Process and Runbooks — Index
Operational documentation for running DCRH as a team process, not just a reference repo.
Every command in these documents is verified against the runnable copy in
../../samples/defending-code-reference-harness/.
| Document | Use it when |
|---|---|
| process-operate-dcrh.md | You need the full picture — who does what, in what order, with what handoffs, from threat model through a merged fix |
| runbooks/setup-and-sandbox.md | Preparing a new host to run the autonomous pipelines (one-time per host) |
| runbooks/run-a-detection-scan.md | Scanning a target — static or execution-verified |
| runbooks/patch-and-verify.md ⚠️ | Drafting and grading a fix — the only runbook that edits code |
| process-optimization.md | Improving the process itself — current-state waste, future-state recommendations, implementation plan |
How these relate
process-operate-dcrh.md ── the SOP: RACI, 5-step flow, exceptions, metrics
│
├── Step 1 (threat model) ── no dedicated runbook, single command in the SOP
├── Step 2 (scan) ── runbooks/run-a-detection-scan.md
│ └── needs ────── runbooks/setup-and-sandbox.md (pipeline path only)
├── Step 3 (triage) ── no dedicated runbook, single command in the SOP
├── Step 4 (patch) ── runbooks/patch-and-verify.md ⚠️
└── Step 5 (act) ── human review, no runbook
process-optimization.md ── separate lens on the same workflow: where is it wasteful,
what would make it faster/safer, in what order to fix it
Source of truth
These documents describe the process for operating the code in
samples/defending-code-reference-harness/, a runnable copy of
anthropics/defending-code-reference-harness
kept alongside this documentation set so every command here can be checked against real
source rather than recalled from a README. See samples/README.md for how to refresh that
copy.
Related: Architecture (HLD) · Low-Level Design ·
Defending Code Harness Vault/DCRH/DCRH — Process and Runbooks.md.