engram-memory-cli

Recall, save, and curate durable project memory through the Engram CLI.

1|Updated Jul 5, 2026
One-click install
npx skills add https://github.com/yersonargotev/packy --skill engram-memory-cli-yersonargotev
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: engram-memory-cli
Source: https://github.com/yersonargotev/packy/tree/main/bundle/skills/engram-memory-cli
Command: npx skills add https://github.com/yersonargotev/packy --skill engram-memory-cli-yersonargotev

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Long-running agent work loses context between sessions, forcing repeated rediscovery of prior decisions, configurations, and known failures. This Skill provides a disciplined protocol for recalling and preserving durable project knowledge with the Engram CLI so history-dependent work starts informed. ## Core Features & Use Cases - Bounded Recall: Search project memory with narrow queries, fetch full memory content with pagination, and pull chronological session context when continuity matters. - Terminal Memory Commit: Finalize each root turn by recording saved, skipped, or needs_review dispositions with preflight duplicate detection and idempotent replay. - Curation Workflows: Edit, review, pin, relate, diagnose, and merge project memories through the commands documented in references/curation.md. - Use Case: Before modifying a release pipeline, recall prior deployment decisions for the exact project, complete the work, then commit a checkpoint memory capturing what changed and why for future sessions. ## Quick Start Ask the agent to search Engram project memory for prior decisions about the current project before starting the task, and to record a checkpoint memory when the work is finished.

Frequently Asked Questions about engram-memory-cli

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search project memory with the Engram CLI?▼

Run engram search with a narrow query, --project, --scope project, --match-mode all, --limit 5, and --json. The response returns up to five candidate summaries within 4 KiB, and you can fetch full content with engram get using the recall_id and result_id.

How do I save a memory from an agent session?▼

Use engram checkpoint record with the host, session ID, root turn ID, a disposition of saved, and repeatable --memory-json values. Run engram checkpoint preflight first to detect duplicates, and use engram save only for explicit curation or loss-risk handoffs.

What happens when the Engram CLI is unavailable?▼

The protocol is best-effort: for tasks unrelated to Engram, continue the primary work without memory after reporting one warning with structured diagnostics. For tasks about Engram itself, keep the CLI failure as task evidence and diagnose it within scope.

Can I use Engram memory across multiple projects?▼

Cross-project recall requires the --all-projects flag and is reserved for explicitly cross-project requests. Automatic recall and writes require a strong or explicit project detection; weak git-based detections must never be promoted to authority.

When should memories be skipped instead of saved?▼

Record a skipped disposition with reason no_durable_knowledge when the disposition rubric finds no durable result after the work settles. Routine self-contained tasks need no search or save, and the record result confirms completion.