Skip to content

    Type a page, a command such as keryn scan, or a rule id.

    github-actions-ai-supply-chain-v2

    View as Markdown

    10 rules · Keryn Software Licence · rules 0.8.0 · all packs

    How to read a rule
    Severity
    How bad the finding is if it is real: error fails a default gate, warning and info do not.
    Confidence
    P0 exploitable as written; P1 dangerous in context; P2 a heuristic or a posture check: confirm by reading the code.
    False-positive risk
    low: an exact API or flag, so hits are almost always real; medium: the pattern is legitimate in many programs, so check each hit; high: it finds a lead to confirm, so expect benign hits.
    Edge
    The relationship the finding sits on. reaches: content or traffic gets to a thing; calls: one thing invokes another; reads: one thing reads from another; writes: one thing writes to another; sends: data leaves for another party; grants: authority is given or left unchecked; depends_on: something relies on a component it does not verify; trusts: one party believes another without evidence. The boundary it crosses, when there is one: user->prompt (untrusted input reaching a prompt); model->code (model output reaching code, a shell, a query or a page); secret->outside (a secret leaving the trusted side); supply-chain (code, models or prompts pulled in from outside).
    Taxonomy chips
    References for readers who work in OWASP LLM 2025, OWASP Agentic, CWE or NIST AI RMF. They say where a rule belongs, not that a framework is satisfied.

    seceng.github.actions-unpinned-third-party-action

    Section titled “seceng.github.actions-unpinned-third-party-action”

    warning generic github-actions

    GitHub Actions workflow uses a third-party action without pinning to a full commit SHA. This can introduce supply-chain risk in AI/ML build pipelines.

    Confidence P1: dangerous in context. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: depends_on something relies on a component it does not verify; it crosses the supply-chain boundary (code, models or prompts pulled in from outside).

    CWE-829 CWE-494 LLM03 ASI04

    seceng.github.pull-request-target-checkout-risk

    Section titled “seceng.github.pull-request-target-checkout-risk”

    warning generic github-actions

    GitHub workflow uses pull_request_target with checkout or script execution. This can expose privileged tokens to untrusted PR code.

    Confidence P2: a heuristic or a posture check: confirm by reading the code. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: grants authority is given or left unchecked.

    CWE-829 LLM03 LLM06 ASI03 ASI04 ASI02

    warning javascript, typescript, python github-actions

    GitHub token or secret appears to flow into LLM prompt, request body, or logs.

    Confidence P2: a heuristic or a posture check: confirm by reading the code. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: reads one thing reads from another.

    CWE-532 CWE-798 LLM02 LLM03 ASI03

    seceng.github-workflow-dispatch-input-to-shell-or-ai

    Section titled “seceng.github-workflow-dispatch-input-to-shell-or-ai”

    warning generic github-actions

    workflow_dispatch input is passed to shell or AI prompt without visible validation.

    Confidence P2: a heuristic or a posture check: confirm by reading the code. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: calls one thing invokes another; it crosses the model->code boundary (model output reaching code, a shell, a query or a page).

    CWE-20 CWE-78 LLM01 LLM06 ASI05 ASI01 ASI02

    seceng.github-ai-pr-bot-commits-without-approval

    Section titled “seceng.github-ai-pr-bot-commits-without-approval”

    info javascript, typescript, python github-actions

    AI/LLM PR bot appears to commit or push generated changes without visible approval gate.

    Confidence P1: dangerous in context. False-positive risk high: it finds a lead to confirm, so expect benign hits.

    Edge inspected: grants authority is given or left unchecked.

    CWE-862 CWE-20 LLM06 LLM09 ASI02 ASI09

    seceng.github-actions-ai-model-download-no-integrity

    Section titled “seceng.github-actions-ai-model-download-no-integrity”

    info generic github-actions

    GitHub workflow downloads model, weights, or AI dependency without visible checksum or signature verification.

    Confidence P1: dangerous in context. False-positive risk high: it finds a lead to confirm, so expect benign hits.

    Edge inspected: depends_on something relies on a component it does not verify; it crosses the supply-chain boundary (code, models or prompts pulled in from outside).

    CWE-494 LLM03 ASI04

    seceng.github-actions-secrets-to-issue-comment-llm

    Section titled “seceng.github-actions-secrets-to-issue-comment-llm”

    warning generic github-actions

    Workflow appears to send issue/PR comment content or secrets to an LLM. Comments are attacker-controlled and may contain prompt injection or secret bait.

    Confidence P1: dangerous in context. False-positive risk high: it finds a lead to confirm, so expect benign hits.

    Edge inspected: reaches content or traffic gets to a thing; it crosses the user->prompt boundary (untrusted input reaching a prompt).

    CWE-20 CWE-200 LLM01 LLM02 ASI01 ASI03

    seceng.github-actions-excessive-token-permissions-ai-job

    Section titled “seceng.github-actions-excessive-token-permissions-ai-job”

    warning generic github-actions

    AI/agent workflow grants broad GitHub token permissions. Restrict contents, issues, pull-requests, actions, and id-token scopes.

    Confidence P1: dangerous in context. False-positive risk high: it finds a lead to confirm, so expect benign hits.

    Edge inspected: grants authority is given or left unchecked.

    CWE-266 LLM06 LLM03 ASI03 ASI02 ASI04

    info generic github-actions

    This job runs on a self-hosted runner, or on a runner named by a configuration variable (vars.*) that may resolve to one. Record it in the build inventory: such runners keep state and network access between jobs, and a variable can move the job onto them without a reviewed change to this file. Keep them off untrusted triggers and use ephemeral (just-in-time) runners where you can.

    Confidence P1: dangerous in context.

    Edge inspected: depends_on something relies on a component it does not verify; it crosses the supply-chain boundary (code, models or prompts pulled in from outside).

    CWE-829 LLM03

    References: https://docs.github.com/en/actions/reference/security/secure-use#hardening-for-self-hosted-runners · https://docs.github.com/en/actions/how-tos/write-workflows/choose-where-workflows-run/choose-the-runner-for-a-job

    seceng.github.actions-self-hosted-runner-untrusted-trigger

    Section titled “seceng.github.actions-self-hosted-runner-untrusted-trigger”

    warning generic github-actions

    A workflow triggered by pull_request or pull_request_target runs this job on a self-hosted runner (or on a runner a vars.* variable picks). Pull-request code, including code from forks, then executes on a machine that persists between jobs and can reach your network. Run pull-request jobs on GitHub-hosted or ephemeral, isolated runners, or restrict the trigger.

    Confidence P2: a heuristic or a posture check: confirm by reading the code. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: depends_on something relies on a component it does not verify.

    CWE-829 LLM03 ASI04

    Fix: Move pull-request jobs to GitHub-hosted runners, or to ephemeral just-in-time runners in an isolated network; require approval for workflows from outside collaborators; keep self-hosted runners for push, schedule and workflow_dispatch on protected branches.

    References: https://docs.github.com/en/actions/reference/security/secure-use#hardening-for-self-hosted-runners · https://docs.github.com/en/actions/concepts/runners/self-hosted-runners#self-hosted-runner-security