Orchestrates lightweight feature implementation via a single subagent with a build-test smoke gate.
npx skills add https://github.com/azure-id/orc --skill orc-mini-azure-id
Please help me install this Agent Skill. Skill: orc-mini Source: https://github.com/azure-id/orc/tree/main/templates/skills/orc-mini Command: npx skills add https://github.com/azure-id/orc --skill orc-mini-azure-id
--json flag to a CLI command. Invoke the mini lane, answer four intake questions, and it plans, dispatches one executor, runs the test suite as a gate, optionally authors extra tests, and ships the commit. ## Quick Start Ask the assistant to use orc-mini to implement a small, well-scoped change such as adding a flag to an existing CLI command.High-intent search queries and answers about installing and using this skill.
Invoke the orc-mini lane with a scoped request such as adding a flag to one command. It runs a light intake, dispatches one Sonnet 5 high-effort executor, runs a build+test smoke gate, and ships the change without full review phases.
orc-mini skips full code review, verification, and summary phases, uses one executor instead of scored parallel waves, and replaces the scoring table with a one-line complexity read. It adds a smoke gate that blocks shipping when build or tests fail.
Avoid it for complex, high-risk work with many interdependencies, core or shared surfaces, or security-sensitive changes. The complexity read recommends switching to the full flow, and you can switch mid-run since both share the same run folder format.
Yes, the Phase M smoke gate runs build and tests read-only after execution. A red result blocks any commit offer, triggers one repair re-dispatch, and a second failure stops the run with an honest report.
Yes, after a green smoke gate it optionally dispatches a test-authoring subagent that writes automated tests, a TEST-PLAN.md, and curl bundles for HTTP APIs into a test-generator folder. It never runs those tests itself.
abzhawRun verification commands and require explicit confirmation before claiming completion.
mvillmowFetch GitHub PR review comments, implement changes, and reply to each comment.
mem0aiSearches and injects relevant memories into context before starting a task.
loomantixOrchestrate multi-lane adversarial code review for high-risk pull requests.