surveys — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited surveys (Agent Skill) and scored it 92/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 2 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 2 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Help the founder use surveys as a quantitative validation layer — to confirm patterns found in interviews, prioritize among competing hypotheses, or reach people they can't interview one-on-one.
Surveys are powerful when they're targeted, short, and posted somewhere with the right audience. They're much weaker when deployed before any qualitative discovery, when the distribution channel is vague, or when the questions fish for validation instead of testing assumptions. Part of this skill's job is to help the founder assess fit before building anything.
This skill covers two modes:
Read startup/core.md to load project context.
Check prior qualitative work: scan startup/hypotheses/ and startup/interviews/. It is sometimes ok to go for a survey without doing interviews, especially if the user knows and/or has access to a good channel to distribute the survey to aquire diverse responses. Procceeding to survey without hypotheses is not recommended. If no hypotheses present, invoke the hypotheses skill and strongly suggest the user to talk about them first.
Scaffold the surveys folder if it doesn't exist yet:
mkdir -p startup/surveysUse judgment when assessing whether surveys make sense right now. This is advisory — never a gate.
Good signals:
Caution signals — raise these, don't block:
If the founder insists on a survey despite caution signals, help them build a good one. The goal is to inform, not gatekeep.
Share a brief applicability assessment (2–3 sentences): what's looking solid, what the risk is, whether this is a good moment for surveys. Then ask if they'd like to proceed.
If proceeding, load the reference file:
.claude/skills/surveys/references/initial-survey-questions.mdThe reference file's instructions take over from this point.
Load and read the relevant files for context. Infer intent from the conversation — don't mechanically ask "what do you want to do?" If the founder is:
initial-survey-questions.md .claude/skills/surveys/references/tally-survey.mdsurvey-analyst subagent) is coming in a later version of this skill.status: archived, propose the change, get confirmation, write backWhen adding or updating surveys, follow these file conventions:
File location: startup/surveys/{YYYY-MM-DD}-{short-descriptor}.md
Slug convention: lowercase, replace spaces and non-alphanumeric characters with hyphens, collapse multiples. "Invoice chasing validation" → invoice-chasing-validation.
Frontmatter:
---
status: draft|ready|active|closed|archived
mode: questions-only|tally
date_created: YYYY-MM-DD
target_persona: One-line segment descriptor
hypothesis_slugs:
- slug-one
- slug-two
response_goal: 30
tally_form_id: abc123 # tally mode only — omit if questions-only
tally_url: https://tally.so/r/abc123 # tally mode only — omit if questions-only
---Required sections:
## Purpose — why this survey exists and what decisions it will inform## Target Audience — who should fill it out, response goal, and why that sample size## Distribution Plan — where to post it and what value is offered to respondents## Questions — numbered list with question type annotatedOptional sections:
## Notes — founder comments, links to related interviews, context for later## Results — populated after fetching Tally dataRead before writing, propose before saving, get confirmation.
Briefly confirm: "Saved to startup/surveys/{slug}.md."
Mention natural next steps without pushing:
mode: questions-only — when ready, the agent can deploy to Tally and return a shareable linkmode: tally and status: active — remind the founder they can ask "how's the survey going?" at any time~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.