Oi: Context and Workflow Loader

VerifiedSafe

Use Oi to load Contexts and Workflows from the connected MCP server, enabling sticky context sessions and multi-step workflows.

Sby Skills Guide Bot
ProductivityIntermediate
7/22/2026
Claude CodeCursorWindsurf
#mcp#context-management#workflows#tool-routing#prompt-library

Recommended for

Automated summary

This skill wires the agent to the Oi MCP server so it can load reusable Contexts and Workflows into the current thread, including as a sticky session.

Strengths

  • Handles selector routing precisely: bare ids, ctx/context(s) and wf/workflow(s) prefixes, multi-select with + and versions via @2/@v2
  • Separates browsing (listing Contexts or Workflows) from running them, and falls back to oi_recommend only when the selector is unknown or unresolvable
  • Manages guardrails through oi_guardrails_confirm using the returned requestId, and never fabricates an id when it is missing
  • Scopes usage reporting to metadata and token counts keyed on usageEventId, explicitly forbidding storage or resending of prompt text

Limitations

  • Fully dependent on the Oi MCP server and its tools being available and connected
  • Usage reporting only works when the host environment (e.g. Cursor) exposes runtime and token accounting
  • Resolving a bare id to a Context or Workflow is delegated to Oi, so the agent may pick the other entity type than the user assumed
When to use it

Use it whenever the user explicitly invokes Oi, names a Context or Workflow, or asks for sticky Context or Workflow state in the current thread.

When not to use it

Avoid it for ordinary coding requests with no stated intent to use Oi, or when the Oi MCP server is not connected.

Security analysis

Safe
Quality score78/100

The skill only orchestrates calls to an Oi MCP server (list/use/start_session/recommend/guardrails/usage). There are no shell commands, no filesystem writes, no obfuscated payloads, and no instructions to exfiltrate secrets; it even explicitly prohibits storing or resending prompt text and forbids inventing requestIds. The only notable surfaces are automated guardrail confirmation on user approval and telemetry-style usage reporting, both of which are legitimate and gated on the user's intent. Nothing destructive or covert is present.

Findings
  • •Automates guardrail confirmation (plain 'yes'/'continue' triggers oi_guardrails_confirm) — user-initiated but relies on the agent interpreting approval correctly
  • •Usage reporting sends metadata/token counts to the remote MCP server, though it explicitly forbids storing or resending prompt text
How this analysis is made

Examples

Load a named context
oi use the backend-conventions context to review this diff and flag anything that breaks our API style.
Run a multi-step workflow
Use Oi workflow release-checklist@v2 to walk through the release steps for version 1.4.0.
Start a sticky session
Start a sticky Oi session with ctx security-review+compliance so every answer in this thread follows those guardrails.

name: oi description: Use Oi to load Contexts and Workflows from the connected MCP server. Use when the user asks for Oi directly, names a Context or Workflow, wants sticky Oi context in the current thread, or uses oi as a trigger.

Oi

When to use

  • When the user says oi ... or use Oi ...
  • When the user names one or more Oi Contexts
  • When the user names an Oi Workflow or asks for a multi-step workflow
  • When the user wants Oi Context or Workflow context loaded into the current thread
  • When the user wants to list available Contexts or Workflows before choosing one
  • When an Oi guardrail response asks for confirmation before continuing
  • When an Oi tool result returns a usage event and Cursor has runtime/token usage to report

Selector routing

  • Pass Context selectors unchanged as contextId and Workflow selectors unchanged as workflowId.
  • Preserve + for multiple Contexts and @2 or @v2 for versions on both selector types.
  • Bare id: when the user gives an installed id without a type prefix, pass it unchanged and let Oi resolve whether it is a Context or Workflow. Use the resolved type to choose oi_contexts_* vs oi_workflows_*.
  • Explicit Context: when the user prefixes with ctx, context, or contexts, strip the prefix and treat the remainder as a Context selector.
  • Explicit Workflow: when the user prefixes with wf, workflow, or workflows, strip the prefix and treat the remainder as a Workflow selector.
  • The reserved Context type selectors ctx, context, and contexts are not Context ids. The reserved Workflow type selectors wf, workflow, and workflows are not Workflow ids. When the user says only one of those words with no id, treat it as a request to list or browse that entity type.

Instructions

  1. Treat Oi and oi as the same plugin surface when plain-text routing is available.
  2. If the user wants to browse available Contexts, call oi_contexts_list.
  3. If the user wants to browse available Workflows, call oi_workflows_list.
  4. If the user names a Context and provides a task, call oi_contexts_use.
  5. If the user names a Workflow and provides a task, call oi_workflows_use.
  6. Apply Selector routing for contextId, workflowId, and sticky sessions.
  7. If the user wants a sticky Context or Workflow for the current thread, call oi_contexts_start_session or oi_workflows_start_session using the routed selector.
  8. Only use oi_recommend when the user does not know which Context or Workflow to use or the provided selector cannot be resolved. It considers both Contexts and Workflows.
  9. Prefer Oi MCP operations over treating the request as plain English when the user clearly intends to use Oi.
  10. If an Oi response returns a guardrail confirmation request, call oi_guardrails_confirm when the user clearly approves continuing. Use the provided requestId; set remember only when the user says not to ask again.
  11. If an Oi response returns usageEventId, call oi_usage_report after the run when Cursor has runtime/token accounting. Pass usageEventId as authoritative, include contextId or workflowId only for clarity, and never store or resend prompt text.

Guardrails

  • Guardrail confirmation is agnostic to Contexts and Workflows. Use oi_guardrails_confirm with the requestId returned by Oi.
  • A plain yes, continue, or equivalent user approval means continue once.
  • Phrases like yes and don't ask again, remember this, or always allow this Context mean set remember: true.
  • Do not invent a requestId; if it is missing, ask the user to rerun the blocked Oi request.

Usage reporting

  • Usage reporting is agnostic to Contexts and Workflows. Use oi_usage_report for both.
  • Prefer usageEventId from the prior Oi tool response. Optional contextId or workflowId are secondary hints.
  • Report only metadata and token counts Cursor actually knows, such as runtime, provider, model, status, latencyMs, inputTokens, cachedInputTokens, cacheWriteInputTokens, outputTokens, reasoningTokens, compressionInputTokens, and compressionOutputTokens.
  • If Cursor has no runtime/token accounting, skip usage reporting instead of guessing.

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/oi-context-and-workflow-loader-mrwakbx1/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/oi-context-and-workflow-loader-mrwakbx1/download' --output '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1/SKILL.md.part' && mv '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1/SKILL.md.part' '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.

Claude Code
test -f '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1/SKILL.md' && test -s '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/oi' && cp -R '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1' '.claude/skills/oi'

Cursor
test -f '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1/SKILL.md' && test -s '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1/SKILL.md' && mkdir -p '.cursor/skills' && test ! -e '.cursor/skills/oi' && cp -R '.skill-downloads/oi-context-and-workflow-loader-mrwakbx1' '.cursor/skills/oi'

OpenSkills

OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.

Official OpenSkills installation
openskills install './.skill-downloads/oi-context-and-workflow-loader-mrwakbx1'

openskills read 'oi-context-and-workflow-loader-mrwakbx1'

Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.

For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.

openskills sync

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/oi-context-and-workflow-loader-mrwakbx1/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.

Related skills