executing-plans
executing-plans
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available
api-serviceOperate the apod-api Flask service locally or from an API client: start it safely, construct valid APOD queries, interpret responses and errors, and troubleshoot upstream, cache, CORS, and static-asset behavior.
Independent GitHub snapshot · Automated classification · Original documentation governs
Use this sub-skill when the task is to run apod-api, query /v1/apod, choose
date, count, start_date, or end_date, diagnose Flask 400/404 responses,
check the service version, or investigate NASA APOD availability.
The service scrapes the live APOD website; a successful request is therefore a
network-dependent observation, not an offline fixture lookup. concept_tags
requires a credential-bound Alchemy integration and is deliberately documented
as degraded when that credential is absent. Do not put API keys or credentials
in commands, scripts, or skill files.
The distribution declares version 1.1.0 and Python >=3.12; the Flask service
itself returns service_version: "v1". The current project metadata is not
safe to treat as a default editable setuptools install: flat-layout discovery
sees top-level apod, apod_parser, static, templates, and skills as
ambiguous. The workflow documents a non-editable runtime setup instead of
hiding this packaging limitation.
The bundled scripts/run_service.py must pass --help and
--inspect-routes without contacting APOD. Whole-skill integration should
later run the safe application-import and custom-404 candidates, then add
synthetic conflicting-query and upstream-outage cases before claiming native
service coverage.