Dr. RAGnos proof lab

One scan that leaves evidence, fixes, and a reviewer handoff.

Dr. RAGnos turns a messy codebase pass into terminal output, structured JSON, vaulted-secret notes, and an agent-ready next step. The proof viewer is the first artifact, not a product shelf.

Scan walkthrough Request scan walkthrough

Next: terminal, report.json, payload.json, reviewer handoff

Sample scan proof viewer

See what one command hands back.

The useful part is not just detection. It is a ranked fix list, structured proof, and reviewer handoff that another operator or agent can use immediately.

dr-ragnos scan --proof

--- terminal / Run scan

$ dr-ragnos scan ./workspace --proof

target redacted-service/ 284 files, 5 scanners

scanners secrets, deps, sast, tests, dead-code

--- Finding / Normalize evidence

before grade D-minus 41 known CVEs, 16 exposed secrets

evidence scanner-output/redacted/* linked to every finding

Finding 9 blockers promoted for operator review

--- Fix list / Generate fixes

after grade A 0 known CVEs, 16 secrets vaulted

Fix list dependency cleanup, secret vaulting, reviewer checks

elapsed 5 hours from first scan to reviewer-ready packet

--- report.json / Release scorecard

wrote .dr-ragnos/report.json

{ "grade_before": "D-minus", "grade_after": "A", "known_cves_remaining": 0 }

{ "critical_findings_closed": 9, "secrets_vaulted": 16 }

--- payload.json / Agent payload

wrote .dr-ragnos/payload.json

{ "owned_paths": ["package.json", "src/auth"], "ready_for_review": true }

{ "constraints": ["vaulted secrets only", "no prod service changes"] }

--- reviewer handoff / Human signoff

Agent handoff .dr-ragnos/reviewer-handoff.md

includes diff summary, reproduction command, evidence links, remaining risk

handoff ready reviewer can approve without re-running the whole scan

terminal

Run scan

Start with one local proof command against a redacted workspace.

Finding

Normalize evidence

Collapse scanner noise into release-facing findings with evidence attached.

Fix list

Generate fixes

Produce the patch order and agent-safe constraints before handoff.

report.json

Release scorecard

Write the structured proof the human reviewer can inspect.

payload.json

Agent payload

Package bounded next steps for another agent without leaking local code.

reviewer handoff

Human signoff

End with a reviewer packet that preserves the diff, evidence, and remaining risk.

This sample is anonymized, but the handoff shape is the point: every terminal line stays connected to evidence, machine-readable context, and a reviewer-friendly narrative.

Anonymized proof band

A messy release became a shippable handoff.

The sample viewer above mirrors a real cleanup pattern: diagnose the risk, vault what should never have lived in code, remove the known exposure, and leave a reviewer-ready trail.

Release grade

D-minus to A

Risk score rebuilt from scanner findings, dependency fixes, and reviewer signoff.

Known CVEs

41 to 0

Dependency exposure collapsed into a clean release gate.

Secrets vaulted

0 to 16

Hardcoded credentials moved into the vault with rotation notes attached.

Time to handoff

Open ended to 5 hours

Report, payload, and reviewer notes were ready in the same working session.

RAGnos universe

The work band around Dr. RAGnos.

RAGnos spans proof-first audits, agent-native implementation, editorial operating memory, and the loops that keep systems from drifting back.

Proof labs

Run focused scans, close the visible risk, and leave an evidence packet that another operator can trust.

Agent-Native Systems

Build workflows where agents receive scoped payloads, constraints, and reviewer context instead of vague prompts.

Editorial field notes

Turn the proof trail into readable operating memory without letting narrative replace the underlying artifacts.

Loops and drift control

Keep the release surface watched after the first cleanup pass so regressions become routed work, not ambient worry.

Field Notes proof

The writeup stays attached to the artifact trail.

Field Notes are not generic thought pieces. They are edited proof records: what was found, what changed, and what the next operator should trust.

Evidence linked Reviewer ready

From D-minus to A without losing the audit trail

A Field Note can cite the terminal pass, the report JSON, and the reviewer handoff without copying raw logs into the story. The result is editorial enough to read and concrete enough to act on.

What changed in the system

Which evidence moved the decision

What an agent or reviewer should do next

Loops

RAGnos Loops

Start with Dr. RAGnos now. Add the monitoring layer when ongoing drift matters more than one cleanup pass.

Monitor

Watch the code and process changes that matter before release review.

Diagnose

Turn fresh drift into an explainable issue instead of another Slack thread.

Route

Let automation handle routine fixes and keep the human decisions obvious.

Request a scan walkthrough

Share the system context you can safely describe. We will use the first conversation to map whether Dr. RAGnos is a fit.