Route broad or ambiguous AgentKit SEO work to the right module while keeping context scoped.
SkillScore100Maintained
Automatic
◆Marketing & Content pack
Last verified 8/18/2026
What it changes
Passed 22 of 22 test cases with the skill, and 9 of 22 without it.
Without skill41%
With skill100%
Checks it passed that the model alone missed
Routes the task to agentkit-seo-github.
Routes the task to agentkit-seo-x-twitter.
Routes the task to agentkit-seo-web-portfolio.
and 11 more in the full report
13 fixed9 passed either way0 still failing0 broke
Pass-rate lift
+59 pp
Tokens
+18%
Agent turns
0%
Response time
-39%
Model gemini-3.6-flashMeasured 8/18/2026v122 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 13 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 13
Prompt
If a user request regarding professional presence is broad or ambiguous, which reference file must be consulted before routing?
What was checked, without and with the skill
Checked
Without skill
With skill
Consults references/module-routing.md for broad or ambiguous requests.
Missed
Passed
case-16 · gemini-3.6-flash · 8/18/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md70 linessingle file
AgentKit SEO
Overview
Use this skill as the orchestrator for the whole repository. Its main job is to select the right module skill, avoid loading irrelevant platform rules, and sequence cross-platform work in a sane order.
Routing workflow
Identify the target surface from the request.
Load only the matching module skill unless the user explicitly asks for a cross-platform pass.
If the request spans multiple surfaces, start with agentkit-seo-agent-context-optimization so the factual source of truth is stable before editing platform outputs.
If the request is only about the skill system itself, read the references in this skill before changing provider adapters or install instructions.
For broad requests with no clear surface:
Active applications or job-description tailoring: route to agentkit-seo-cv-ats.
Recruiter discovery or profile search: route to agentkit-seo-linkedin.
Proof-of-work, repositories, or developer credibility: route to agentkit-seo-github or agentkit-seo-web-portfolio, based on the supplied asset.
Audience building, posting strategy, or public conversation loops: route to agentkit-seo-x-twitter.
Conflicting, scattered, or cross-platform facts: route to agentkit-seo-agent-context-optimization first.
Token discipline
Route to one module by default.
Load the agent context file before platform references only when facts, consistency, or cross-surface rewriting matter.
Prefer public URL inspection, local search, or a compact pasted section over asking the user to dump every asset into the prompt.
Summarize inspected inputs and ask for the smallest missing input set.
Do not expand into algorithm explanation unless the user asks why.
Intake workflow
If the user already has an agent context file, ask for or use its explicit path before rewriting platform assets.
If the task spans multiple surfaces, or the user's facts are scattered, recommend creating or repairing the agent context file first.
Do not block a narrow one-off edit on a full context file when the supplied material is already enough.
For public URLs, fetch or inspect public material when tools allow it and cite which source was used.
For private or login-gated surfaces, ask the user for pasted section text, screenshots, exports, or a local text file instead of guessing.
If critical facts are missing, ask only for the minimum extra inputs needed to proceed.