AgentLint

Errors0

Warnings0

Instructions1

Skills1

Plugins1

MCP servers2

01 · Precedence

Effective Instruction Graph

Instructions are ordered from project root toward the working directory. Nearer guidance may alter the effective boundary.

.AGENTS.md:3

denyfilesystem.delete- Never delete files without explicit user approval.

.AGENTS.md:4

approve-beforegit.push- Ask for confirmation before `git push`.

02 · Authority

Evidence-linked routes below show supported scanned relationships, never positional correspondence. No tool is invoked.

Evidence-linked multi-step routes

safe-demoplugin/demo/.codex-plugin/plugin.json.mcp.jsonplugin/demo/.mcp.jsondocsconfigures · plugin/demo/.mcp.json

safe-demoplugin/demo/.codex-plugin/plugin.json.mcp.jsonplugin/demo/.mcp.jsonworkspace-filesconfigures · plugin/demo/.mcp.json

Other evidence-linked edges

AGENTS.mdAGENTS.mddenyfilesystem.deleterequested authority

AGENTS.mdAGENTS.mdapprove-beforegit.pushrequested authority

safe-demoplugin/demo/.codex-plugin/plugin.jsonbundlesread-workspaceplugin/demo/skills/read-workspace/SKILL.md

03 · Evidence

Filter the audit list, search its escaped evidence, then expand a row for impact and a bounded remediation.

No deterministic findings

The scanned configuration passed the current rule set. This does not certify that every behavior is safe.

AgentLint is a deterministic preflight, not a security certification. A clean report means no current rule matched—not that every behavior is safe.

04 · Coverage

The configuration surface included in this scan.

4config files

1AGENTS files

1skills

1plugins

1MCP configs