New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Nostr

by OpenAgentsInc in OpenAgentsInc/openagents

How Nostr works in this repository: the three NIP lanes under nips/, the wire protocol (NIP-01), relay authentication (NIP-42), payload encryption (NIP-44), the Block agent NIPs the relay serves…

Apache-2.0Auto-check passedDevelopment

Install Nostr

skills CLI
$ npx skills add OpenAgentsInc/openagents --skill nostr -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install OpenAgentsInc/openagents nostr --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/OpenAgentsInc/openagents.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/nostr .claude/skills/nostr && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
nostr
GitHub stars
455
Token cost
~6.2k tokens
SKILL.md length
3,032 words
Files
1
Skills in repo
5
Repo updated
First seen
Licence
Apache-2.0

At a glance

How Nostr works in this repository: the three NIP lanes under nips/, the wire protocol (NIP-01), relay authentication (NIP-42), payload encryption (NIP-44), the Block agent NIPs the relay serves…

  • Works in 5 steps: The worker connects, answers AUTH, and… → The terminal connects, answers AUTH,… → The relay fans the request out to the… → …
  • Development work in your project
  • SKILL.md covers Where the specifications live, The protocol in one page…, Authentication (NIP-42) and Encryption (NIP-44), plus 5 more sections
  • Needs CODER_SECRET_KEY and CODER_WORKER_SECRET

What it does

Nostr is an agent skill from OpenAgentsInc/openagents. How Nostr works in this repository: the three NIP lanes under nips/, the wire protocol (NIP-01), relay authentication (NIP-42), payload encryption (NIP-44), the Block agent NIPs the relay serves, and the OpenAgents NIPs (NIP-CJ jobs, NIP-CAP capabilities, NIP-PRG programs) that carry a Coder turn to a worker. Read it before touching crates/nostr, crates/nostr-relay, crates/coder/src/relay.rs, coder-worker, or any file under nips/, and before debugging a relay handoff.

Its SKILL.md is about 6.2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Development. The licence is Apache-2.0.

When your agent uses it

  • Development work in your project

Example prompts

  • “/nostr”

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. The worker connects, answers AUTH, and subscribes
  2. The terminal connects, answers AUTH, signs a kind-25900 request
  3. The relay fans the request out to the worker's subscription. The
  4. The terminal accepts an event only if the signature covers what it
  5. The job ends on the first result or status: error. The terminal

What it can do on your machine

Read from SKILL.md and the folder at commit ad29644. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are bash).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Links to these hosts (documentation or services it may open):

    • github.com

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names these keys or tokens, usually read from environment variables:

    • CODER_SECRET_KEY
    • CODER_WORKER_SECRET
    • CODER_DOOR_KEY

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Nostr loads about 6.2k tokens when it runs. Until then it costs about 123 tokens; SKILL.md has 3,032 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~123
When it runs · the whole SKILL.md, loaded when a task matches
~6.2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from OpenAgentsInc/openagents at commit ad29644, republished under its Apache-2.0 licence (© OpenAgentsInc). 3,032 words, ~6,158 tokens.

Download SKILL.mdSave it as .claude/skills/nostr/SKILL.md (or your agent's skills folder).
name
nostr
description
How Nostr works in this repository: the three NIP lanes under `nips/`, the wire protocol (NIP-01), relay authentication (NIP-42), payload encryption (NIP-44), the Block agent NIPs the relay serves, and the OpenAgents NIPs (NIP-CJ jobs, NIP-CAP capabilities, NIP-PRG programs) that carry a Coder turn to a worker. Read it before touching `crates/nostr`, `crates/nostr-relay`, `crates/coder/src/relay.rs`, `coder-worker`, or any file under `nips/`, and before debugging a relay handoff.

Nostr in OpenAgents

Use this skill when you change or debug anything that touches a relay: the protocol crates, the relay binary, the Coder relay door, the worker, or a specification under nips/. The specifications are the source of truth; this skill tells you where they are and how the pieces fit, so you don't rediscover the flow from code.

