pulseengine-claude

Plugin pulseengine-claude from marketplace pulseengine-eu in pulseengine/pulseengine.eu. Commands read from the repo's manifest — today.

Skills (22)

• bootstrap-verification

  Use to stand up verification scaffolding for a NEW or not-yet-built piece (greenfield repo or component) so it is traceable from commit one ("bootstrap this", "set this up properly from the start").

• capture-session-learnings

  Use at the end of substantive work or before compaction to distill durable knowledge into the working-context checkpoint and memory ("remember this", "capture what we learned"). Also the monthly skill-usage measurement.

• ci-status

  Use before opening a pull request and before waiting on CI ("is CI green yet", "merge when green"), and when the disk-pressure hook says the disk is low. Checks queue admission, polls the CI status API (falling back to gh), and frees disk by asking the named holders.

• claim-verification

  Use before a claim is reported or written, such as findings, audit results, PR or issue text, "0 found", "all green", "because", "verify this", an AI review used as evidence, or a README, badge or doc asserting what the system proves.

• delegate-by-cost

  Use BEFORE EVERY Agent (Task) call to choose `model:`, and when a delegated tier fails the same check twice. Mechanical work with an exact check goes to haiku.

• factory-notice

  Use when an unattended agent hits a blocker, a fork it would ask about, something shipped or the end of a session that the owner must know about ("notify me", "ask me later"). Posts a notice and does not wait.

• gate-potency

  Use to check that checks and evidence can fail and are about the right system ("audit our gates", "can CI actually fail", vacuous or skipped required checks, or an emulator, staging or self-test result filed as "on-target" or "in production" evidence).

• issue-hunt

  Use to run an incremental bug/feature hunt over a repo's issue tracker, handling only what is NEW since the last pass ("hunt the issues", "loop through the issues", a standing loop).

• oracle-gate-a-change

  Use when proposing, landing or evaluating a consequential change on a PulseEngine project ("is this safe to land", "what's the gate", "before merging this"), and before claiming a property holds.

• outside-in-recheck

  Use when the user shares outside material (talk, paper, upstream spec change, peer design doc) or says "recheck us against this". Turn it into checkable questions about our own repos instead of summarizing.

• proof-synthesis

  Use to write, repair or strengthen a machine-checked proof, spec or invariant (Verus, Rocq/Coq, Lean 4, Dafny, Kani, scry), or when a proof obligation fails and needs a generate, verify, refine loop.

• publication-review

  Use before a blog post or other public, claim-heavy prose goes live (ready = true, draft = false, "is this post ready", "fact-check the draft").

• pulseengine-feature-loop

  Use for a feature end-to-end on a PulseEngine project that touches more than one file ("implement a feature", "add a new requirement", "ship it with traceability").

• refactor-decision

  Use when deciding whether to stop patching and refactor, or when asked to "refactor X" or "tidy first". It does NOT apply in pulseengine/rivet or pulseengine/spar before 2027-02-01.

• release-artifact-pipeline

  Use to set up, audit or change a release artifact pipeline (release.yml, cosign signing, SBOM, SLSA, crates.io or npm publish, sigil, witness or scry gates, the Pages dashboard).

• release-execution

  Use to cut, ship or finish a release ("ship it", "cut a release", "tag this", "work the PR queue to green") and for autonomous release campaigns.

• release-planning

  Use to plan releases in rivet (what goes in v0.X, triaging an incoming issue toward a release, "is v0.X ready to cut") and to run the issue-driven delivery loop.

• repo-hygiene

  Use for a hygiene sweep after a release or on a cadence (stale worktrees, dead branches, old stashes, the open-issue board; "clean up", "what's stale").

• report-tool-friction

  Use the moment a PulseEngine tool errors, gives wrong output, lacks a capability or forces a workaround. File the issue in the tool's own repo.

• seam-contract

  Use at a seam between two repos, before building against another repo's interface, changing something another repo binds to, or at release-execution step 4c. It writes a seam block in the owning issue and checks it.

• stpa-audit

  Use to conduct or audit an STPA or STPA-Sec hazard analysis ("do an STPA", "are our losses, hazards and UCAs complete") or before a release whose V-model gate depends on it.

• traceability-audit

  Use to check the rivet trace is complete and bidirectional across the V for the target safety standard ("are all artifacts linked"), while authoring and as the blocking check before tagging.

Hooks (15)

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/inject-pulseengine-memory.sh

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/inject-situational-awareness.sh

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/inject-rivet-context.sh

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/persist-shell-options.sh

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/signing-probe.sh

• SessionStart

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/factory-notice.sh start

• PreToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/agent-model-reminder.sh

• PreToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/commit-merge-guard.sh

• PreToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/disk-pressure.sh

• PreToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/private-host-guard.sh

• PreCompact

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/save-working-context.sh

• SessionEnd

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/save-working-context.sh

• SessionEnd

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/factory-notice.sh end

• PostToolUseFailure

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/loop-guard.sh

• PostToolUse

  bash ${CLAUDE_PLUGIN_ROOT}/hooks/loop-guard.sh

Similar plugins