The settings overwrite is an active installation-time change to another agent's control surface, rather than merely registering package-owned extension files.

AI rationale

Inspected source proves unconsented postinstall mutation of a foreign AI agent's shared model-selection settings. This meets the supplied blocking policy even without evidence of credential exfiltration.

Evidence

package.jsonscripts/postinstall.mjssettings.json.pi-gigachat-default-model

Decision evidence

public snapshot

with false-positive risk.

@dev-sergeev/pi-gigachat@0.4.0

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Source & flagged code