Where the specifications live

LanePathSource of truthSynced
Official NIPsnips/official/nostr-protocol/nipsyes, ./scripts/sync-nips.sh
Block (Buzz) extension NIPsnips/block/block/buzzyes
OpenAgents NIPsnips/openagents/this repositoryno — the files here are authoritative

nips/manifest.json pins the upstream commit of each synced lane. A sync never changes the implementation without review and a fixture update. nips/README.md states the mandate: every pinned spec that applies to a relay is an implementation target for crates/nostr and crates/nostr-relay, optional features stay fail-closed and out of NIP-11 until they run, and client-only NIPs are implemented as fixture-backed clients rather than pretended.

Read the September 26 upstream review and implementation coverage before assuming complete support. Source inventory checks now track the current pins separately from behavioral evidence. PMA 30179 is rejected, client-authored 39007 is refused, and app data and private artifacts have explicit visibility. RS has an opt-in atomic HTTP snapshot. The PL executor is off by default and needs a credential-holding push gateway; no real device delivery is recorded. Client helpers do not imply complete AP, FI, CW-thread, or RS merge integration.

Read the spec before the code. The files you need most often:

  • nips/official/01.md — the protocol: events, kinds, filters, and the client-relay messages.
  • nips/official/42.md — client authentication to a relay.
  • nips/official/44.md — encrypted payloads.
  • nips/official/11.md — the relay information document.
  • nips/official/40.md — the expiration tag.
  • nips/official/19.md — npub and nsec encoding.
  • nips/block/README.md — a per-spec summary of all 17 Block NIPs.
  • nips/openagents/NIP-CJ.md — Coder jobs, the protocol between coder and coder-worker.
  • nips/openagents/NIP-DEC.md — decisions: typed questions with structured (EntryType) instructions and criteria, carried by the CJ decision family.
  • nips/openagents/NIP-CAP.md — capability manifests and presence.
  • nips/openagents/NIP-PRG.md — programs.
  • nips/openagents/NIP-EXT.md — extension releases, discovery, revocation, foreign-format import, and compatible host component-set assessments.
  • nips/openagents/NIP-REG.md — curated plugin registries, exact publisher release pins, per-registry trust, and signed Nostr and GitHub/HTTPS sources.
  • nips/openagents/NIP-RUN.md — encrypted durable journals and recovery.
  • nips/openagents/NIP-CTX.md — task frames, evidence views, and expansion.
  • nips/openagents/NIP-POL.md — instructions, approvals, disclosure, routing records, and governed activation of learned private preferences.
  • nips/openagents/NIP-COORD.md — task claims, fencing, and background findings.
  • nips/openagents/NIP-EVAL.md — attributable workload evaluation and promotion evidence.
  • nips/openagents/NIP-OPT.md — semantic AI contracts, immutable implementations, bounded studies, actual candidate execution, and optimization evidence.
  • nips/openagents/NIP-XP.md — quests, referee awards of accepted outcomes, revocations, and per-reader XP ledgers; crates/nostr (xp) checks them.
  • nips/openagents/NIP-KB.md — shared knowledge entries, heads, withdrawals, and evidence; crates/nostr (kb) checks them.
  • nips/openagents/NIP-CTRL.md — task-scoped client pairing, observation, steering, and cancellation; no approval or wallet authority from pairing.
  • nips/openagents/NIP-MKT.md — negotiated offerings, bilateral orders, private records, and separately admitted fixed-price Lightning settlement.
  • nips/openagents/NIP-LAB.md — bounded agent labor, exact execution linkage, deliverables, independent checks, rework, disputes, acceptance, and rights.
  • nips/openagents/NIP-X402.md — Designed Lightning-paid operations before execution, standard x402 HTTP/MCP bindings, and an opt-in native Nostr profile.
  • nips/openagents/NIP-SESS.md — engine sessions, actual adapter support, queues, turn control, pending interactions, and retained native history.
  • nips/openagents/NIP-WS.md — workspace/resource identity, conditional edits, checkpoints, and bounded projection snapshots, deltas, and cuts.
  • nips/openagents/NIP-WORK.md — tracked objectives, planning relationships, accountable ownership, delegation, exact revisions, and work disposition.
  • nips/openagents/NIP-AUTO.md — durable schedules, source watches, bounded goal continuations, occurrence identity, and restart-safe accounting.
  • nips/openagents/NIP-ENV.md — execution environment leases, actual materialization, attachment, cleanup, and uncertain resource accounting.
  • nips/openagents/NIP-LIVE.md — admitted media/capture, participants and recipients, input and speaking floors, evidence, and scoped device input.
  • nips/openagents/NIP-REACH.md — private host directory, bounded presence, reachability, and authenticated direct channels. Placement grants no rights.
  • nips/openagents/NIP-HOST.md — host-wide device enrollment and scoped grants; access rights do not supply POL approval or spending authority.
  • nips/openagents/NIP-TERM.md — host-owned interactive terminals, bounded output replay, explicit gaps, and current access checks.
  • nips/openagents/NIP-SOV.md — Designed successor to historical SA: durable identity, supported custody, bounded AUTO lifecycle, POL guardians, treasury policy, and retained recovery evidence. It allocates no new event kinds.
  • nips/openagents/NIP-MV.md — Verse world presence, entity state, and gestures; independent of the shared agent-artifact contracts.
  • docs/optimization/README.md — DSPy/GEPA concepts, Gym evaluation, host boundaries, and a consolidated unfiled integration backlog.
  • nips/openagents/contracts.md — pinned identities, schemas, locks, evidence, context, effects, outcomes, and private artifact envelopes for the v1 contracts.
  • docs/coder/design/typesafe-agent-protocol-addendum.md — what belongs in Nostr and what hosts/clients must implement; complete source-proposal coverage.
  • docs/protocol/implementation-plan.md — implementation across all lanes.
  • docs/agents/README.md and docs/agents/roadmap.md — general agent infrastructure, Coder's domain boundary, and remaining non-code contracts.
  • docs/protocol/block-nips.md — what the relay does with each Block NIP, including what it deliberately doesn't advertise.

