Help us improve
Share bugs, ideas, or general feedback.
From agentkit-seo
Builds and maintains a personal professional source-of-truth context file from CVs, LinkedIn exports, GitHub history, and project summaries for consistent platform outputs.
npx claudepluginhub agentkit-seo/agentkit-seo --plugin agentkit-seoHow this skill is triggered — by the user, by Claude, or both
Slash command
/agentkit-seo:agentkit-seo-agent-context-optimizationThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Work through the lens of a meticulous biographer and fact-checker assembling the user's professional source of truth. Use this skill before any cross-platform optimization pass that depends on a stable factual record. The goal is to turn scattered inputs into one reliable, agent-readable context file.
Builds ATS-optimized resumes for developers and product managers from PDFs/DOCX, LinkedIn PDFs, GitHub profiles, or guided interview.
Generates ATS-optimized CVs from LinkedIn, GitHub, or portfolio, tailored to job descriptions. Outputs paste-ready text with flaw report.
Generates ATS-optimized resumes tailored to job postings from master resumes or experience data, producing .docx files via Python rendering.
Share bugs, ideas, or general feedback.
Work through the lens of a meticulous biographer and fact-checker assembling the user's professional source of truth. Use this skill before any cross-platform optimization pass that depends on a stable factual record. The goal is to turn scattered inputs into one reliable, agent-readable context file.
Normalize the user's facts before writing any LinkedIn, CV, GitHub, web portfolio, or X/Twitter output.
VERIFIED FACTS work, or how to handle context-file failure modes.references/ file and mark wiki-specific guidance as unavailable when it affects confidence.QUICK REFERENCE block first when an existing context file is long.Use the smallest honest context pass:
Quick scan: check whether a context file exists, read QUICK REFERENCE, and identify obvious structural gaps.Default pass: quick scan plus relevant entries for the requested platform, supplied source material, and hard-fact consistency checks.Deep reconciliation: full context file review, all supplied sources, chronology checks, platform conflicts, unsupported claims, and targeted repairs across sections.Default to Default pass for broad context-file work. Offer Deep reconciliation as an optional next step when the current answer would benefit from more evidence. Do not choose Deep reconciliation silently unless the user asks for full normalization, complete validation, or cross-platform reconciliation.
~/.agentkit-seo/<name-surname>-seo-context.md.Before returning, check the draft and fix or flag any failure:
If a check fails and cannot be resolved from the available inputs, say so explicitly instead of smoothing it over.
Once the context file is clean, hand off to exactly one target platform skill unless the user explicitly requests a multi-surface pass.
Return:
For audits or validation passes, use concise labels such as Verified, From context, From source, Inference, and Needs evidence when a claim could otherwise be ambiguous. When the pass is intentionally bounded, include a one-line Depth note that says what sources were not inspected and what deeper reconciliation would add.
Human playbook: hub/agent-context-optimization/README.md.