| Internet-Draft | Safe-IOC Sharing | April 2026 |
| Grimminck | Expires 30 October 2026 | [Page] |
This document codifies a consistent and reversible convention used in the threat intelligence and security communities for sharing potentially malicious indicators of compromise (IOCs), such as URLs, IP addresses, email addresses, and domain names. It describes a safe obfuscation format that reduces the risk of accidental execution or activation when IOCs are displayed or transmitted. The recommended form brackets the URI scheme name so that the string is not syntactically a valid URI per generic URI parsers, and extends the same bracket treatment to colons inside IPv6 literals; recognizable nested indicators within the Path, Query, or Fragment of a URI are obfuscated in place, and legacy scheme-substitution tokens are defined for de-obfuscation interoperability. Safe-IOC strings are a textual rendering convention, not URIs, and are not intended to be processed by generic URI parsers. These conventions aim to improve interoperability among tools and feeds that exchange threat intelligence data.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 30 October 2026.¶
Copyright (c) 2026 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document.¶
The secure sharing of malicious artifacts is vital to threat intelligence, open-source intelligence (OSINT), and incident response. However, sharing raw URLs, IP addresses, and email addresses associated with malware or threat actors poses a risk of accidental activation.¶
Participants who routinely share indicators of compromise (IOCs) include security operations center (SOC) analysts, computer security incident response teams (CSIRTs), OSINT researchers, incident responders, and vendors of threat intelligence platforms and feeds. IOCs appear in email threads, instant-messaging channels, ticketing systems, PDF and HTML reports, blog posts, paste sites, machine-readable formats such as STIX [STIX21] / TAXII [TAXII21], and platforms such as MISP [MISP]. Both human readers and automated pipelines consume this material.¶
When a raw URI such as "https://malicious-host.example/path" is embedded in those channels, many systems automatically detect it and render it as a clickable or otherwise actionable link. An analyst may then activate the resource unintentionally: navigating to an attacker-controlled URI can reveal the analyst's IP address and organizational affiliation, trigger delivery of malware, or alert the threat actor that a particular indicator is under active investigation. Some mail and web infrastructure pre-fetches or resolves links for scanning or preview purposes, producing the same exposure without any deliberate user action. PDF viewers and rich-text editors may turn strings that resemble URIs into hyperlinks even when the author intended plain text.¶
A longstanding practice in the security community is to alter IOCs so that they remain human-readable but are not treated as live URIs by typical software: for example, replacing "." with "[.]" and wrapping the scheme name in brackets so that "http" becomes "[http]". Older conventions substituted characters within the scheme name (e.g., "http" to "hxxp"), but many variant spellings emerged (e.g., "h**p", "hXXp"), hindering reliable parsing and automation. This document defines a canonical form that uses uniform square-bracket wrapping for the scheme, dot, at-sign, and (inside IPv6 literals) colon delimiters, and a strict order of operations so that independent implementations can interoperate. Legacy scheme substitutions ("hxxp", "hxxps") are documented for de-obfuscation interoperability but are NOT RECOMMENDED for new output.¶
Safe-IOC strings produced by this specification are a textual rendering convention intended for human consumption and for lossy text channels. They are deliberately not valid URIs per [RFC3986] and are not intended to be passed to generic URI parsers, resolvers, or dereferencing libraries.¶
The canonical "[scheme]" form does not occupy the URI scheme namespace (see Section 4). The legacy tokens "hxxp" and "hxxps" do occupy that syntactic position; this document requests an update to their existing entries in the URI Schemes registry (see Section 12). The security implications are discussed in Section 11. Implementations MUST NOT treat any obfuscated form as a resolvable URI scheme.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
Obfuscating: The process of altering an indicator so that it cannot be accidentally activated or clicked. The goal is to prevent automatic execution or resolution, not to conceal the content from human readers; the original indicator remains visually recognizable.¶
De-obfuscating: The process of restoring an obfuscated indicator to its original, actionable form.¶
IOC: Indicator of Compromise - data such as a URL, IP address, domain name, email address, or hash associated with malicious activity.¶
IP-literal: The bracketed form of an IPv6 address as it appears in a URI authority component (e.g., "[2001:db8::1]"), per Section 3.2.2 of [RFC3986].¶
