con un clic
harness-remember
// Manages SSOT memory — decisions.md, patterns.md, and cross-session learnings. Use when recording decisions, searching memory, or promoting learnings.
// Manages SSOT memory — decisions.md, patterns.md, and cross-session learnings. Use when recording decisions, searching memory, or promoting learnings.
Orchestrates the full powerball-harness plugin release with build, validation, version bump, CHANGELOG, tag, and GitHub Release. Use when releasing this plugin.
Plans and tracks tasks in .claude/harness/plans.json. Use when creating plans, adding tasks, updating markers, checking progress, or brainstorming an idea into tasks.
Periodic housekeeping and session lifecycle management. Use when performing cleanup, pruning, or session commands.
Runs iterate-until-pass loops with ralph-worker agents in a worktree. Use when running [ralph]-marked tasks or looping until verify passes.
Initializes and configures Harness in a project — CI, memory, duo workflow, Codex. Use when setting up a new project or adding Harness configuration.
Executes plans.json tasks — solo, parallel, or breezing team mode. Use when implementing tasks or running the work loop.
| name | harness-remember |
| description | Manages SSOT memory — decisions.md, patterns.md, and cross-session learnings. Use when recording decisions, searching memory, or promoting learnings. |
| when_to_use | record decision, search memory, update patterns, SSOT, promote learning, decisions.md |
| allowed-tools | ["Read","Write","Edit","Bash","Grep","mcp__harness__harness_mem_*"] |
| argument-hint | [ssot|sync|sync-across|migrate|merge|search|record] |
| context | fork |
| model | sonnet |
| effort | high |
SSOT (Single Source of Truth) and cross-tool memory management for Harness.
| User Input | Subcommand | Behavior |
|---|---|---|
"Init SSOT" / memory ssot | ssot | Bootstrap .claude/memory/decisions.md + patterns.md (see references/ssot-initialization.md) |
"Save what we learned" / "promote to SSOT" / /harness-remember sync | sync | Pull learnings from Claude Code auto memory (Layer 1) into SSOT (Layer 2) decisions/patterns (see references/sync-ssot-from-memory.md) |
"Sync across agents" / /harness-remember sync-across | sync-across | Reconcile memory artifacts across agent workspaces / project specs (see references/sync-project-specs.md) |
"Migrate from AGENTS.md" / memory migrate | migrate | Run interactive workflow migration (see references/workflow-migration.md) |
"Merge Plans.md" / memory merge | merge | Consolidate multiple Plans.md files (see references/plans-merging.md) |
"Search memory for X" / memory search <term> | search | Keyword/regex search; local first, MCP extends (see references/search.md) |
"Record this decision" / memory record | record | Validate SSOT-worthiness, write local first, MCP mirrors (see references/record.md) |
For
searchandrecord, local SSOT is authoritative and MCP extends reach — see each reference file for the full contract.
Unified Harness Memory (shared DB through harness_mem_* MCP)
Relationship with Claude Code Auto Memory (D22): Harness SSOT (Layer 2) coexists with Claude Code's auto memory (Layer 1). Auto memory records general learnings passively; SSOT explicitly curates project-specific decisions. Use /harness-remember sync when a Layer 1 observation has become important enough to preserve across sessions and contributors.
decisions.md in the project's .claude/memory/ directory (project-root), entry D22: 3-Layer Memory Architecture.