acacian/Aegis AI Agent Security Gate

Scan for ungoverned AI calls, enforce policies, and detect prompt injection in CI/CD.

Trust Signals

Scorecard Score: not yet scored
Maintenance Recency: Activelast commit May 5, 2026
License: MIT

Pinned Snippet

workflow.ymlSHA-pinned

uses: acacian/aegis@58d5fa5f50e8135bf4c64d4ce91908596249e1dc # v0.9.5

tags can be moved; commit SHAs can't. why a SHA?

Inputs

name	description	required	default
command	Command to run: scan, score, validate, plan, test, or all (default runs scan+validate+score)	no	all
policy-file	Path to policy YAML file (for score/validate/test)	no	policy.yaml
scan-directory	Directory to scan for ungoverned AI calls	no	.
min-score	Minimum governance score to pass (0-100)	no	0
fail-on-ungoverned	Fail if ungoverned AI calls are found	no	false
python-version	Python version to use	no	3.11
old-policy	Path to current policy YAML (for plan command)	no	""
new-policy	Path to proposed policy YAML (for plan command)	no	""
audit-db	Path to SQLite audit database for plan replay	no	""
replay	Path to JSONL file of recorded actions for plan replay	no	""
test-suite	Path to test suite YAML (for test command)	no	""
regression-policy	Path to old policy for test --regression comparison	no	""
comment	Post results as a PR comment (requires github-token)	no	true
github-token	GitHub token for posting PR comments (defaults to GITHUB_TOKEN)	no	""

Outputs

name	description
score	Governance score (0-100)
grade	Governance grade (A+, A, B, C, D, F)
ungoverned-count	Number of ungoverned AI calls found
badge-url	Shields.io badge URL for governance score
plan-newly-blocked	Number of actions newly blocked by policy change (plan command)
test-passed	Number of tests passed (test command)
test-failed	Number of tests failed (test command)
test-total	Total number of tests (test command)