Nested Indicator: A URI, email address, or IP address literal that appears inside the Path, Query, or Fragment of another URI (for example, an open-redirect target carried as a query parameter, or an email address embedded in a redirect URL).¶
Inconsistent obfuscation practices hinder the reliable and automated exchange of threat intelligence. For example:¶
Such inconsistencies reduce the effectiveness of threat detection and response.¶
To prevent double-bracketing (e.g., "[[https]]://example[[.]]example") when a tool processes the same string twice or in the wrong order, implementations MUST apply transformations in the following strict order of operations. Implementations MUST treat already-obfuscated substrings (the tokens "[scheme]", "[.]", "[@]", and "[:]") as opaque and MUST NOT apply transformations to them again; thus, the transformation is idempotent. Implementations SHOULD NOT use percent-encoded characters (such as %2e for ".") in obfuscated output to prevent ambiguity. Percent-encoded delimiters in the Host or Userinfo of an input URI MAY be decoded before applying Steps 2 and 3; percent-encoded content inside a Path, Query, or Fragment MUST NOT be decoded during obfuscation, so that independent implementations produce the same output for the same input.¶
Identify the URI scheme and wrap it in square brackets. For example, "http" becomes "[http]", "https" becomes "[https]", "ftp" becomes "[ftp]". This is the RECOMMENDED canonical form. Only the scheme name is wrapped; the scheme delimiter ("://" for hierarchical URIs, ":" for schemes such as mailto: or tel:) is preserved unchanged. Scheme names SHOULD be normalized to lowercase before wrapping, consistent with Section 3.1 of [RFC3986]. The leading "[" character is not permitted in a URI scheme name (Section 3.1 of [RFC3986]), so the result cannot be mistaken for a valid URI by compliant parsers. Because the transformation is purely syntactic wrapping, it extends to any current or future scheme without requiring a per-scheme mapping table.¶
The legacy tokens "hxxp" and "hxxps" are in widespread operational use as obfuscated forms of "http" and "https" respectively. Implementations encountering these tokens in existing data SHOULD recognize them as obfuscated indicators during de-obfuscation (see Section 6). These legacy forms are NOT RECOMMENDED for new output. No other legacy substitutions (e.g., "h**p", "fxxps") are defined by this specification.¶
Identify the "@" symbol in the userinfo subcomponent (per [RFC3986]) and replace it with "[@]". This applies to email addresses and URIs containing userinfo (e.g., "username:password@host").¶
Replace all "." (period) characters in the Host subcomponent with "[.]". This applies to domain names and IPv4 addresses, including standalone values (e.g., "evil.example" or "198.51.100.1" without a scheme). IPv4 addresses use dotted-decimal notation that overlaps with domain name syntax, so the "[.]" substitution applies to them naturally. Inside an IPv6 literal, replace every ":" (colon) with "[:]"; if the literal contains an embedded IPv4 address (e.g., "::ffff:192.0.2.1" per Section 2.5.5 of [RFC4291]), each "." in that embedded IPv4 MUST also be replaced with "[.]". The outer square brackets that surround an IP-literal in URI syntax (Section 3.2.2 of [RFC3986]) are part of the URI grammar, not part of the obfuscation, and MUST be preserved verbatim. Port numbers following the host (e.g., ":8080" or the port following an IP-literal) are not part of the Host subcomponent and MUST NOT be altered.¶
Zone identifiers attached to IPv6 addresses (the "%25<zone-id>" suffix inside a URI IP-literal per [RFC9844], or the "%<zone-id>" suffix in bare form) are preserved verbatim.¶
Bare IPv6 literals without surrounding URI brackets (e.g., "2001:db8::1" appearing on its own line in a report) MUST receive the same colon-bracketing treatment. Implementations SHOULD recognize bare IPv6 by the presence of either a "::" shorthand or the fully expanded eight-group colon-hexadecimal form so that strings such as "host:port" (a single colon) are not misclassified.¶
For schemes whose content after the scheme delimiter is not a hierarchical authority (e.g., "mailto:" per [RFC6068], or "urn:"), implementations apply Step 2 to every "@" and Step 3 to every "." within the scheme body, up to the first "?" or "#" delimiter. This ensures that mail addresses and dotted labels carried in opaque URI bodies are obfuscated by the same rules that apply to the corresponding bare indicators.¶
When a scheme is present, host subcomponent boundaries are determined by the URI syntax defined in [RFC3986]. For bare indicators without a scheme (e.g., a standalone domain or IP address), host identification is implementation-dependent; implementations SHOULD apply reasonable heuristics such as recognizing dotted-decimal IPv4 addresses, colon-hexadecimal IPv6 addresses, and domain-name patterns.¶
