Recall ships its command surface as Agent Skills — one canonical SKILL.md per skill under agent-skills/. Claude Code and omp receive per-file links in their skill homes; Pi discovers the canonicals directly through Recall’s native package manifest. Codex receives generated native adapters in hosts/plugins/recall/skills/; Grok and JCode can load the same canonical skill bodies through their host setup. Byte-identical skill text is not treated as behavioral equivalence across hosts. In Claude Code, invoke skills as /do-recall-<name>; the model can also trigger most of them itself when the conversation calls for it.
Migrating from
/Recall:*slash commands? The old namespaced slash commands were retired in favor of these skills (issue #228) — same bodies, one namespace across all hosts.install.sh/update.shremove the stale~/.claude/commands/Recall/symlinks automatically. See Upgrading.
Available Skills
do-recall-dump
Flush the current session to the database and capture a Library of Alexandria entry. Use this at the end of every session.
Usage: /do-recall-dump Session Title Here
What it does:
- Imports the current session’s messages into SQLite
- Creates a Curated LoA entry via the fabric Extractor (or a basic summary with
--skip-fabric) - Makes everything immediately searchable
This skill is user-invoked only — dumping a session is always your call, never the model’s. Claude uses disable-model-invocation: true; the generated Codex adapter uses agents/openai.yaml with allow_implicit_invocation: false.
do-recall-search
FTS5 full-text search across all memory tables.
Usage: /do-recall-search kubernetes auth
Searches messages, LoA entries, decisions, learnings, and breadcrumbs. The skill wraps recall search, so human CLI flags work here too:
/do-recall-search database choice -t decisions— hard-filter to decisions only/do-recall-search database choice --bias-type decisions— prefer decisions first, while still returning matching learnings/messages/LoA/breadcrumbs/do-recall-search database choice --show-provenance— show Record Provenance for every result (by default only unknown provenance is flagged)
Rule of thumb: use -t when you want only one table; use --bias-type when you want one table first without hiding other context.
do-recall-recent
Show recent memory records.
Usage:
/do-recall-recent— all tables/do-recall-recent decisions— specific table
do-recall-scout
Memory-first orientation for an unfamiliar codebase. Searches Recall before touching git history or source, then produces a structured scout report in chat.
Usage:
/do-recall-scout— scout the whole repo/do-recall-scout auth— narrow the scout to a subsystem, path, or question
The report covers: memory summary (what Recall already knew), repo map, tech stack (declared versions from manifests; lockfiles excluded), key paths, decisions & learnings (the memory tied to each key file), conventions (declared + observed + Recall-learned house style), tests, risks, and next steps. It enforces a strict sensitive-data boundary — no secrets, generated/vendored files, the live database, or cross-project memory — and is chat-only by default. Reports are persisted only when the repo endorses it, and then only to .agents/atlas/artifacts/ (never .agents/atlas/handoffs/). Scout grounds its repo map, key paths, and risks in CodeGraph (primary) via the external codegraph CLI: a codegraph status --json probe, a cheap orientation bundle, and at most two narrow explore calls — on an unindexed repo scout offers codegraph init (runs it only if you say yes), with a grep/tree-walk backstop. CodeGraph is optional; scout works fully without one.
do-recall-stats
Database statistics — record counts, size, and table breakdown.
Usage: /do-recall-stats
do-recall-add
Add a structured memory record — breadcrumb, decision, or learning.
Usage: /do-recall-add
Claude will prompt for record type and details.
do-recall-doctor
Health check all subsystems — database, MCP registration, hooks, extraction, embeddings.
Usage: /do-recall-doctor
Run this first when troubleshooting any issue.
do-recall-update
Check whether a newer Recall release is available on GitHub.
Usage: /do-recall-update
Check-only — prints the current version, the latest release tag, a short excerpt of the release notes, and the exact cd <path> && ./packaging/update.sh recipe. It never runs update.sh inline because rebuilding the recall binary mid-session can corrupt in-flight hook invocations. Exit the coding agent, run the recipe, and restart.
Rate limit: GitHub’s anonymous API is 60 requests/hour per IP. On throttle, the skill falls back to pointing at https://github.com/edheltzel/Recall/releases.
do-recall-loa
Browse and view Library of Alexandria entries.
Usage: /do-recall-loa
Installation
Skill names use the canonical do-recall-* namespace. Claude and Codex load them from their native plugins; Pi loads them from the native package; omp loads them from ~/.omp/agent/skills/do-recall-*. Installer per-file links remain for hosts without a plugin attach. Canonicals live under ~/.agents/Recall/shared/skills/<name>/. Source files are in agent-skills/ in the Recall repository.
The Codex plugin adapters are generated with bun run build:codex-plugin and packaged with the native plugin bundle documented in Codex Integration. Claude copies are generated with bun run build:claude-plugin; see Claude Integration.
If installer-linked skills are missing after an update, re-run ./packaging/install.sh to relink them. See omp Integration for its skill and extension ownership. Claude, Codex, and Pi prefer their native attachments.