m365-mailbox-compromise

Microsoft 365 mailbox compromise chain — OAuth consent phishing, delegate access abuse, mail rule persistence, and token theft via device code phishing. Full kill chain from initial access to persistent email collection.

Quality

66%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Fix and improve this skill with Tessl

tessl review fix ./packages/decepticon/decepticon/skills/standard/cloud/m365-mailbox-compromise/SKILL.md

Quality

Content

65%Weight 40%Scale 1-3

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The content is highly actionable with executable commands across the full kill chain and clear phase sequencing. It is held back by redundancy, missing validation checkpoints for batch/destructive operations, and a monolithic structure with no progressive disclosure.

Suggestions

Add explicit validation checkpoints before destructive/batch steps (e.g., 'Verify the forwarding rule was created via GET .../messageRules before proceeding; confirm exfil target reachability before bulk download').

Split large reference material out of SKILL.md into bundle files (e.g., references/detection-signatures.md, references/tools.md) and link to them so the overview stays lean.

De-duplicate the token-refresh logic (present in both §2 Step 4 and §6) and consider folding the Decision Gate into the phase sections rather than restating them as a separate block.

Dimension	Reasoning	Score
Conciseness	The body is mostly lean and code-dense without explaining concepts Claude already knows, but it carries redundant material — the intro echoes the description, the 'Decision Gate' re-states the section logic as IF-blocks, and token refresh appears in both §2 and §6 — that could be tightened.	2 / 3
Actionability	Provides fully executable curl commands against real Graph/EWS/login endpoints with a real first-party client_id, jq pipelines, and a polling loop; commented PowerShell cmdlets are explicitly justified by admin-context constraints rather than being pseudocode, matching the copy-paste-ready anchor.	3 / 3
Workflow Clarity	Phases are clearly numbered (§1–§6) with a Quick Reference and Decision Gate for branching, but there are no explicit verify-then-proceed checkpoints for the batch and destructive operations (bulk mail/attachment download loops, permanentDelete of security alerts), which the rubric caps at 2.	2 / 3
Progressive Disclosure	Sections are well-organized, but the ~300-line skill is monolithic with no bundle files and no split references, so content that could be separate (Detection Signatures, Tools & Resources, the full EWS XML) is inline; the under-50-line exception does not apply here.	2 / 3
	Total	9 / 12 Passed

Description

67%Weight 40%Scale 1-3

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific and distinctive, clearly conveying a concrete M365 mailbox-compromise kill chain. It loses points for lacking an explicit 'Use when...' trigger and for missing common term variations (M365, Office 365, Outlook).

Suggestions

Add an explicit 'Use when...' clause, e.g. 'Use when simulating or responding to M365/Office 365 mailbox compromise, OAuth consent phishing, or device code phishing incidents.'

Include natural user-facing variations such as 'M365', 'Office 365', 'Outlook', and 'Exchange Online' alongside 'Microsoft 365' to broaden trigger coverage.

Move some of the keyword breadth currently only present in metadata.when_to_use into the description itself so triggers are surfaced at the top level.

Dimension	Reasoning	Score
Specificity	Quotes 'OAuth consent phishing, delegate access abuse, mail rule persistence, and token theft via device code phishing' — four concrete, specific techniques plus 'Full kill chain from initial access to persistent email collection', matching the multiple-specific-actions anchor rather than the domain-and-some-actions level below.	3 / 3
Completeness	The WHAT is comprehensive, but there is no 'Use when...' clause or equivalent explicit trigger guidance; per the rubric a missing explicit trigger caps completeness at 2 even though the capability description is strong.	2 / 3
Trigger Term Quality	Includes relevant terms like 'Microsoft 365 mailbox compromise', 'OAuth consent phishing', and 'device code phishing', but omits common variations a user would naturally say (M365, Office 365, Outlook, Exchange), so it does not reach full natural-term coverage.	2 / 3
Distinctiveness Conflict Risk	'Microsoft 365 mailbox compromise chain' tied to specific OAuth/device-code/delegate techniques is a clear niche with distinct triggers unlikely to conflict with other skills, rather than a generic overlapping description.	3 / 3
	Total	10 / 12 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning

	Total	15 / 16 Passed

Repository: PurpleAILAB/Decepticon
Path: packages/decepticon/decepticon/skills/standard/cloud/m365-mailbox-compromise/SKILL.md
Commit: e34afba

Reviewed: 6 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.