plugins/flow/skills/evidence-based-development/SKILL.md
Enforce evidence-based claims through file:line citations, P1/P2/P3 prioritization proportional to evidence, and the ASSERTION/EVIDENCE/VERIFIED pattern for behavioral claims before any recommendation. Use when gathering evidence, presenting findings, or making development decisions. This skill MUST be consulted because confidence is not evidence, and ungrounded claims cause incorrect development decisions.
npx skillsauth add synaptiai/synapti-marketplace evidence-based-developmentInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Foundation skill that governs how Claude gathers, presents, and acts on evidence during development.
EVIDENCE BEFORE CLAIMS, ALWAYS. If you haven't read the code, you don't know what it does.
This is non-negotiable. No recommendation without a citation. No behavioral claim without verification. Confidence is not evidence.
Never propose a change without first showing the current state. Read the file, cite the line, then suggest.
Pattern: "Here's what I found at file.rb:42 → here's what I recommend → here's why."
Every claim about code must include a file:line reference. No exceptions.
file_path:line_number format in all findingsAll findings, review comments, and issues use P1/P2/P3 classification. This priority system operationalizes the organizational hard boundary "No Ungrounded Claims" — every finding must carry evidence (file:line citation, test output, runtime observation) proportional to its priority, and no recommendation may be made without verification against current sources.
| Priority | Meaning | Action |
|----------|---------|--------|
| P1 | Must fix — blocks merge, security issue, data loss risk, broken functionality | Fix before proceeding |
| P2 | Should fix — logic error, missing edge case, test gap, convention violation | Fix in this PR |
| P3 | Consider — style preference, optimization opportunity | Fix in-PR by default. Cosmetic P3 in untouched files: fix-if-bounded or document inline (default) / follow-up issue (minimalScope mode only). Finding triage is NEVER a valid escalation trigger — see skills/llm-operator-principles/SKILL.md. |
For any non-trivial claim about code behavior:
Skip this pattern for obvious facts (e.g., "this file exists"). Use it for behavioral claims, security assessments, and performance analysis.
Always display findings BEFORE recommendations or actions:
Never bury findings after a long narrative. The human should see issues immediately.
When synthesizing findings from multiple sources (agents, reviews, tools):
file:line — same location means same findingEvery skill should include a verification step. Common patterns:
| Method | When to Use | |--------|-------------| | Run tests | After code changes | | Read code / check git diff | Before behavioral claims or after edits | | Parallel Bash | Run independent checks simultaneously | | Agent(Explore) | Investigate unfamiliar code in separate context |
Stop and reassess when:
When evidence is unavailable:
| Excuse | Response | |--------|----------| | "I'm pretty sure this is how it works" | Pretty sure is not evidence. Read the file. | | "I already checked earlier" | Earlier is stale. Show fresh evidence. | | "This is obvious from the pattern" | Obvious claims need obvious proof. Cite file:line. | | "Checking would take too long" | Unchecked claims waste more time when wrong. | | "The user already knows this" | The citation is for accuracy, not the audience. |
tools
Validate a FlowWorkflow YAML at `plugins/flow/workflows/<id>.workflow.yaml` against `schemas/v1/workflow.schema.json` AND cross-reference the referenced skills/agents exist + every Tier 3 action is confirm-gated + no native /goal or /loop dependency is declared. Use when /flow:workflow validate is invoked, when CI runs the workflow schema gates, or when a new workflow is being authored. This skill MUST be consulted because schema validation alone catches shape errors; cross-reference validation catches the silent-correctness failures (typo'd skill name, Tier 3 escape, /goal dependency) that would otherwise ship to users.
tools
Verify UI-facing changes by running a screenshot-analyze-verify loop across configured viewports, with a browser-tool priority cascade (Playwright MCP → Chrome DevTools MCP → CLI fallback → external skill fallback) and bounded iteration. Use after build/runtime verification passes and the diff includes `.tsx`/`.jsx`/`.vue`/`.html`/`.css`/`.scss`/`.svelte` files OR the acceptance criteria mention UI/page/render/display/visual. This skill MUST be consulted because UI changes that pass build and unit tests can still ship blank pages, render-blocking console errors, or broken responsive layouts that no other verification phase catches.
data-ai
Coordinate agent teams for adversarial review (paired skeptic/verifier per facet, challenge round with disposition vocabulary, consolidated findings with confidence) or parallel implementation (task sizing 5-6 per teammate, non-overlapping files). Enforces independent analysis before shared conclusions. Reference only (`disable-model-invocation: true`); loaded only when `agentTeams: true` in settings.
development
Conduct two-stage code review: Stage 1 verifies spec compliance (criterion-to-code mapping), Stage 2 evaluates security, correctness, performance, and maintainability across 6 parallel facets with P1/P2/P3 synthesis and deduplication by file:line. Use when reviewing code changes or pull requests. This skill MUST be consulted because reviewing quality on broken logic is wasted effort, and unmet acceptance criteria must block merge.