The OpenAgents lane contains 30 NIPs plus the shared contracts. X402, SESS, WS, WORK, AUTO, ENV, LIVE, and SOV are Designed drafts, as are POL's learned preference lifecycle and EXT's import/component-set assessment additions. REG is a Designed profile using inert EXT catalogs and existing release/head kinds; no registry client is implemented. These profiles allocate no new event kinds: private artifacts use 3188, remote host operations use admitted CAP/CJ profiles, and effects leave RUN evidence. Keeping an encrypted envelope does not implement the contract inside it. SOV does not restore the legacy 392xx allocations or imply working key custody, guardians, or payments. REACH has a narrower coder-reach implementation; consult its role limits before treating reachability as availability or controller-transfer authority. ATIF is a Designed draft that does allocate kinds: public trajectory declarations 3198 and chunks 3199; private trajectories ride 3188. It replaces historical 39230/39231; no component publishes trajectories yet. MKT/LAB has a bounded free-order host in crates/coder-labor, with retained relay and process fixtures. Paid settlement, independently operated service delivery, and broader market roles remain unfinished. CTRL has a bounded local-owner bridge in crates/coder-control; read docs/coder/runtime/nostr-task-control.md for its rights, expiry, exact retry, and disclosure limits. It does not complete mobile control or a cross-device product. Check the glossary for the narrower implemented and partial roles.

