Skill orientation

Use this panel to select and sequence the skill. The canonical source follows below.

Purpose: Install repository-specific controls, adapters, and coding-agent entry points.
Category and sequence: Lifecycle; sequence 1
Primary use: Commission A Repository, Refresh A Commissioned Project Pack, Install Valdris Front Doors
Required gates: Skill Registry, Provenance, Neutrality, Privacy
Conditional gates: commissioned-portability before a generated pack is handed off
Red Zone triggers: No skill-specific triggers are declared
Next route: Route and Control the Goal
Source identifier: skills/valdris-commission/SKILL.md

Canonical pathskills/valdris-commission/SKILL.mdRevision69bab1cInspect source

Valdris Commission

This skill owns the Commissioning system. Its job is to make a repository ready to use the Valdris repository-level assurance harness. It does not route or implement product work.

Deterministic flow

Confirm the target repository and keep all project-specific facts in its generated adapter. Do not add customer, company, provider, topology, credential, or private incident facts to the public Valdris core.
Inspect the target's existing AGENTS.md, CLAUDE.md, source-of-truth files, branch policy, validation commands, deployment boundaries, risk paths, and operator authority. Reuse stable answers from an existing commissioned pack.
For a fresh install, run the canonical commission command from the Valdris source:

npm run commission -- --repo <target> --project-name "<name>" --out <target>/.valdris-harness --yes

To refresh an existing pack, first verify that .valdris-harness is a reviewed, generated Valdris pack and preserve its stable commissioned answers. Then run the explicit replacement form:

npm run commission -- --repo <target> --project-name "<name>" --out <target>/.valdris-harness --yes --force

For a recognized generated pack, --force preserves reviewed commissioning answers and re-detects values that previously came from generated defaults. A reviewed --answers <json> file overlays only intentional changes. Never use --force on an unrecognized directory.
Review .valdris-harness/commissioning-review.md, .valdris-harness/project-adapter.json, and the bounded loaders added to target-root AGENTS.md and CLAUDE.md.
Verify the generated workflow and lifecycle registries, all fifteen generated skills, controls, scripts, trust-store placeholders, and portable paths. Run the commissioned portability verifier and the skill-registry, provenance, neutrality, and privacy gates.
Commission operator-owned review and authority keys outside agent reach. Never generate, trust, or approve an agent-owned key.
Commit the complete .valdris-harness directory and the two bounded root loaders before creating a route. Do not claim a portable or reviewable pack from an uncommitted tree.

Completion criterion

Commissioning is complete only when:

the target contains a reviewed and committed .valdris-harness/project-adapter.json;
.valdris-harness/commissioning-review.md identifies no unresolved blocking answer;
target-root AGENTS.md and CLAUDE.md discover the pack without overwriting unrelated instructions;
.valdris-harness/skills/registry.json, codex-routing.yaml, all 8 work-type skills, and all 7 lifecycle skills pass skill-registry-gate.mjs;
generated Codex and Claude mirrors are byte-identical to the commissioned canonical skills;
portability, provenance, neutrality, and privacy checks pass.

If any item fails, remain in this skill and repair commissioning. Otherwise hand off to $valdris-route-goal.