nostr-event-builder
Nostr Event Builder
Overview
Construct correct Nostr event structures from natural language descriptions. This skill handles the non-obvious parts: choosing the right kind, building proper tag arrays with correct markers, enforcing NIP-10 vs NIP-22 threading rules, and producing valid event JSON ready for signing.
When to Use
- The task is to construct or validate a concrete Nostr event JSON object and its tag array.
- The user describes what they want to publish and needs the exact event shape, kind, tags, and content rules.
- The problem is correctness of event structure for replies, comments, metadata, or other publishable events.
- The output should be an event payload, not protocol-planning advice.
Do NOT use when:
- The task is deciding which NIPs apply at a high level.
- The work is subscription filter design or relay query construction.
- The problem is relay implementation rather than event payload construction.
Response format
Always structure the final response with these top-level sections, in this order:
- Summary — state the task, scope, and main conclusion in 1-3 sentences.
- Decision / Approach — state the key classification, assumptions, or chosen path.
- Artifacts — provide the primary deliverable(s) for this skill. Use clear subheadings for multiple files, commands, JSON payloads, queries, or documents.
- Validation — state checks performed, important risks, caveats, or unresolved questions.
- Next steps — list concrete follow-up actions, or write
Noneif nothing remains.
Rules:
- Do not omit a section; write
Nonewhen a section does not apply. - If files are produced, list each file path under Artifacts before its contents.
- If commands, JSON, SQL, YAML, or code are produced, put each artifact in fenced code blocks with the correct language tag when possible.
- Keep section names exactly as written above so output stays predictable across skills.
Workflow
1. Identify the Event Kind
Ask: "What is the developer trying to publish?"
| Intent | Kind | Category | Key NIP |
|---|---|---|---|
| Post a short text note | 1 | Regular | NIP-10 |
| Reply to a kind:1 note | 1 | Regular | NIP-10 |
| Comment on non-kind:1 content | 1111 | Regular | NIP-22 |
| Set/update user profile | 0 | Replaceable | NIP-01 |
| Update follow list | 3 | Replaceable | NIP-02 |
| Delete events | 5 | Regular | NIP-09 |
| Repost a note | 6 | Regular | NIP-18 |
| React to an event | 7 | Regular | NIP-25 |
| Publish a long-form article | 30023 | Addressable | NIP-23 |
See references/event-kinds.md for full kind-to-structure mapping.
Critical routing rule:
Is the target a kind:1 note?
YES → Use kind:1 reply with NIP-10 e-tag markers
NO → Use kind:1111 comment with NIP-22 uppercase/lowercase tags
2. Build the Tag Array
Tags are the hardest part. Follow the tag guide for your kind:
For kind:1 replies (NIP-10):
Direct reply to root (no intermediate replies):
{
"kind": 1,
"tags": [
["e", "<root-event-id>", "<relay-url>", "root", "<root-author-pubkey>"],
["p", "<root-author-pubkey>"]
],
"content": "Your reply text"
}
Reply to a reply in a thread:
{
"kind": 1,
"tags": [
["e", "<root-event-id>", "<relay-url>", "root", "<root-author-pubkey>"],
[
"e",
"<parent-event-id>",
"<relay-url>",
"reply",
"<parent-author-pubkey>"
],
["p", "<root-author-pubkey>"],
["p", "<parent-author-pubkey>"]
],
"content": "Your reply text"
}
For kind:1111 comments (NIP-22):
Top-level comment on a regular event:
{
"kind": 1111,
"tags": [
["E", "<root-event-id>", "<relay-url>", "<root-author-pubkey>"],
["K", "<root-event-kind>"],
["P", "<root-author-pubkey>", "<relay-url>"],
["e", "<root-event-id>", "<relay-url>", "<root-author-pubkey>"],
["k", "<root-event-kind>"],
["p", "<root-author-pubkey>", "<relay-url>"]
],
"content": "Your comment text"
}
Top-level comment on an addressable event (kind 30000-39999):
{
"kind": 1111,
"tags": [
["A", "<kind>:<pubkey>:<d-tag>", "<relay-url>"],
["K", "<root-event-kind>"],
["P", "<root-author-pubkey>", "<relay-url>"],
["a", "<kind>:<pubkey>:<d-tag>", "<relay-url>"],
["e", "<event-id>", "<relay-url>"],
["k", "<root-event-kind>"],
["p", "<root-author-pubkey>", "<relay-url>"]
],
"content": "Your comment text"
}
Comment on a URL or external identifier:
{
"kind": 1111,
"tags": [
["I", "<url-or-identifier>"],
["K", "<identifier-type>"],
["i", "<url-or-identifier>"],
["k", "<identifier-type>"]
],
"content": "Your comment text"
}
Reply to an existing comment:
{
"kind": 1111,
"tags": [
["E", "<original-root-event-id>", "<relay-url>", "<root-author-pubkey>"],
["K", "<original-root-kind>"],
["P", "<root-author-pubkey>"],
[
"e",
"<parent-comment-id>",
"<relay-url>",
"<parent-comment-author-pubkey>"
],
["k", "1111"],
["p", "<parent-comment-author-pubkey>"]
],
"content": "Your reply to the comment"
}
See references/tag-guide.md for complete tag semantics.
