Implement

Implement

Entry gate

Require:

• approved scope, non-goals and acceptance criteria;
• costly-to-reverse decisions approved or expressly parked;
• build ceiling and write/external authority;
• config/risk-policy.json risk/authority profile: minimum tier, bounded paths,

disclosure, secrets and external-action constraints;

• evaluate need for stochastic/judgement-bearing behaviour.

If missing, use scope; never infer owner decisions.

For active service/safety impact, approved mode: expedited-incident may parallelise containment verification/review. Gates and authority still apply; name a reconciliation run.

Loop

1. For substantial+ work, create the canonical delivery-run from

../deliver/templates/RUN.template.json, set profile software, and follow [run-contract.md](references/run-contract.md). Routine minor work may proceed without RUN.json unless the user or project policy requests one.

1. Keep an adaptive plan. Use tdd for observable change, refactor

for approved behaviour-preserving structure and diagnose for unknown causes. Migrations may also need behaviour tests and equivalence evidence. Use orchestrate when decomposition or independent coverage helps; adapt topology/order inside authority. For a version-sensitive external interface or migration, apply [source grounding](references/source-grounding.md) and [migration compatibility](references/migration-compatibility.md).

1. Run deterministic checks; map each criterion to evidence. When required, run

evaluate and attach its passing receipt.

1. Invoke read-only code-review under the current HARNESS.md risk ladder.

Its adaptive topology and recorded substitution path load-bear. Record adapter, actual family, omissions, and substitutions.

1. Repair blockers, then repeat verification and review. Stop at

the scaled repair budget (guardrail-not-target: 1-2 routine, up to ~5 complex; see references/run-contract.md) or on scope/design drift; return evidence to the user or scope.

1. Update owned docs for behavioural, architectural, operational or decision

change. For substantial+ apply session: refresh the recovery checkpoint, audit context, graduate durable findings and classify retained/ephemeral artifacts in RUN.json.

1. When a receipt exists, validate with

"${AGENTS_HOME:-$HOME/.agents}/skills/deliver/scripts/validate_delivery.py" \ .agent-run//RUN.json --workspace-root "$PWD" --verify-hashes. Hand off only after this machine gate.

1. User final acceptance is mandatory; promotion needs separate release

authority. When a run directory exists, after acceptance, failure or cancellation, terminalise with ${AGENTS_HOME:-$HOME/.agents}/skills/orchestrate/scripts/run_dir_finalize.py; an outer orchestrator's awaiting-user transport remains active and does not rename the canonical receipt state.

Authority and completion

• Reviewed source stays immutable; separately authorised artifacts and Fabric

communication stay bounded.

• Record missing legs as failed/unavailable; never filter them. Review coverage

blocks only when the current HARNESS ladder and its substitution path remain unsatisfied.

• Objective evidence outranks reviewer confidence; adjudicate conclusions,

never vote.

• Substantial+ starts a fresh implementation session bound to approved digests.
• awaiting_acceptance is the successful machine-gate state. Move the

canonical receipt to accepted only after explicit user acceptance.

Adapter-absent path

Without optional Console, Herdr or GitHub, use canonical project artifacts and emit the skill-owned kind in [portable-workflow.v1.json](portable-workflow.v1.json). It records evidence but grants no acceptance or promotion authority.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: mblauberg
• Source: mblauberg/provenant
• License: MIT

Install and usage instructions live in the source repository linked above.