agents-remember

The agent lifecycles: one lifecycle per agent type, under one roof. Routes every session by exactly three conditions (spawn-role env -> fresh role brief -> otherwise free-chat launcher), carries the minimal lifecycle frame (the six lifecycle signals every session shares), and houses the self-contained per-role lifecycles (architect, orchestrator, designer, strategist, manager, worker, curator, system-specialist, adversarial reviewer) plus the report-template library and the reviewer criteria catalogs. Solo work is the degenerate portfolio. Supersedes and replaces both l-01-session-job-lifecycle and l-02-agent-orchestration.

After the harness starter package and MCP server are wired and the harness has restarted, run/verify runtime_install, set up the memory repo, bootstrap onboarding when scaffolding new, and configure providers to index the code and memory.

Create, attach to, report on, integrate, finalize, and clean up Agents Remember worktree-backed tasks while preserving delegated/human approval gates and external-memory compatibility.

Close out approved Agents Remember edits by preserving the applicable approval authority, missing-onboarding checks, external-memory onboarding refresh, memory quality, ledger alignment, and no automatic push for worktree-backed tasks.

Create and maintain file-level onboarding and repo entity catalogs. Enforces strict 1-to-1 source mapping, governing overview links, metadata, and references, and routes package/module/source-slice create, refresh, move, or delete cases to the c-03-repo-bootstrap skill.

In-between task workflow for work that needs a durable task file, distinct implementation examples, and an approval gate but still fits a single-page implementation plan as a rule of thumb.

Choose retrieval strategies across memory substrates: semantics for known concepts with unknown structure, relationships for known anchors with unknown connections, and intent for hidden contracts and code truths.

Capture confirmed findings, route them to the right durable task-local path, and propagate factual current-state clarifications into onboarding immediately when allowed.

Bootstrap onboarding for undocumented repos or existing memory slices. Builds root overviews, route-local overview pillars, evidence packs, file cards, onboarding waves, deleted-slice cleanup, curator reviews, and handoff while keeping the orchestrator thin.

Resolve the active Agents Remember context for a target repository, including topology, coordination root, memory root, settings, storage, pathRules, worktree contract facts, ledger path, and branch-gated cross-repo allowances.

Review the agents-remember cockpit dashboard like a user: discover the intended workflow scenarios, detect missing views and where users get stuck, and judge whether the UI is self-explanatory at a glance / leads the user / progressively discloses / has clear hierarchy / is a good observability surface. Conducts the installed design, accessibility, data-viz, and motion skills and owns only the workflow-completeness layer none of them cover. Findings only — never auto-fixes.

Carry richer onboarding memory from a source branch into official memory only when the corresponding code has landed on the official branch.

Initialize or repair the Agents Remember memory root for a target repository. Defaults to repo-local internal memory; creates an external memory repo only when the developer explicitly asks for external memory.

Control Agents Remember memory quality with task-start drift checks, pre-code-commit missing-onboarding checks, and closeout memory quality gates.

Adopt existing external-memory onboarding as the first ledgered memory baseline after resolving context, checking drift, and requiring explicit acceptance when onboarding is not proven current.