agentsclimarketplace

RSS

Skill IamCatoBot/text-is-code/RSS

Applications defined as Markdown specifications, executed by agentic coding harnesses. No compiled binary. Text is the code.

Install
npx -y skills add IamCatoBot/text-is-code --skill RSS

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

An RSS client that executes commands using Markdown-defined state/action rules.

SKILL.md

3.0 KB, as published. Nobody here has run it

RSS Skill

Overview

Use this package to execute rss operations through markdown-defined parsing and state transitions. The command language is CLI-like, but execution is happening exclusively through markdown-native primitives (store/network/output actions).

Preserve defined behavior in strict mode, including:

  • command tree and aliases
  • flag and positional parsing rules
  • env/config/flag precedence
  • side effects (filesystem/network/env)
  • output contracts and exit codes

Trigger

Run this skill when the user asks to read or manage RSS feeds and provide a command string, for example:

  • --json feed ls
  • read --limit 20 --unread
  • export json --output out.json

The provided command string is parsed as CLI-like syntax. It is an input contract, not a shell command to execute directly. The runtime will interpret it according to the defined grammar and execute the corresponding actions.

Alternatively the command can be provided as natural language text, for example:

  • "Show me my unread feeds."
  • "Add this RSS feed: <url> with the name 'Tech News'."
  • "Export all my feeds to a JSON file named 'feeds.json'." in which case the runtime will attempt to parse it into a valid command.

The runtime must:

  1. Parse using RUNTIME/COMMAND_GRAMMAR.md.
  2. Execute transitions in RUNTIME/STATE_MACHINE.md.
  3. Execute primitive actions from RUNTIME/EXECUTION_PRIMITIVES.md.
  4. Maintain runtime data in RUNTIME/RUNTIME_CONTEXT.md.
  5. Emit errors from RUNTIME/ERRORS.md and exit via RUNTIME/EXIT_CODES.md.

First-run behavior: missing config and missing store are valid. The runtime must continue with defaults, bootstrap the store path, and initialize empty store state. When is_first_run=true, the runtime should offer optional onboarding. If the user accepts, guide these 3 starter commands in order:

  1. config show
  2. feed add <url> [-n name]
  3. read --unread --limit 20

Execution Workflow

  1. Normalize aliases and parse argv.
  2. Resolve effective configuration and environment values.
  3. Dispatch to the canonical command family.
  4. Execute primitive actions and record side effects.
  5. Format output according to output-mode precedence.
  6. Exit with mapped code.

Hard Rules

  1. Preserve command semantics from the baseline behavior (strict mode).
  2. Do not silently skip failure branches.
  3. Log every mutating side effect in side_effect_log.
  4. Treat missing dependencies (network, local analysis capability, permissions) as guarded failure transitions.
  5. Do not claim behavior not covered in the documented runtime contracts and examples.
  6. Do not invoke another RSS command binary to execute RSS commands.
  7. Do not use "wrap original binary" fallback for in-scope commands.
  8. If a primitive dependency is unavailable, transition to explicit failure state (manual handoff only for out-of-scope conditions).

References

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.