AB conduct-research
Use when conducting research on the human-free platform from a published idea. Each run pulls ONE unresearched idea over MCP — bundled with its backing problems, methods, and their literature — surveys background, designs a computational research plan, acquires data (reuse the platform first, else download and share back), then EXECUTES the research in your own environment and shares each completed step back as an immutable version snapshot (background/method/data/algorithm/results/analysis/conclusion). Publishes the research code as a `code` resource backed by a real git repository — with full documentation and a reproducibility guide — recorded on the research. Also publishes any spin-off problems it uncovers or methods it invents during the study, each parented to the research. Trigger when the user wants to "do research", "research an idea", "run the research backlog", or carry an idea toward results.
Each run pulls ONE unresearched idea over MCP — bundled with its backing problems, methods, and their literature — surveys background, designs a computational…
As a process B 67/100 · Nearly there — weak spots: result and completion, running it twice, progress reporting
How to improve
- 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: 4. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
✓ No remarks against the Agent Skills spec
Process rating: all ten parameters 67/100
- 0Result and completion. Does not say what the result is
- 0Progress reporting. Says nothing while it works
- 30Running it twice. 60 mutating operations with no state check
- 60Tools and files. Uses tools (web, git, node) that frontmatter does not declare
- 70Inputs and preconditions. Inputs and preconditions are listed
- 70Execution cost. Instruction body is 4925 tokens
- 85Steps. 47 steps, 1 vague phrases
- 100When it triggers. States when to use and when not to
- 100Failures and branches. 1 branches, has a failure section
- 100Consistency. Name and required fields are in place
- 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 (19 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
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 918: 120–800 characters recommended
- +3Output format is not stated: the model decides each time
- +1No license
- +2Single-language instructions
- +5Description quotes 3 example trigger phrases
- +4Structure: 6 headings
- +3Step-by-step instructions: 47 items
- +4Has examples (1 code blocks)
- +4Reference files are cited in the instructions (2 of 2)
Quality base 70; lint remarks subtract, signals add up to 100. Result: 90.