Read the x402 integration assessment before adding a payment adapter. NIP-X402 preserves upstream http:1 and mcp:1; nostr:openagents:1 is an opt-in OpenAgents extension that requires explicit facilitator support. It pays before execution; MKT/LAB pays after acceptance. NWC is wallet transport, not purchase authority. Zap invoices hash the zap request, while x402 invoices hash the bound operation; do not reuse one invoice as both. L402/LSAT is a distinct macaroon-based protocol. These paths have no operational OpenAgents wallet or settlement adapter. Pure validation components do not consume proofs or authorize spending.

Read the teardown integration plan and file-by-file archive coverage when carrying historical ideas forward. The archive supplies design evidence; its old TypeScript, hosting, and product decisions are not current instructions. A tracked work item, session, run, environment, view, and commercial order have separate identities. Assignment does not dispatch; completion does not accept labor; acceptance does not pay; a live view does not grant mutation. Hosts must enforce these boundaries before advertising the corresponding role.

The protocol in one page (NIP-01)

An event is a signed JSON object: id (SHA-256 of the serialized fields), pubkey (the author's x-only secp256k1 key, hex), created_at (unix seconds), kind, tags (arrays of strings; the first element is the tag name), content, and sig (Schnorr). The id and sig are what make an event evidence: a relay can relabel or drop an event, but it can't forge one.

Kind ranges decide how a relay stores an event:

RangeClassRelay behavior
everything not listed belowregularstored, all kept
0, 3, 10000–19999replaceablelatest per pubkey+kind kept
20000–29999ephemeralfanned out to open subscriptions, never stored
30000–39999addressablelatest per pubkey+kind+d tag kept

crates/nostr implements this as EventClass::from_kind in domain/replacement.rs. Every NIP-CJ job kind is ephemeral; NIP-CAP and NIP-PRG discovery kinds are addressable. NIP-EXT and NIP-RUN add regular immutable records with separate addressable heads. A discovery head is never an execution version pin.

Messages, client to relay:

  • ["EVENT", <event>] — publish.
  • ["REQ", <sub_id>, <filter>, ...] — subscribe; the relay sends stored matches, then EOSE, then live matches until CLOSE.
  • ["CLOSE", <sub_id>] — end a subscription.
  • ["AUTH", <event>] — answer a NIP-42 challenge.

Relay to client:

  • ["EVENT", <sub_id>, <event>] — a match.
  • ["OK", <event_id>, true|false, <message>] — the verdict on an EVENT or AUTH. A false message starts with a machine-readable prefix: duplicate:, pow:, blocked:, rate-limited:, invalid:, restricted:, mute:, auth-required:, or error:.
  • ["EOSE", <sub_id>] — end of stored events; what follows is live.
  • ["CLOSED", <sub_id>, <message>] — the relay ended the subscription, with the same prefixes.
  • ["NOTICE", <message>] — human-readable.
  • ["AUTH", <challenge>] — a NIP-42 challenge.

Filters select on ids, authors, kinds, #<single-letter-tag>, since, until, and limit. Only single-letter tags are indexed, which is why NIP-CJ routes on e and p.

Authentication (NIP-42)

A relay may send ["AUTH", <challenge>] at any time; this relay sends one on connect whenever NOSTR_RELAY_URL is set. The client answers with a kind-22242 event carrying ["relay", <url>] and ["challenge", <challenge>] tags, created_at within ten minutes of now, and the relay replies ["OK", <auth_event_id>, true, ""]. Kind 22242 is never stored or broadcast.

An unauthenticated client isn't refused by default. Only what the relay gates needs auth:

  • Everything, when NOSTR_RELAY_AUTH_REQUIRED=true — the relay answers auth-required: on EVENT and CLOSED ... auth-required: on REQ.
  • Reads of private Block kinds (gift wraps 1059, engrams 30174, turn metrics 44200, and others listed in event_visible_to_reader in crates/nostr-relay/src/gateway/subscription.rs). A subscription's read_pubkeys is the set of keys the connection has authenticated as.
  • Protected events (NIP-70 - tag), closed-membership relays, and every relay command in docs/protocol/block-nips.md.

