Taddy API skills

Agent skills for building on Taddy API.

Skill	What it covers
`taddy-podcast-api`	Podcast and episode lookup, full-text search, latest-episode feeds, top charts, episode transcripts, webhooks. Bundles the GraphQL schema and the podcast docs.

Install

Ask your AI Agent to: Download the Taddy API skill from https://github.com/taddyorg/taddy-api-skills into your project's skills folder

Architecture

data.json                 name, version, description, author, logo, categories, marketplace: the source for every manifest below
plugin.json               Agent Plugins manifest (agent-plugins.org), read by Cursor and Codex
assets/taddy-logo.png     square 1024x1024 logo referenced by the Claude, Codex and Cursor manifests
.claude-plugin/           Claude Code plugin + marketplace manifests
.cursor-plugin/           Cursor plugin manifest
.agents/plugins/          Codex marketplace manifest
scripts/
└── sync-manifests.py     regenerates every manifest from data.json (--check to verify)
skills/taddy-podcast-api/
├── SKILL.md              triage, three workflows, decision table, gotchas, routing
├── references/
│   ├── pagination.md     limits and stop conditions for every list query
│   ├── errors-and-quota.md
│   ├── graphql-client-libraries.md   plain HTTP client vs typed GraphQL client, per language
│   ├── skill-feedback.md     when and how to report a skill/API mismatch or a blocked use case to Taddy
│   ├── schema.graphql    podcast-only SDL for the agent to read (codegen introspects api.taddy.org instead)
│   └── *.md              the podcast docs, one page per file
└── scripts/
    ├── taddy_client.js   minimal client, Node 20+
    ├── taddy_client.py   minimal client, Python 3.9+
    ├── taddy_throttled.js   one request then a 250 ms pause, for loops and polling, Node 20+
    ├── taddy_throttled.py   same, Python 3.9+
    ├── taddy_client_typed.ts   graphql-request transport for generated documents
    ├── codegen.ts        GraphQL Code Generator config (introspects api.taddy.org)
    ├── graphql/          starter fragments/ and queries/ for the typed setup, copied into the project
    ├── taddy_feedback.js   validates and scrubs a feedback report, posts it only with --send, Node 20.19+
    └── webhook-express.js

Releasing a new version

Skill versions track the bundled schema. The schema changes rarely, and a skill release is the only time it is updated. Before tagging a release:

Copy the docs mirror into references/ and refresh the schema:
```
curl -sSfL https://ax0.taddy.org/docs/taddy-podcast-api-schema.graphql -o skills/taddy-podcast-api/references/schema.graphql
```
Both files are podcast-only exports of the Taddy docs (the full docs and schema, covering every medium, are taddy-api-docs.md and schema.graphql under the same path).
Run python3 scripts/sync-manifests.py --check, plus evals/check-refs.sh if you have the local evals/ folder (it is not in git).
If you have the local evals/ folder, run the eval loop with the skill-creator skill in Claude Code, using evals/evals.json.
Bump version in data.json and in SKILL.md, run python3 scripts/sync-manifests.py to regenerate the manifests, and add a CHANGELOG.md entry. Never edit the generated .json manifests by hand.

Distribution: installs via npx skills add are what list the skill on skills.sh (nothing to submit). Listing on the public Cursor Marketplace is a separate submission at cursor.com/marketplace/publish.

Announce skill releases alongside the API changelog.

License

MIT