Amm save pending

Skill frankenmason/amm-skills/amm-save-pending

Selective curated long-term memory skill bundle for agentmemory MCP. Works with Claude Code, Codex, Cursor, Gemini CLI, Copilot, Antigravity.

Install

npx -y skills add frankenmason/amm-skills --skill amm-save-pending

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Persist an unresolved / deferred item to long-term memory with a priority level (P0/P1/P2), so it is not forgotten across session boundaries. Trigger when the user says "on hold", "pending", "later", "deferred", or when the agent detects an unfinished task that cannot be completed this turn.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

3.0 KB, 630 tokens by cl100k_base, as published. Nobody here has run it

AMM — Save Pending Item

Purpose

Multi-session projects accumulate deferred items. Without an explicit pending list, they silently fall off the agent's radar. This skill records each one with a priority and resolution condition, so amm-recall can surface them at session start.

When to Trigger

Keywords: "pending", "on hold", "later", "deferred", "TODO for next time".
User offers multiple options and moves on without selecting.
Agent cannot complete a task this turn (waiting for input, blocking dependency).

Apply Conditions

The item has a clear description (what + why deferred).
A priority can be assigned (P0 = critical / P1 = important / P2 = nice-to-have).
Expected resolution within ~14 days (longer horizons may not belong in AMM — use a project tracker instead).

Execute

memory_save({
  content:  "PENDING (#{{SESSION_ID}}, P<0|1|2>): <what is deferred>.
             Reason: <why not now>.
             Resolution condition: <criterion for 'done'>.
             Context: <1-line session context for future self>",
  type:     "workflow",
  concepts: "pending, P<0|1|2>, <topic>, {{SESSION_ID}}",
  files:    "<comma-separated absolute paths or empty>"
})

Optional TTL

Tag a TTL in concepts as ttl_<YYYY-MM-DD>. With AGENTMEMORY_TOOLS=all, a governance routine can purge expired pendings via memory_governance_delete. Without it, purge manually.

Warnings

Do NOT save items that can be resolved this turn.
Do NOT save permanently abandoned items — use type: fact with a cancelled concept instead.
When a pending is resolved, save a follow-up entry with the same concepts and a resolved tag for traceability.

Example

User: "let's put the caching optimization on hold until we have real traffic data."

memory_save({
  content:  "PENDING (#gamma-5, P1): caching optimization on hold.
             Reason: need real traffic data to pick the right layer.
             Resolution condition: 1 week of prod traffic logs ingested.
             Context: weighing LRU vs write-through on /api/users.",
  type:     "workflow",
  concepts: "pending, P1, caching, performance, ttl_{{EXPIRY_DATE}}, gamma-5",
  files:    "/workspace/src/cache/strategy.ts"
})

amm — umbrella (Bootstrap, references, license)
amm-save-branch — when a branch creates pendings
amm-save-resume — the pending list is part of resume

Change Log

1.0.0 — Initial release.