Implementations MUST NOT apply Steps 2 and 3 wholesale to the Path, Query, or Fragment of the primary URI; doing so would alter semantics (for example, by bracketing dots in file extensions or query values) and is not reversible without ambiguity.¶
If a Path, Query, or Fragment contains a recognizable nested indicator, implementations SHOULD obfuscate that nested indicator in place by recursively applying Steps 1 through 3 to the matched span. Nested-indicator detection operates on the literal (undecoded) text of the enclosing component; percent-encoded nested indicators are not matched by this step, and implementations that wish to surface them SHOULD emit them as separate obfuscated indicators. For the purposes of this step, a nested indicator is one of the following:¶
Only spans matching one of the grammars above are obfuscated; dots, "@", and ":" characters that appear elsewhere in the Path, Query, or Fragment MUST be preserved verbatim.¶
Implementations MAY instead extract each nested indicator and emit it as a separate obfuscated indicator alongside the primary one.¶
The following uses Augmented BNF (ABNF) per [RFC5234] to define the tokens used in obfuscated IOC strings. The scheme production matches the syntax in Section 3.1 of [RFC3986]. An implementation MAY use this grammar to validate whether a string is already obfuscated or still requires processing.¶
safe-scheme = "[" scheme "]"
scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
legacy-scheme = "hxxp" / "hxxps"
safe-dot = "[" "." "]"
safe-at = "[" "@" "]"
safe-colon = "[" ":" "]"
obf-label = 1*( ALPHA / DIGIT / "-" )
obf-host = obf-label 1*( safe-dot obf-label )
obf-local = 1*( ALPHA / DIGIT / "." / "_" / "%"
/ "+" / "-" )
obf-email = obf-local safe-at obf-host
The obf-label production permits any character allowed in a DNS label; applications SHOULD additionally enforce label format rules (no leading or trailing hyphen, 63-octet maximum) per Section 3.5 of [RFC1035]. The composition of a full URI (scheme, authority, path, query, fragment) and of an obfuscated IPv6 literal is not restated here; both are defined algorithmically in Section 4 and rely on the URI grammar of [RFC3986] and the address grammar of [RFC4291].¶
A compliant implementation MUST recognize a string as obfuscated when it contains any of the tokens safe-scheme, legacy-scheme, safe-dot, safe-at, or safe-colon in the roles described in this document.¶
Tools designed to ingest obfuscated data SHOULD automatically reverse these transformations in a deterministic manner:¶
For legacy tokens, longer strings MUST be reversed before shorter prefixes that are substrings of them (e.g., reverse "hxxps" before "hxxp"). De-obfuscation MUST maintain the original semantics of the data to avoid misinterpretation.¶
Safe-IOC strings may contain two kinds of square brackets: the URI IP-literal brackets (Section 3.2.2 of [RFC3986]) and the obfuscation tokens "[scheme]", "[.]", "[@]", and "[:]". When an IPv6 URI is obfuscated, adjacent "[" or "]" characters may appear (e.g., "[http]://[[:][:]1]"). These are adjacent tokens, not nested brackets. Parsers SHOULD tokenize left-to-right, attempting to match the obfuscation tokens first and treating any remaining "[" or "]" as a URI IP-literal delimiter.¶
De-obfuscation MUST only be performed when the output is written to a non-executable buffer (e.g., a variable, string, or file) that cannot be automatically interpreted, executed, or rendered as a clickable link by the system or application. The tool MUST NOT de-obfuscate a string if it is currently being rendered in a "live" environment (e.g., a web browser preview, an active document viewer, or any context where the resulting string could be automatically executed, resolved, or displayed as a clickable link).¶
De-obfuscation SHOULD only occur in controlled contexts such as:¶
Accidental activation during the de-obfuscation process poses a security risk and MUST be prevented.¶
Common scenarios include:¶
Safe-IOC strings are intended for human-readable contexts (reports, email, tickets, chat) and for free-text fields in structured formats such as STIX 2.1 [STIX21]; they are not a substitute for the typed observable and indicator objects those formats provide.¶
Software designed to parse threat intelligence feeds should explicitly support these obfuscation and de-obfuscation conventions. Implementations SHOULD verify correct behavior through unit tests and validation scripts using the test vectors in Section 10.¶
Internationalized Domain Names (IDNs): Apply Step 3 to the IDN as presented, in either A-label (ASCII-Compatible Encoding, "xn--" prefixed) or U-label (native Unicode) form per [RFC5890]; for example, "xn--n3h.example" becomes "xn--n3h[.]example".¶
Non-Standard URI Schemes: The "[scheme]" canonical form extends to any URI scheme. For example, "smb://fileserver.example/share" becomes "[smb]://fileserver[.]example/share".¶
