iam-deceptive-escalation-auditor

Name: iam-deceptive-escalation-auditor
Author: anyshift-io

16stars0forks

How to install iam-deceptive-escalation-auditor?

anyshift-io/sre-skills/iam-deceptive-escalation-auditor

$npx -y skills add anyshift-io/sre-skills --skill iam-deceptive-escalation-auditor

Installs into the current project.

Summary

Audit the union of every IAM policy attached to one principal for privilege-escalation paths that no single statement reveals, and for apparent escalations that are already neutralised. Resolves the effective permission set across all attached policies (Allow minus blanket Deny),

Files · 1

View on GitHub

SKILL.md

1	# iam-deceptive-escalation-auditor
2
3	Privilege-escalation audit skill for one AWS IAM principal. Takes every permissions policy
4	attached to a role or user (plus the trust policy and permissions boundary if supplied),
5	resolves the effective permission set across all of them, and answers one question a
6	per-statement read cannot: can this principal escalate to a privilege it was not granted,
7	and is an apparent escalation real or already neutralised. It returns findings with severity
8	and a fix, then names exactly where a single principal's policy documents stop being able to
9	answer the question.
10
11	The escalation combinations this skill exists to catch are precisely the ones that **span
12	two statements or two attached policies**, so that no single statement looks guilty on its
13	own. `iam:PassRole` in one policy and `sagemaker:CreateTrainingJob` in another are each
14	routine; together they let the principal launch compute with any role attached and inherit
15	it. A per-statement read clears every statement and misses the union. The other half of the
16	skill is the inverse discipline: an explicit `Deny`, a resource scope, a broken trust, or an
17	unsatisfiable `Condition` can neutralise an escalation that still reads as critical, and
18	the audit must not fabricate a finding the effective permissions do not support.
19
20	## When to invoke
21
22	- An agent is asked to audit an IAM role or user for privilege escalation, over-broad
23	grants, or "can this principal become administrator."
24	- A policy is being shipped or reviewed and the question is whether two individually-fine
25	grants combine into an escalation.
26	- A policy looks dangerous (a full mutation kit, a cross-account assume, an `Action '*'`)
27	and the claim "but it's capped / scoped / denied" needs to be confirmed against the
28	effective permissions, not taken on trust.
29	- An incident assumes a principal is compromised and the question is what it can escalate to.
30
31	## What this skill reads, and what it does not
32
33	It reads the static policy documents attached to one principal: every permissions policy,
34	plus the trust policy (`AssumeRolePolicyDocument`) and the permissions boundary if supplied.
35	That is the entire input. The audit is correct and complete *for the effective permissions
36	those documents express*, and it is explicit about the rest. Every audit ends by naming the
37	joins it cannot make:
38
39	- It does not see the principal's other attached policies if only some were supplied.
40	Effective permissions are the union of every managed and inline policy. Join: principal to
41	its full set of attached policies.
42	- It does not know the permissions boundary unless one is supplied. A boundary caps
43	what any Allow can actually grant. Join: principal to its permissions boundary.
44	- It does not see org SCPs. A Service Control Policy can Deny actions this policy
45	Allows and is invisible from the account. Join: account to its organization's SCPs.
46	- It does not contain the privileges of a targeted role. An escalation that passes,
47	assumes, or hijacks a role only matters if that role is more privileged than this
48	principal, and those privileges live in other documents. Join: this policy to the roles
49	and resources it references.
50
51	A clean (neutralised) policy still gets a boundary section, because a capped policy is not a
52	proven-safe principal.
53
54	## The model
55
56	Build the effective permission set across all attached policies. An action is granted
57	when some Allow statement matches it (by case-insensitive glob on `Action`, or by
58	`NotAction`) and no blanket `Deny` (on `Resource "*"`) matches it. Deny w

People who liked this also viewed

6 picks

Type

iam-deceptive-escalation-auditor

16stars0forks

View on GitHub

How to install iam-deceptive-escalation-auditor?

anyshift-io/sre-skills/iam-deceptive-escalation-auditor

$npx -y skills add anyshift-io/sre-skills --skill iam-deceptive-escalation-auditor

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# iam-deceptive-escalation-auditor
2
3	Privilege-escalation audit skill for one AWS IAM principal. Takes every permissions policy
4	attached to a role or user (plus the trust policy and permissions boundary if supplied),
5	resolves the effective permission set across all of them, and answers one question a
6	per-statement read cannot: can this principal escalate to a privilege it was not granted,
7	and is an apparent escalation real or already neutralised. It returns findings with severity
8	and a fix, then names exactly where a single principal's policy documents stop being able to
9	answer the question.
10
11	The escalation combinations this skill exists to catch are precisely the ones that **span
12	two statements or two attached policies**, so that no single statement looks guilty on its
13	own. `iam:PassRole` in one policy and `sagemaker:CreateTrainingJob` in another are each
14	routine; together they let the principal launch compute with any role attached and inherit
15	it. A per-statement read clears every statement and misses the union. The other half of the
16	skill is the inverse discipline: an explicit `Deny`, a resource scope, a broken trust, or an
17	unsatisfiable `Condition` can neutralise an escalation that still reads as critical, and
18	the audit must not fabricate a finding the effective permissions do not support.
19
20	## When to invoke
21
22	- An agent is asked to audit an IAM role or user for privilege escalation, over-broad
23	grants, or "can this principal become administrator."
24	- A policy is being shipped or reviewed and the question is whether two individually-fine
25	grants combine into an escalation.
26	- A policy looks dangerous (a full mutation kit, a cross-account assume, an `Action '*'`)
27	and the claim "but it's capped / scoped / denied" needs to be confirmed against the
28	effective permissions, not taken on trust.
29	- An incident assumes a principal is compromised and the question is what it can escalate to.
30
31	## What this skill reads, and what it does not
32
33	It reads the static policy documents attached to one principal: every permissions policy,
34	plus the trust policy (`AssumeRolePolicyDocument`) and the permissions boundary if supplied.
35	That is the entire input. The audit is correct and complete *for the effective permissions
36	those documents express*, and it is explicit about the rest. Every audit ends by naming the
37	joins it cannot make:
38
39	- It does not see the principal's other attached policies if only some were supplied.
40	Effective permissions are the union of every managed and inline policy. Join: principal to
41	its full set of attached policies.
42	- It does not know the permissions boundary unless one is supplied. A boundary caps
43	what any Allow can actually grant. Join: principal to its permissions boundary.
44	- It does not see org SCPs. A Service Control Policy can Deny actions this policy
45	Allows and is invisible from the account. Join: account to its organization's SCPs.
46	- It does not contain the privileges of a targeted role. An escalation that passes,
47	assumes, or hijacks a role only matters if that role is more privileged than this
48	principal, and those privileges live in other documents. Join: this policy to the roles
49	and resources it references.
50
51	A clean (neutralised) policy still gets a boundary section, because a capped policy is not a
52	proven-safe principal.
53
54	## The model
55
56	Build the effective permission set across all attached policies. An action is granted
57	when some Allow statement matches it (by case-insensitive glob on `Action`, or by
58	`NotAction`) and no blanket `Deny` (on `Resource "*"`) matches it. Deny w

People who liked this also viewed

6 picks

Type