Find hidden Unicode (zero-width, bidi, tag payloads), homoglyphs and lookalike names in text.
Scored across 3 tools
The three tools have distinct actions: clean_text mutates, scan_text reports findings, compare_names compares against a protected list. clean_text and scan_text target the same character domain and share terminology, but the descriptions clearly separate remediation from detection, leaving little real ambiguity.
All three tools follow a clean verb_noun snake_case pattern (clean_text, compare_names, scan_text) with no mixed conventions or vague verbs.
Three tools is well within the appropriate range for a focused Unicode-safety utility, and each tool covers a distinct operation (detect, remediate, compare) without redundancy.
The surface covers the core lifecycle: scanning for hidden characters, cleaning them, and comparing confusables against protected names. Minor gaps exist, such as no batch/multi-input endpoints or a standalone normalisation/skeleton preview, but agents can work around these.
Return the text with invisible characters, bidi controls, tag characters and control characters removed and unusual spaces turned into normal spaces. By default ZWJ/ZWNJ/VS16 are kept inside emoji sequences, between letters of joining scripts (Arabic, Persian) and after an Indic virama (keep_joiners=true); strict=true removes them everywhere. Optional: map lookalike letters to Latin (lookalikes) and apply NFKC normalisation (nfkc). Read-only. Same engine as GET /api/clean.
| Name | Required | Description | Default |
|---|---|---|---|
| nfkc | No | Apply Unicode NFKC first (fullwidth and math letters → plain). | |
| text | Yes | Text to inspect (up to 2000 UTF-16 code units). | |
| spaces | No | Turn unusual spaces into a normal space. | |
| strict | No | Strict mode: remove every joiner and variation selector, even in emoji and Persian/Indic text (overrides keep_joiners). | |
| lookalikes | No | Map lookalike letters (e.g. Cyrillic а) to Latin. Changes legitimate non-Latin text too. | |
| keep_joiners | No | Keep ZWJ/ZWNJ/VS16 where they are legitimate: emoji ZWJ sequences and VS16 emoji presentation, between letters of joining scripts (Arabic, Persian), and after an Indic virama. Joiners elsewhere are always removed. |
| Name | Required | Description |
|---|---|---|
| text | Yes | |
| notes | No | |
| length | No | |
| changed | Yes | |
| options | No | |
| removed | Yes | |
| version | Yes | |
| joinersKept | No | |
| lookalikesMapped | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint, idempotentHint and destructiveHint=false, so the description's 'Read-only' line is redundant, but it adds real behavioural detail beyond annotations: exactly what is stripped, that ZWJ/ZWNJ/VS16 survive in emoji, joining scripts and after a virama by default, and that lookalikes will rewrite legitimate non-Latin text. That last point is a genuinely useful side-effect warning.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Three front-loaded sentences that pack the default joiner policy first and the optional flags second; almost every clause carries information. The 'Read-only' clause restates an annotation and is the one piece of waste.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With an output schema present, the description need not explain return values, and it adequately covers the default vs. strict joiner policy, the two opt-in transforms, and the risk of lookalike mapping. The only gap is guidance on choosing this tool over scan_text, which is a usage rather than completeness issue.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so all six parameters are already documented, including the strict-overrides-keep_joiners interaction. The description restates the lookalikes and nfkc options without adding syntax or format detail beyond the schema, so the baseline 3 applies.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with a precise verb + resource: return the text with invisible characters, bidi controls, tag characters and control characters removed, plus space normalisation. That is far more specific than the name 'clean_text' alone. It does not, however, say how it differs from the sibling scan_text (detect vs. clean) or compare_names.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Usage is implied by contrast with the siblings' names and by the option set (sanitise text before further processing), but there is no explicit when-to-use / when-not-to-use statement and no named alternative. 'Same engine as GET /api/clean' helps an agent that knows the HTTP surface, not one that only sees the tool list.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Check whether a name, username, domain label or brand looks the same as any of up to 50 protected names (for example "pаypal" with a Cyrillic а vs "paypal"), using a UTS #39-style confusable skeleton (curated subset, case and accents ignored). Read-only. Same engine as GET /api/compare.