The Coder client and worker always authenticate when challenged — the npub is the account — and both refuse a connection whose challenge gets no OK. NIP-AA lets an agent key satisfy AUTH with an owner attestation (nips/block/NIP-OA.md) instead of its own membership.

Encryption (NIP-44)

NIP-44 version 2 encrypts a payload under a conversation key derived from the sender's secret and the recipient's public key (ECDH, then HKDF), so either party can decrypt and nobody else can. It hides the content and pads its length; it does not hide the tags, so kind, e, and p stay visible to the relay as routing metadata. Gift wraps (NIP-59) and every NIP-CJ payload use it. The implementation is crates/nostr's nip44 module.

Show full SKILL.md (1,397 more words)Show less

Coder jobs (NIP-CJ)

nips/openagents/NIP-CJ.md is the contract between coder (the customer) and coder-worker (the fulfiller). The relay is transport only: it holds no job state and sees only ciphertext.

KindNameDirectionPayload type
25900job requestterminal → workertask, transcript, instructions, client, v
27000job feedbackworker → terminaljudgment, partial (with seq under v: 1), status
26900job resultworker → terminalresult with text, optional usage and model

The flow, in the order the sockets speak it:

  1. The worker connects, answers AUTH, and subscribes {"kinds": [25900], "#p": [<worker pubkey>]}.
  2. The terminal connects, answers AUTH, signs a kind-25900 request p-tagged to the worker with NIP-44 content encrypted to the worker's key, subscribes {"kinds": [26900, 27000], "#e": [<request id>]}, and only then publishes the request. Subscribing first closes the race where fast feedback would be missed.
  3. The relay fans the request out to the worker's subscription. The worker decrypts, generates through its door, and publishes feedback and exactly one result, each e-tagged to the request and p-tagged to the customer, encrypted to the customer's key.
  4. The terminal accepts an event only if the signature covers what it expects: the kind is 26900 or 27000, the signer is the configured worker, an e tag names this request, and a p tag names this terminal. The subscription label is a routing hint, never identity.
  5. The job ends on the first result or status: error. The terminal then sends CLOSE for the job's subscription and keeps the socket for the next turn. A relay CLOSED on that subscription ends the job as a relay error with the relay's reason. A socket that broke, or a wait that ran out with a subscription that might still deliver, is dropped, and the next turn opens a fresh one. A turn holds the socket for its duration, so a caller that cancels the turn drops the socket with it; the relay frees the subscription when the connection ends.

The conversation specification uses integer v1 with sequenced partials. Decision and execution bodies use their named v1 schemas. Validate each family against its own complete contract; do not infer support from a version alone.

Deadlines and failure words, on the terminal side (crates/coder/src/relay.rs): if nothing bound to the request arrives within the contact deadline, the turn fails with cause worker_absent ("no worker answered"). If a worker was heard but no result arrives within the answer deadline, the cause names the worker as silent. A relay OK … false on the request fails the turn with the relay's reason. Opening a socket, handshake and AUTH together, is bounded by CONNECT_TIMEOUT. A worker that fails upstream and publishes nothing looks identical to an absent worker from the terminal — check the worker's stderr before blaming the relay.

A worker whose door is a local executor (CODER_EXECUTOR) publishes one encrypted kind-27000 feedback event, {"type": "status", "status": "processing"}, as soon as it admits a delegation and before the executor starts. The terminal counts any well-formed kind-27000 status of queued, processing, or error as contact, so its 30-second contact deadline ends at admission rather than at the executor's answer.

Environment, terminal side: CODER_WORKER (worker npub or hex pubkey) and CODER_RELAY (ws:// or wss:// URL). The identity is CODER_SECRET_KEY or CODER_NSEC, or, when neither is set, the key at ~/.openagents/nostr-secret, created on first use with mode 0600. Only a missing file is a first use: an unreadable one is an error, so a permissions accident cannot silently mint a new npub. Creation is exclusive and atomic, and concurrent first runs agree on one key. A fresh keypair is a valid anonymous account; never print, log, or commit the secret.

