| name | auditing-devforgeai |
| description | Perform, resume, refresh, validate, or synthesize the governed DevForgeAI enterprise audit across Prompts 0–8 with immutable provenance. Use for framework skills, hooks, commands, agents, process gates, Claude/Codex parity, and Codex-native behavior. |
Auditing DevForgeAI
Follow references/audit-protocol.md completely. Use the bundled Prompt Suite,
templates, and audit-bundle script for the same nine-phase contract as Codex.
After Prompts 0–8, run audit-content-validate against the exact managed report
directory and audited snapshot. Do not finalize an untouched template,
placeholder, incomplete Appendix B record, incomplete phase map, or synthesis
whose Prompt 0–7 hashes do not match.
For durable intake, call audit-import-completed, then
audit-findings-build. Never mutate archive content to correct a malformed
source; import a new managed copy with correction provenance. Return only the
validated run ID to remediation.
HALT on missing CLI, exit 127, malformed JSON, snapshot drift, or unsupported
Confirmed evidence. Audit execution never edits remediation status or closure.
