Baselines

--baseline compares a scan against a prior --json report — not the lockfile — and reports what is new and what is resolved since then.

npx toolprint scan --json > baseline.json
# …later…
npx toolprint scan --baseline baseline.json

Since baseline (baseline.json): 1 new finding, 0 resolved.
  NEW  Instruction-override phrase in tool "helper"

Findings are matched by their stable id, so a report written by an older toolprint still compares correctly.

Baseline vs lockfile

They answer different questions, and you generally want both.

	Lockfile	Baseline
Compares	Capability definitions	Findings
Answers	"Did this server change?"	"Did our security posture change?"
Catches	A silent rug-pull	A newly-introduced injection or leak
Gates by default	Yes (high)	No — informational

--fail-on-new

On its own, --baseline never changes the exit code. --fail-on-new makes it gate — but only on findings that are new since the baseline:

npx toolprint scan --baseline baseline.json --fail-on-new

This is how you adopt toolprint on a repo that is not clean yet. Pre-existing findings are still reported in full; they just do not fail the build, so you can block regressions today and work through the backlog on your own schedule.

Passed --fail-on high: 1 finding at or above high but not new, not enforced (exit 0).

The outcome line distinguishes "at or above the gate but not new" from "below the gate" and from "suppressed", so a passing run never misrepresents why it passed.

--fail-on-new requires --baseline; using it alone is an error rather than a silent no-op.

Adopting the pattern

1. Get a baseline and commit it:

   npx toolprint scan --json > baseline.json
   git add baseline.json
   

2. Gate on new findings only:

   - uses: jestatsio/toolprint@v1
     with:
       config: ./.vscode/mcp.json
       baseline: baseline.json
       fail-on-new: true
   

3. As you fix things, refresh the baseline so resolved findings cannot silently come back.

Baseline vs suppressions

Use a baseline for a bulk backlog you intend to work through. Use suppressions for a specific finding you have reviewed and decided is fine permanently — each one carries a written reason and an optional expiry.

A baseline is a schedule; a suppression is a decision.