BC blog-topic-research
Stop writing blog posts nobody searches for. This skill builds your editorial backlog from real, verifiable user demand - never from AI vibes. It mines candidates from Google Suggest, People Also Ask, Reddit, Stack Overflow, GitHub issues, vendor forums, and changelogs; captures every signal as a citable URL with verbatim evidence; classifies each topic by post format (how-to-fix, x-vs-y, listicle, migration, release-recap, ...); checks against your existing backlog so you don't cannibalize what you already published; and outputs a writer-ready scaffold with primary sources, problem summary, confirmed fixes, version context, and FAQ variants. Built for content marketers, founders, indie hackers, and dev-tool teams who want a long-tail SEO pipeline backed by evidence instead of hallucinated keyword volumes. Trigger when the user says: 'research blog topics', 'find topics with real demand', 'expand the editorial backlog', 'research N long-tail topics', or any variant of growing a content pipeline with verified candidates.
As a process C 60/100 · Has gaps — weak spots: result and completion, inputs and preconditions, running it twice
How to improve
- Shorten the description to 1024 characters.
- The SKILL.md body is over 5,000 tokens: move reference detail into references/ and load it when needed.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 2. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- error
description-longdescription is 1035 chars, limit 1024 - warning
body-longSKILL.md body ≈ 7751 tokens (recommended < 5000); move details to references/ - note
frontmatter-keyunknown frontmatter key "emoji" - note
frontmatter-keyunknown frontmatter key "homepage"
Process rating: all ten parameters 60/100
- 0Result and completion. Does not say what the result is
- 0Inputs and preconditions. Does not say what the process needs to start
- 30Running it twice. 32 mutating operations with no state check
- 60Tools and files. Uses tools (bash, web, python, node) that frontmatter does not declare
- 70When it triggers. States when to use, but not when not to
- 70Execution cost. Instruction body is 7751 tokens
- 100Steps. 88 steps
- 100Failures and branches. 37 branches, has a failure section
- 100Consistency. Name and required fields are in place
- 100Progress reporting. Reports progress
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low The response is described with custom markup (60 tags): a typed call is more reliable
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 1035: 120–800 characters recommended
- +3Output format is not stated: the model decides each time
- +1No license
- +2Single-language instructions
- +4Structure: 20 headings
- +3Step-by-step instructions: 88 items
- +4Has examples (7 code blocks)
Quality base 70; lint remarks subtract, signals add up to 100. Result: 44.