affinda-automation
by ComposioHQaffinda-automation helps agents run Affinda workflows through Composio Rube MCP by searching live tool schemas first, checking the Affinda connection, and executing document automation tasks with less guesswork.
This skill scores 66/100, which makes it acceptable but limited for directory listing. Directory users can understand that it is a Rube MCP-based Affinda automation helper and how an agent should begin tool discovery and connection setup, but they should not expect rich, task-specific Affinda workflows or bundled implementation assets.
- Valid frontmatter clearly declares the MCP dependency on Rube and describes Affinda automation via Composio.
- Prerequisites and setup steps explain how to verify Rube MCP, manage the Affinda connection, and require ACTIVE status before workflows.
- The skill gives a repeatable discovery-first pattern using RUBE_SEARCH_TOOLS and RUBE_MANAGE_CONNECTIONS, which should reduce schema guesswork for agents.
- No support files, scripts, references, or concrete Affinda task examples are included beyond the SKILL.md instructions.
- Execution depends on live Rube MCP tool discovery and an active Affinda connection, so the skill gives limited standalone operational detail.
Overview of affinda-automation skill
What affinda-automation does
The affinda-automation skill helps an AI agent automate Affinda tasks through Composio’s Rube MCP server. It is designed for workflows where the agent must discover current Affinda tool schemas, verify an active Affinda connection, and then execute document-processing actions through the available Rube tools instead of guessing API parameters from memory.
Best fit for Affinda workflow automation
Use this skill if you already use Affinda for resume parsing, document extraction, classification, or related document AI workflows and want Claude or another MCP-capable agent to operate those tasks through Composio. The best-fit reader is someone setting up agentic workflow automation rather than someone looking for a standalone Affinda SDK wrapper.
Key differentiator: search tools first
The main value of the affinda-automation skill is its insistence on calling RUBE_SEARCH_TOOLS before execution. That matters because Composio tool schemas can change, and Affinda operations often require precise field names, document IDs, collection IDs, or file inputs. The skill’s pattern reduces failed calls caused by stale assumptions.
What to check before installing
This skill has a narrow dependency surface: the repository path contains only SKILL.md, with no helper scripts or bundled examples. That is acceptable if your client already supports MCP and you are comfortable driving Rube tools interactively. If you need a turnkey CLI, sample app, or prebuilt Affinda workflow templates, this skill will feel minimal.
How to Use affinda-automation skill
affinda-automation install and MCP setup
Install the skill from the Composio skill collection with your skill manager, for example:
npx skills add ComposioHQ/awesome-claude-skills --skill affinda-automation
Then configure Rube MCP in your AI client by adding the MCP server endpoint:
https://rube.app/mcp
The skill assumes RUBE_SEARCH_TOOLS and RUBE_MANAGE_CONNECTIONS are available. No local scripts are included in the skill folder, so most setup happens in your MCP client and Composio/Rube connection flow, not in the repository checkout.
Required inputs before running a workflow
Before asking the agent to automate Affinda, provide the job context it cannot infer reliably:
- The exact Affinda outcome: parse a resume, upload a document, retrieve extracted data, manage a collection, check document status, or similar.
- The relevant identifiers you already have: workspace, organization, collection, document, candidate, or job IDs.
- File location or upload source if a document is involved.
- Output format: raw JSON, normalized table, summary, validation report, or next-step action.
- Any constraints: do not create new records, only read data, retry failed extraction, redact personal data, or confirm before deletion.
A weak prompt is: “Use Affinda to process this resume.”
A stronger prompt is: “Use the affinda-automation skill. First search Rube tools for the current Affinda schema for uploading and parsing a resume. Check the Affinda connection status. If active, upload /files/candidate-jane.pdf to the resume parsing workflow, wait for or retrieve the parsed result if supported, and return normalized JSON with name, email, phone, skills, education, and work history. Ask before creating any new collection.”
Practical affinda-automation usage flow
A reliable workflow usually looks like this:
- Ask the agent to invoke the skill and search tools for the exact Affinda task.
- Have it call
RUBE_MANAGE_CONNECTIONSfor toolkitaffinda. - If the connection is not active, complete the returned authentication link.
- Re-run the connection check.
- Execute the selected Affinda tool using the schema returned by
RUBE_SEARCH_TOOLS. - Inspect the response and ask the agent to continue with retrieval, transformation, validation, or export.
Do not skip discovery even if you know the Affinda API. The skill is oriented around Composio tool slugs and MCP schemas, which may differ from raw Affinda API documentation.
Files to read first in the repository
Start with composio-skills/affinda-automation/SKILL.md. It contains the actual operational contract: prerequisites, setup, tool discovery, connection checking, workflow sequence, and the warning to search tools first. There are no README.md, rules/, resources/, references/, or scripts/ folders in the current skill path, so the repository review is quick. For deeper product behavior, compare the discovered Rube schemas with the Affinda toolkit documentation at composio.dev/toolkits/affinda.
affinda-automation skill FAQ
Is affinda-automation for Workflow Automation or API coding?
affinda-automation is primarily for Workflow Automation through an MCP-enabled agent. It does not generate a full Affinda integration library, manage infrastructure, or replace an SDK in a production backend. Its strength is letting an agent discover available Affinda actions and run them safely inside a Rube/Composio tool session.
How is this better than an ordinary prompt?
A normal prompt may ask the model to “use Affinda,” but the model might invent parameters or rely on outdated API knowledge. The affinda-automation skill gives the agent a concrete operating pattern: discover tools, check connection, use live schemas, and only then execute. That structure is the main reason to install it.
Do beginners need to know Affinda first?
You do not need to know every Affinda endpoint, but you should understand your business workflow and the document type you are processing. Beginners should start with read-only or low-risk tasks, such as checking available tools or retrieving existing document data, before asking the agent to upload, create, update, or delete records.
When should I not use this skill?
Do not use it if you need offline processing, a non-MCP integration, deterministic batch infrastructure, or extensive local examples. Also avoid it for vague goals like “automate HR” without specifying which Affinda operation should run. The skill works best when the user can name the target action and supply the required IDs, files, and output expectations.
How to Improve affinda-automation skill
Improve prompts with exact Affinda intent
The most common failure mode is an underspecified request. Improve results by naming the operation and the desired end state: “find the correct tool for listing documents in a collection,” “extract parsed resume fields from document ID X,” or “upload this invoice-like document and return extraction confidence fields.” Specificity helps RUBE_SEARCH_TOOLS return a better execution plan.
Provide stronger schema and connection context
When iterating after the first run, paste back the discovered tool slug, required fields, and any error message. For example: “The discovered tool requires collection_id and file; I have collection_id=abc123 and the file path is /tmp/resume.pdf. Retry using those fields only.” This is more effective than asking the agent to “try again” because it narrows the repair path.
Add guardrails for risky operations
If the workflow may create, modify, or delete Affinda resources, add explicit confirmation rules to your prompt. Useful guardrails include: “read-only unless I approve,” “show the planned tool call before execution,” “do not upload duplicate files,” and “mask candidate contact details in the final response.” These instructions improve adoption in teams handling personal or recruitment data.
Iterate on affinda-automation outputs
After the first output, review both the Affinda response and the agent’s transformation. Ask for targeted refinements such as confidence filtering, field normalization, deduplication, missing-field checks, or export formatting. The affinda-automation skill gets you to live tool execution; output quality still depends on clear validation criteria and follow-up instructions.