Environment, worker side (crates/coder/src/bin/coder-worker.rs): CODER_WORKER_SECRET (64 hex or nsec), CODER_RELAY, and the door: CODER_DOOR_KEY with CODER_DOOR_URL and CODER_WORKER_MODEL for an Open Responses door, or CODER_EXECUTOR=<capability slug> for a local approved executor such as devin-local (see docs/coder/guides/worker-executor.md). Setting both is refused. The worker prints its pubkey on start; that value is what the terminal's CODER_WORKER names. --once answers one job and exits; --decline <CODE> refuses every job with a typed status error, for measuring the refusal path. CODER_WORKER_ALLOW (comma-separated npub or hex keys) limits which customers the worker answers; anyone else gets a typed not_admitted status, never silence. Unset admits everyone, which is right only on a local relay. CODER_WORKER_JOBS bounds how many jobs the worker runs at once; unset, an executor door runs as many as its manifest's concurrent_max and a model door runs four. A request past the bound is refused before anything runs, with a typed busy status, and the refusal releases the slot. The worker's key and the terminal's key must differ.

docs/coder/measurements/relay-transport.md is the measured proof that both ends meet, with per-transport latency and refusal causes.

Capabilities and programs (NIP-CAP, NIP-PRG)

CAP describes portable interfaces, host bindings, presence, and operator preferences. Reading a manifest is inert; probes and invocations have separate admission. PRG defines typed workflows, seven step kinds, bounded composition, and the Wasm packet ABI. EXT distributes immutable components and RUN records durable execution.

OPT separates semantic AI signatures from their concrete implementations. Programs can invoke a supported pinned implementation through the same host boundary. Compiler, optimizer, evaluator, and inference capabilities remain distinct. EVAL supplies workload evidence and scoped promotion; neither a score nor a package signature grants execution.

Follow the normative NIPs for exact fields and validation. Implement only supported roles and advertise them after conformance. Artifact identity, actual materialization, protected evaluation, and operator adoption are separate checks. Local operation need not publish an event for every action.

The relay (crates/nostr-relay)

One binary, one Postgres database. Configuration is environment only; deploy/nostr-relay.env.example and docs/deployment/configuration.md list every variable. The ones you need for a local run:

sh
DATABASE_URL=postgres://<user>:<password>@127.0.0.1:5432/<db>
NOSTR_RELAY_BIND_ADDR=127.0.0.1
NOSTR_RELAY_PORT=7447
NOSTR_RELAY_URL=ws://127.0.0.1:7447      # enables the NIP-42 challenge
NOSTR_RELAY_MEDIA_ROOT=$HOME/relay-media  # optional; absolute path
NOSTR_RELAY_AUTH_REQUIRED=false           # true to gate EVENT and REQ
NOSTR_RELAY_LOG_LEVEL=info

Migrations run on first boot and the start fails closed if the ledger disagrees with the database. GET /health and a NIP-11 request (Accept: application/nostr+json on /) confirm it's up. docs/deployment/runbook-local-dev.md walks the same steps with a disposable Postgres cluster and the nak client.

Live fan-out is in gateway/subscription.rs: a published event is indexed by id, author, kind, and each single-letter tag; a subscription is matched against its filters, then checked with event_visible_to_reader for the private kinds; a subscription still loading history buffers live matches until EOSE. Ephemeral events are deduplicated by id over a short window and never reach storage.

The Block NIPs the relay serves, and how, are in docs/protocol/block-nips.md. The relay-signed kinds (39005, 39006, 39007, 30622) are refused from clients with restricted:.