| Name | Required | Description | Default |
|---|---|---|---|
| name | Yes | Name, username, domain label or brand to test. | |
| targets | Yes | 1 to 50 protected names. |
| Name | Required | Description |
|---|---|---|
| name | No | |
| notes | No | |
| results | Yes | |
| version | Yes | |
| nameRisk | No | |
| skeleton | Yes | |
| nameScripts | No | |
| confusableWith | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnly/idempotent/non-destructive, so 'Read-only' is redundant, but the description adds genuinely useful behavior: the UTS #39-style confusable skeleton, that it is a curated subset, and that case and accents are ignored. This tells the agent what counts as a match and where recall may be limited.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Core purpose is front-loaded in the first clause, with caveats parenthesized. The parenthetical example and algorithm notes are dense but each earns its place; 'Read-only' duplicates the annotation slightly.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
An output schema exists so return values need no explanation, and annotations cover the safety profile. The description supplies matching semantics and limits; the only gap is sibling differentiation against clean_text/scan_text.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% so baseline is 3, but the description adds semantic detail beyond the schema: the 50-target cap is surfaced and the normalization rules (case/accent-insensitive confusable skeleton) clarify what values 'name' really compares on.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a concrete verb ('check whether... looks the same') with the resource (name/username/domain label/brand) and the comparison target (up to 50 protected names). The homoglyph example ('pаypal' with Cyrillic а) makes the operation immediately distinguishable from the sibling text tools.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The impersonation-check framing implies the use case, but there is no explicit when-to-use/when-not guidance and no routing to clean_text or scan_text. The 'Same engine as GET /api/compare' note is provenance, not selection guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Find invisible characters, bidi controls (Trojan Source), hidden Unicode tag payloads (decoded), variation-selector runs, control characters, unusual spaces, mixed-script words and lookalike letters. ZWJ/ZWNJ/VS16 in legitimate context (emoji sequences, Persian/Arabic joining, Indic virama) are reported as info joiner_in_context and do not raise risk; elsewhere they stay medium. Lone surrogates are flagged. Returns a risk level (none/low/medium/high), counts, and findings with UTF-16 offsets and code points. Read-only. Same engine as GET /api/scan.
| Name | Required | Description | Default |
|---|---|---|---|
| text | Yes | Text to inspect (up to 2000 UTF-16 code units). |
| Name | Required | Description |
|---|---|---|
| risk | Yes | |
| notes | No | |
| total | Yes | |
| counts | Yes | |
| length | No | |
| version | Yes | |
| findings | Yes | |
| truncated | No | |
| codePoints | No | |
| tagPayload | No | |
| normalizedNfkcDiffers | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare read-only/idempotent/non-destructive, so the bar is lower, and the description still adds real value: it explains the risk-tiering logic (ZWJ/ZWNJ/VS16 in legitimate emoji, Persian/Arabic joining, or Indic virama contexts downgrade to info joiner_in_context, otherwise medium) and notes lone surrogates are flagged. That conditional-severity policy is behavioral nuance not present in any structured field.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The detection list is front-loaded and every clause names a distinct finding category, so density is earned. The trailing 'Read-only. Same engine as GET /api/scan.' is slightly redundant against the annotations but is short and provides consistency context.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With an output schema present, the description needn't restate return shape, yet it still summarizes risk level, counts, findings, UTF-16 offsets and code points. Detection categories, edge-case exceptions, and severity policy are all covered; the only gap is how it fits into a workflow with clean_text.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Only one parameter and schema coverage is 100% including the 2000 UTF-16 unit limit, so the schema does all the work. The description never mentions the input bound or what counts as valid text, so it adds nothing beyond structured data — baseline 3 applies.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Opens with a specific verb ('Find') and enumerates the concrete resource classes it detects: invisible characters, bidi controls, tag payloads, variation-selector runs, mixed-script words, lookalikes. That enumeration makes it unmistakably a detection-only scanner, distinct from clean_text (which would mutate) and compare_names (which compares).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Usage is implied by the detection framing and 'Read-only', but there is no explicit when-to-use statement and no routing to siblings — nothing says whether to run this before clean_text or how it relates to compare_names. The 'Same engine as GET /api/scan' line is endpoint parity, not usage guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
clean_textcompare_namesscan_textFind hidden prompt-injection text: ASCII smuggling, zero-width/emoji stego, bidi, homoglyphs.
Scan AI agent skills and configs for hidden Unicode, prompt injection and exfiltration.
Check text before your agent reads it: hidden instructions, invisible text and links that leak data.
Scans untrusted web pages, emails, MCP responses, or A2A payloads for hidden prompt injections an.
Add one secure layer between your agents and this server.