agentsclimarketplace
Sign in

Using ninea

Skill gopact-ai/9a/internal/builtin/skills/using-ninea

Use when an agent needs to connect, find, run, inspect, or disconnect capabilities managed by NineA.From its SKILL.md

Install
npx -y skills add gopact-ai/9a --skill using-ninea

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

2 things to look at

  • 29 days oldThe repository was created 29 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 2 stars2 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.

SKILL.md

2.6 KB, 591 tokens by cl100k_base, as published. Nobody here has run it

Using NineA

NineA is a local capability runtime. Search before loading details, and run a capability only when the user's task requires an upstream action.

Find and run

9a search "what the user needs" --json
9a search <integration>/<capability> --json
9a run <integration>/<capability> --input '<json>' --json

Use --input @file for an existing request file or pipe one JSON value to stdin. As an agent, use --json for every search and run; the structured error fields are part of the safety contract.

Reading this Skill, running search, and inspecting a contract have no upstream side effects. run is the execution boundary.

If NineA reports approval_required, explain the proposed upstream action and ask the user to approve it. Preserve data.approvalToken and the exact input. Only after explicit approval, retry that unchanged input with --approve <approvalToken>. Never use a token from earlier or implied approval. The token is single-use, expires after 10 minutes, and is invalidated by a daemon restart. If it expires, was already used, or the input or capability changes, run a new preflight and ask again.

On a run error, inspect data.sideEffect. none means no upstream request was sent, so follow nextAction. possible means the outcome is unknown: never retry automatically, inspect upstream state when possible, and tell the user what may have happened. retryable never overrides these rules.

Connect an HTTP API

On a fresh workspace where this Skill has not been projected yet, the same embedded contract is available through 9a connect --guide http --json. Read references/manifest.md, create an integration manifest, then run:

9a connect <manifest.yaml>

Treat .9a/integrations/<name>.yaml as the source of truth. The generated .agents/skills/using-ninea gateway is disposable and must not be edited. After connecting, run the integration search --json command printed by NineA, select a capability, then inspect its exact <integration>/<capability> contract before constructing input.

For local MCP servers and remote A2A agents, use the shortcuts in references/integrations.md.

Remove runtime access

9a disconnect <integration>

This keeps the source manifest. Never place credential values in manifests, prompts, command arguments, or generated files. 9a secret values are scoped to the current workspace.

Read references/troubleshooting.md when a command fails. Stay within the seven public command groups and follow the next action printed by NineA.

What ships with it: 4 files

6.8 KB alongside SKILL.md

agents/

references/

Keep looking

Skills are one crate of 326,790. 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.