Debugging a handoff

When coder -p reports worker_absent:

  1. Confirm the worker is connected and subscribed: its stderr shows worker <pubkey>, relay <url>, door …, and waiting for jobs. CODER_WORKER must equal that printed pubkey.
  2. Confirm the relay fans out: open a second subscription with the worker's filter (a nak req or a scripted WebSocket client) and run the turn again. Seeing the kind-25900 event there proves the relay and the terminal; not seeing it means the request was refused (read the OK) or the terminal's CODER_RELAY points elsewhere.
  3. Read the worker's stderr for job <id> failed: …. A worker whose door fails upstream publishes nothing, and the terminal can't tell that from silence.
  4. Run the relay with NOSTR_RELAY_LOG_LEVEL=debug. It then logs one line per admitted ephemeral event: kind, ID, author, e and p tags, and the content's byte length, never the content. A 25900 with no 26900 tagged e to it is a job the worker never answered. Ephemeral kinds are not stored, so the database holds nothing to query afterwards; the log is the only relay-side record.
  5. Only then read relay code.

Rules that don't bend

  • Product code is Rust. crates/nostr has no storage, no network, and no third-party Nostr crate.
  • Never print, log, commit, or paste a secret key, nsec, or door key. Test fixtures use throwaway keys generated in the test.
  • The relay is transport, not authority. Anything a relay could relabel or replay must be checked against the signature — kind, signer, e, and p — before the payload is read.
  • Unverifiable state is refused, not assumed: an unknown probe is not present, an unsigned answer is not an answer, and a feature that isn't configured isn't advertised in NIP-11.
  • A NIP change goes spec first (nips/openagents/ or a synced upstream commit), then validators and fixtures, then the applicable host, client, or relay role. A client-only or encrypted host contract does not become relay behavior merely because its envelope passes validation.

© OpenAgentsInc, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .agents/skills/nostr of OpenAgentsInc/openagents.

Open the folder on GitHubat commit ad29644

Compare with similar skills

Nostr next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Nostr compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Nostr this skillOpenAgentsInc/openagents455—~6.2kAutomated safety check: PassApache-2.0
Vercel Composition Patternssupabase/supabase111k59 repos~726Automated safety check: PassMIT
Finishing a Development Branchobra/superpowers296k5 repos~1.9kAutomated safety check: PassMIT
Typescript Advanced Typesrolling-scopes/rsschool-app10k25 repos~4.2kAutomated safety check: PassMPL-2.0
PR Babysitteropeninterpreter/openinterpreter69k3 repos~4.2kAutomated safety check: PassApache-2.0
Code Review ChecklistshareAI-lab/learn-claude-code78k5 repos~1.1kAutomated safety check: PassMIT

Similar skills

  • Vercel Composition Patterns

    supabase/supabase

    Official

    React composition patterns that scale. An agent skill from supabase/supabase.

    111k GitHub starsUsed in 59 repos~726 tokens
    DevelopmentAuto-check passed

  • Finishing a Development Branch

    obra/superpowers

    Walks the last step of a branch: confirm tests pass, detect the git environment, ask how to integrate, carry out your choice and clean up the worktree.

    296k GitHub starsUsed in 5 repos~1.9k tokens
    DevelopmentAuto-check passed

  • Typescript Advanced Types

    rolling-scopes/rsschool-app

    Master TypeScript's advanced type system including generics, conditional types, mapped types, template literals, and utility types for building type-safe applications.

    10k GitHub starsUsed in 25 repos~4.2k tokens
    DevelopmentAuto-check passed

  • PR Babysitter

    openinterpreter/openinterpreter

    Watches an open GitHub pull request until it merges, handling review comments, diagnosing CI failures and retrying flaky checks along the way.

    69k GitHub starsUsed in 3 repos~4.2k tokens
    DevelopmentAuto-check passed

  • Code Review Checklist

    shareAI-lab/learn-claude-code

    Reviews code against a five-part checklist covering security, correctness, performance, maintainability and testing, and reports findings in a fixed format.

    78k GitHub starsUsed in 5 repos~1.1k tokens
    DevelopmentAuto-check passed

  • Greploop

    onyx-dot-app/onyx

    Iteratively improves a PR (GitHub), MR (GitLab), or shelved changelist (Perforce) until Greptile gives it a 5/5 confidence score with zero unresolved comments.