IPv6 Literals in URIs: Bracket every ":" inside the IP-literal as "[:]" and every "." in any embedded IPv4 portion as "[.]". Preserve the outer URI brackets, any "%25<zone-id>" suffix, and any following port. For example, "http://[2001:db8::1]:8080" becomes "[http]://[2001[:]db8[:][:]1]:8080", and "[::ffff:192.0.2.1]" becomes "[[:][:]ffff[:]192[.]0[.]2[.]1]". The "[[" and "]]" sequences that appear when an IP-literal begins or ends with "::" are adjacent tokens, not nested brackets.¶
Bare IPv6 Literals: Addresses appearing without URI brackets (e.g., "2001:db8::1" on its own line) are obfuscated by colon-bracketing alone: "2001[:]db8[:][:]1". Implementations SHOULD restrict bare-IPv6 detection to strings that contain "::" or the fully expanded eight-group form so that "host:port" and similar single-colon constructs are not misclassified.¶
The following provides a "golden set" of inputs and expected outputs. Domain names use the "example" reserved space per [RFC2606]; IPv4 addresses use documentation ranges per [RFC5737]; IPv6 addresses use the documentation prefix per [RFC3849]. Implementations SHOULD use these vectors to ensure correct behavior and to avoid under-obfuscation (e.g., missing email addresses or unaltered IPv6 colons) or over-obfuscation (e.g., bracketing a port colon).¶
Note: The IPv6 rows show colon bracketing inside the IP-literal and "[.]" bracketing of dots in any embedded IPv4 portion; outer URI brackets, "%25<zone-id>" suffixes, and port separators are preserved. The nested-indicator rows illustrate Step 4 in-place obfuscation. The Punycode row shows that IDN labels in punycode form receive the same "[.]" treatment as ASCII labels. The idempotency rows confirm that applying the transformation to an already-obfuscated string produces no change. The legacy de-obfuscation row confirms backward compatibility with the older "hxxp" convention.¶
While these obfuscation techniques reduce the risk of accidental activation, systems processing obfuscated indicators SHOULD apply the safeguards described in this section.¶
The canonical "[scheme]" form does not occupy the URI scheme namespace (see Section 4). Software that encounters bracketed schemes MUST NOT attempt to resolve or dereference them.¶
The legacy tokens "hxxp" and "hxxps" do occupy the syntactic position of a URI scheme. The "Uniform Resource Identifier (URI) Schemes" registry per [RFC7595] contains existing Provisional entries for both names; this document requests that those entries be updated to cite this document and to change status to "Permanent" (see Section 12). Implementations should consider surrounding context when disambiguating these tokens.¶
When a scheme is present, a compliant tool MUST obfuscate both the scheme (Step 1) and the host-level delimiters (Steps 2-3). Partial obfuscation (for example, replacing only "." with "[.]" while leaving "https" unchanged) creates a false sense of security: the scheme remains active and could still trigger automatic linkification or execution. Implementations MUST NOT produce partially obfuscated output when full obfuscation is intended. For bare indicators without a scheme (e.g., a standalone IP address or email address), host-level and userinfo obfuscation (Steps 2-3) alone constitutes valid Safe-IOC output.¶
Implementations that parse Safe-IOC strings may become confused by malformed or inconsistently obfuscated input. For example, "[https]://example.example" (scheme obfuscated but dots not) or "https://example[.]example" (dots obfuscated but scheme not) are not canonical Safe-IOC output and SHOULD be treated as partially obfuscated: such strings retain activation risk and implementations SHOULD either complete the obfuscation before display or flag the input as non-conforming. Parsers SHOULD validate that obfuscated strings conform to the canonical transformation rule and the ABNF before de-obfuscation. Rejecting or flagging ambiguous input reduces the risk of misinterpretation.¶
As stated in Section 6.1, de-obfuscation MUST only occur when the result is placed in a non-executable buffer. A non-executable buffer is one that cannot be automatically interpreted by the system (e.g., as a URI to fetch, a command to run, or a link to display). Writing de-obfuscated output into a live document, rich-text editor, or browser address bar before explicit user action creates an unacceptable risk of accidental activation.¶
The "Uniform Resource Identifier (URI) Schemes" registry per [RFC7595] contains Provisional entries for "hxxp" and "hxxps", originally registered with reference to draft-salgado-hxxp-01. IANA is requested to update both entries to cite this document as the reference and to change the status of both entries from "Provisional" to "Permanent".¶
Scheme name: hxxp
Status: Permanent
Reference: This document
Description: Legacy obfuscated form of "http" used in threat-intelligence sharing. Not to be dereferenced as a URI scheme. NOT RECOMMENDED for new output (see Section 4).¶
Scheme name: hxxps
Status: Permanent
Reference: This document
Description: Legacy obfuscated form of "https" used in threat-intelligence sharing. Not to be dereferenced as a URI scheme. NOT RECOMMENDED for new output (see Section 4).¶