Examples
Working reefpods you can copy and adapt. Each is a complete
pod.toml validated against the v0.1 schema. Pair each with a
prompts/system.md directive and run
konareef pod validate to confirm your copy.
Hello World
Greets a name and writes a one-line greeting. Beginner · lobster · 1000 sats.
Research brief
Researches a topic into a short markdown brief. Intermediate · orca · 3000 sats.
Meeting summary
Turns notes into a summary with action items. Beginner · lobster · 800 sats.
Hello World
What it does: greets a supplied name and
writes the greeting to a file. The smallest useful pod — start here.
Full walkthrough in the Quickstart.
Difficulty: Beginner · Runtime: lobster · Budget: 1000 sats · Uses: input, file deliverable, budget
pod_spec_version = "0.1"
[pod]
name = "hello-world"
version = "0.1.0"
description = "A tiny reefpod that greets a user by name."
[runtime]
kind = "lobster"
version = "^0.1"
[model]
provider = "anthropic"
name = "claude-sonnet-4-5"
[inputs]
name = { type = "string", required = true, description = "The name to greet" }
[directive]
template = "./prompts/system.md"
max_iterations = 3
[output]
[[output.success]]
kind = "file_exists"
path = "./greeting.md"
[[output.failure]]
kind = "timeout_seconds"
value = 120
[[output.deliverables]]
path = "./greeting.md"
description = "The generated greeting."
[budget]
max_sats = 1000
Research brief
What it does: researches a topic and writes
a markdown brief of at least 150 words. Uses the iterative
orca runtime and a min_words success predicate.
Difficulty: Intermediate · Runtime: orca · Budget: 3000 sats · Uses: input, min_words predicate, deliverable
pod_spec_version = "0.1"
[pod]
name = "research-brief"
version = "0.1.0"
description = "Researches a topic and writes a short markdown brief."
tags = ["research", "writing"]
[runtime]
kind = "orca"
version = "^0.1"
[model]
provider = "anthropic"
name = "claude-sonnet-4-5"
[inputs]
topic = { type = "string", required = true, description = "The topic to research." }
[directive]
template = "./prompts/system.md"
max_iterations = 6
[output]
format = "markdown"
[[output.success]]
kind = "min_words"
path = "./brief.md"
count = 150
[[output.failure]]
kind = "timeout_seconds"
value = 600
[[output.deliverables]]
path = "./brief.md"
description = "The research brief."
[budget]
max_sats = 3000
Meeting summary
What it does: turns raw meeting notes into a
summary that must contain an “Action items” section (enforced with a
contains_regex predicate) within a tight per-call budget.
Difficulty: Beginner · Runtime: lobster · Budget: 800 sats (400/call) · Uses: input, contains_regex predicate, per-call cap
pod_spec_version = "0.1"
[pod]
name = "meeting-summary"
version = "0.1.0"
description = "Turns raw meeting notes into a summary with action items."
tags = ["productivity", "summary"]
[runtime]
kind = "lobster"
version = "^0.1"
[model]
provider = "anthropic"
name = "claude-sonnet-4-5"
[inputs]
notes = { type = "string", required = true, description = "Raw meeting notes." }
[directive]
template = "./prompts/system.md"
max_iterations = 2
[output]
[[output.success]]
kind = "contains_regex"
path = "./summary.md"
pattern = "Action items"
[[output.failure]]
kind = "budget_exhausted"
[[output.deliverables]]
path = "./summary.md"
description = "Meeting summary with action items."
[budget]
max_sats = 800
per_call_cap_sats = 400
Save any example as pod.toml, add a
prompts/system.md directive, and run
konareef pod validate ./pod.toml. Tighten the budget and
predicates to match your workflow — see the
pod.toml reference and
Security & trust.