    32k GitHub starsUsed in 4 repos~3.3k tokens
    DevelopmentAuto-check passed

More from OpenAgentsInc/openagents

  • Typesafe AI

    OpenAgentsInc/openagents

    Build AI-powered software with TypeSafe: small units of AI intelligence you can use like programming primitives.

    455 GitHub starsUsed in 9 repos~2.5k tokens
    Auto-check passed

  • Decision API

    OpenAgentsInc/openagents

    Call the OpenAgents decision API: authenticate with an oak bearer key, post state + typed questions to POST /v1/systemone, read Noul, Choice, and Score answers with their probabilities, and handle…

    455 GitHub stars~1k tokensUpdated today
    Auto-check passed

  • Google Developer Style

    OpenAgentsInc/openagents

    Apply the Google Developer Documentation Style Guide to user-facing docs, READMEs, AGENTS.md, and commit messages.

    455 GitHub stars~630 tokensUpdated today
    Auto-check passed

  • Openagents Decision API

    OpenAgentsInc/openagents

    Authenticate to and call the OpenAgents decision API — typed decisions, batch classification, durable jobs, and the bundled documentation — through the oak CLI, the oak-mcp server, or raw HTTP.

    455 GitHub stars~680 tokensUpdated today
    Auto-check passed

Categories

Questions about Nostr

What does Nostr do?

How Nostr works in this repository: the three NIP lanes under nips/, the wire protocol (NIP-01), relay authentication (NIP-42), payload encryption (NIP-44), the Block agent NIPs the relay serves…. Nostr is an agent skill from OpenAgentsInc/openagents. How Nostr works in this repository: the three NIP lanes under nips/, the wire protocol (NIP-01), relay authentication (NIP-42), payload encryption (NIP-44), the Block agent NIPs the relay serves, and the OpenAgents NIPs (NIP-CJ jobs, NIP-CAP capabilities, NIP-PRG programs) that carry a Coder turn to a worker.

When should I use Nostr?

Nostr fits situations like: development work in your project.

How do I install Nostr in Claude Code?

Run `npx skills add OpenAgentsInc/openagents --skill nostr -a claude-code`. Or copy the skill folder (.agents/skills/nostr in OpenAgentsInc/openagents) into .claude/skills/nostr in your project. Claude Code loads it when a task matches its description.

How do I install Nostr in Codex?

Run `npx skills add OpenAgentsInc/openagents --skill nostr -a codex`. Or copy the skill folder (.agents/skills/nostr in OpenAgentsInc/openagents) into .agents/skills/nostr in your project. Codex loads it when a task matches its description.

Can I use Nostr in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add OpenAgentsInc/openagents --skill nostr -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/nostr, .gemini/skills/nostr, .github/skills/nostr and .opencode/skills/nostr in your project.

What does Nostr need to run?

Going by SKILL.md and its folder, Nostr needs credentials named CODER_SECRET_KEY, CODER_WORKER_SECRET and CODER_DOOR_KEY.

Does Nostr access the network?

SKILL.md names 1 domain. As links in the text: github.com. This is read from the text; nothing was executed.

Is Nostr safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Nostr use?

Nostr is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Nostr use?

About 6.2k tokens (SKILL.md is roughly 25k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Nostr?

Skills that share tags, products or a category with Nostr: Vercel Composition Patterns (supabase/supabase, 111k stars), Finishing a Development Branch (obra/superpowers, 296k stars), Typescript Advanced Types (rolling-scopes/rsschool-app, 10k stars) and PR Babysitter (openinterpreter/openinterpreter, 69k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Nostr?

OpenAgentsInc (a GitHub organization) maintains it in OpenAgentsInc/openagents, which has 455 GitHub stars. The repository holds 5 skills in this directory. The repository was last updated on October 8, 2026.

Source: OpenAgentsInc/openagents on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.