rpi
Run one bounded Plan, Implement, and fresh Validate experiment, then report and stop. Triggers: "run rpi", "feed this through the loop", "research-plan-implement".
How do I install this agent skill?
npx skills add https://github.com/boshu2/agentops --skill rpiIs this agent skill safe to install?
- Gen Agent Trust Hubwarn
The skill acts as a development workflow orchestrator that automates discovery, implementation, and validation phases. It requires high autonomy and executes local shell commands and scripts to manage the environment and project state. The primary security considerations involve the use of scripts that perform file system operations (including directory removal) and instructions that explicitly direct the agent to bypass human-in-the-loop confirmation steps.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
- Runlayerwarn
8/12 files flagged
- ZeroLeakspass
Score: 93/100 · 2 sections analyzed
What does this agent skill do?
RPI
Run one experiment through three responsibilities and stop:
Plan -> Implement -> fresh Validate -> report
RPI preserves the original intent and dispatches each core phase at most once. It does not own retries, budgets, queues, claims, leases, Git, delivery, release, closure, or the caller's next decision.
The pure scripts/run_once.py reference behavior makes
the dispatch and stop semantics executable without Git, ao, or a tracker.
Contract
- Invoke
/planonce with the caller's intent. Preserve its exactPlanPacketand digest. If planning cannot produce a complete packet, reportNOT_PLANNEDand stop. - Invoke
/implementonce with that packet. It performs one bounded experiment and returns aCandidatePacket. If no candidate is built, reportNOT_BUILTand stop. - Invoke
/validateonce in a context distinct from the candidate'sauthor_context_id. Pass only the PlanPacket, CandidatePacket, factual evidence, validator identity, and freshness attestation. - Return the durable
verdict.v2reference and a short report. Stop regardless ofPASS,FAIL, orNOT_PROVEN.
NOT_PLANNED and NOT_BUILT are report statuses, never semantic verdicts.
A caller may later create a revision-packet.v1 and start a new invocation;
RPI never creates, selects, or consumes a revision automatically.
Invariants
- Acceptance and its digest do not change between phases.
- The candidate reports complete changed-path coverage or Validate returns
NOT_PROVEN. - A proven change outside
write_scopemakes the verdictFAIL. - PASS requires nonempty distinct author and validator context IDs plus an explicit freshness attestation.
- Optional Premortem, Postmortem, Council, genie, factory, tracker, and runtime adapters are caller-selected. They do not alter phase order or core outcomes.
- Learn is an optional later consumer of verdict collections and is not part of this invocation.
Report
Return exactly the useful boundary facts:
schema_version: rpi-report.v1
status: PASS | FAIL | NOT_PROVEN | NOT_PLANNED | NOT_BUILT
plan_packet_digest: <sha256 or null>
subject_manifest_digest: <sha256 or null>
verdict_ref: <path or null>
verdict_digest: <sha256 or null>
checked: []
not_checked: []
Do not append a next action. The caller owns continuation.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/boshu2/agentops/rpi">View rpi on skillZs</a>