3. Set the Content Field
Content format depends on the kind:
| Kind | Content Format |
|---|---|
| 0 | Stringified JSON: {"name":"...","about":"...","picture":"..."} |
| 1 | Plaintext (no markdown, no HTML) |
| 5 | Optional deletion reason text |
| 6 | Stringified JSON of the reposted event |
| 7 | + (like), - (dislike), or emoji |
| 1111 | Plaintext comment |
| 30023 | Markdown-formatted article body |
4. Construct the Complete Event
Assemble the unsigned event object:
{
"pubkey": "<32-bytes-lowercase-hex-public-key>",
"created_at": "<unix-timestamp-seconds>",
"kind": "<integer>",
"tags": [["..."]],
"content": "<string>"
}
The id is computed as SHA-256 of the serialized form:
[0, "<pubkey>", <created_at>, <kind>, <tags>, "<content>"]
Serialization rules:
- UTF-8 encoding, no whitespace/formatting
- Escape in content:
\n,\",\\,\r,\t,\b,\f - All other characters verbatim
The sig is a Schnorr signature (secp256k1) of the id.
5. Validate Before Signing
Checklist before the event is ready:
-
kindis correct for the intent - All required tags present for this kind
-
etags have correct markers (root/replyfor kind:1) -
ptags include ALL participants in the thread - NIP-10 kind:1 replies only target other kind:1 events
- NIP-22 kind:1111 comments do NOT target kind:1 events
-
Kandktags present for kind:1111 comments - Uppercase tags (E/A/I/K/P) point to root scope in kind:1111
- Lowercase tags (e/a/i/k/p) point to parent item in kind:1111
-
contentformat matches the kind's requirements -
created_atis a Unix timestamp in seconds (not milliseconds) - All hex values are 32-byte lowercase
Common Mistakes
| Mistake | Why It Breaks | Fix |
|---|---|---|
| Using kind:1 to reply to a kind:30023 article | NIP-10 kind:1 replies MUST only reply to other kind:1 events | Use kind:1111 (NIP-22 comment) for non-kind:1 targets |
| Using kind:1111 to reply to a kind:1 note | NIP-22 comments MUST NOT reply to kind:1 | Use kind:1 with NIP-10 e-tag markers |
Missing root marker on e tags in kind:1 |
Clients can't reconstruct the thread tree | Always use marked e tags: ["e", "<id>", "<relay>", "root"] |
Only one e tag with reply marker (no root) |
Direct replies to root need root marker, not reply |
Single e tag = use root marker only |
| Missing p tags for thread participants | Users don't get notified of replies | Include p tags for ALL pubkeys in the thread |
| Lowercase e/k/p tags for root scope in kind:1111 | Root scope MUST use uppercase E/K/P tags | Uppercase = root scope, lowercase = parent item |
| Missing K or k tags in kind:1111 | Both are REQUIRED by NIP-22 | Always include ["K", "<root-kind>"] and ["k", "<parent-kind>"] |
created_at in milliseconds |
Nostr uses seconds, not milliseconds | Use Math.floor(Date.now() / 1000) |
| Content as object instead of string for kind:0 | Content must be stringified JSON | Use JSON.stringify({name: "...", ...}) |
Missing d tag on addressable events (30000-39999) |
Relay can't address the event properly | Always include ["d", "<identifier>"] |
| Positional e tags without markers | Deprecated; creates ambiguity in thread reconstruction | Always use marked e tags with root/reply |
Kind Category Quick Reference
| Range | Category | Behavior |
|---|---|---|
| 1000-9999, 4-44, 1, 2 | Regular | Stored by relays, all kept |
| 10000-19999, 0, 3 | Replaceable | Latest per pubkey+kind kept |
| 20000-29999 | Ephemeral | Not stored by relays |
| 30000-39999 | Addressable | Latest per pubkey+kind+d-tag kept |
Example: Complete Event Construction
User says: "I want to reply to my friend's note in an existing thread"
Step 1 — Identify kind: Replying to a kind:1 note → use kind:1 (NIP-10)
Step 2 — Determine thread position: This is a reply to a reply (not the
root), so we need both root and reply e tags.
Step 3 — Build the event:
{
"pubkey": "a1b2c3d4e5f6a7b8c9d0e1f2a3b4c5d6e7f8a9b0c1d2e3f4a5b6c7d8e9f0a1b2",
"created_at": 1709827200,
"kind": 1,
"tags": [
["e", "aaa111...", "wss://relay.example.com", "root", "f7234bd4..."],
["e", "bbb222...", "wss://relay.example.com", "reply", "93ef2eba..."],
["p", "f7234bd4...", "wss://relay.example.com"],
["p", "93ef2eba...", "wss://relay.example.com"]
],
"content": "Great point! I totally agree with this take."
}
Step 4 — Validate: Root e tag has root marker ✓, reply e tag has reply
marker ✓, p tags include both the root author and the parent author ✓, content
is plaintext ✓.
More from accolver/skill-maker
skill-maker
Create or iteratively improve agent skills with eval-driven refinement when the task is to build a new SKILL.md package or tune an existing skill’s trigger accuracy and performance.
26git-conventional-commits
Generate conventional commit messages from staged git changes when the task is to name or format a commit, not to review code or write release notes.
16pdf-toolkit
Operate the bundled PDF scripts to extract, OCR, create, merge, split, or convert PDFs when the task explicitly involves PDF document processing.
14error-handling
Standardize application error handling—taxonomy, propagation, response shapes, logging, and correlation IDs—when the task is to improve consistency of existing error behavior across a codebase.
12pr-description
Generate pull-request descriptions from branch diffs when the task is to explain change scope, motivation, testing, risk, and reviewer guidance for a reviewable branch.
12code-reviewer
Review existing code or diffs for bugs, security issues, performance problems, maintainability risks, and test gaps when the user wants evaluation, not new implementation.
12