Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/zircote/git-notes-memory/capturegit clone --depth 1 https://github.com/zircote/git-notes-memoryWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00020 | $0.01599 |
| Opus 5 | $0.00010 | $0.00800 |
| Sonnet 5 | $0.00004 | $0.00320 |
| Haiku 4.5 | $0.00002 | $0.00160 |
Grade A, and why
capture scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 201 lines — stays where its author put it; the contents beside it link to each section on GitHub.
<help_check>
Help Check
If $ARGUMENTS contains --help or -h:
Output this help and HALT (do not proceed further):
<help_output>
CAPTURE(1) User Commands CAPTURE(1)
NAME
capture - Capture a memory (decision, learning, blocker, progress...
SYNOPSIS
/memory:capture [namespace] <summary> -- <content>
DESCRIPTION
Capture a memory (decision, learning, blocker, progress, etc.) to the git-backed memory system
OPTIONS
--help, -h Show this help message
EXAMPLES
/memory:capture
/memory:capture <namespace>
/memory:capture --help
SEE ALSO
/memory:* for related commands
CAPTURE(1)
</help_output>
After outputting help, HALT immediately. Do not proceed with command execution. </help_check>
/memory:capture - Capture a Memory
Capture information to the git-backed memory system for later retrieval.
Your Task
You will help the user capture a memory. The memory will be stored as a git note and indexed for semantic search.
Arguments format: $ARGUMENTS
Parse the arguments:
- If first word matches a known namespace, use it as the namespace
- Look for
--separator: text before is summary, text after is content - If no
--separator, use the entire text as both summary (truncated) and content
Valid namespaces: decisions, learnings, blockers, progress, reviews, patterns, retrospective, inception, elicitation, research
If no namespace specified, auto-detect based on content patterns:
- Contains "decided", "chose", "will use" →
decisions - Contains "learned", "discovered", "TIL", "found out" →
learnings - Contains "blocked", "stuck", "cannot", "impediment" →
blockers - Contains "completed", "finished", "milestone" →
progress - Contains "pattern", "recurring", "often" →
patterns - Default →
learnings
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 201 lines · 20 tokens per session scan A 1e127cf54779
capture is a command published in the GitHub repository zircote/git-notes-memory (4 stars, last pushed 8mo ago), licensed MIT. It adds 20 tokens to every session and 1,599 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
grill-me
Run an interactive one-question-at-a-time oral drill for research defence or active-recall study, escalating around weak answers and ending with a study sheet. Use when the user asks to be grilled, quizzed, or prepared for a viva, job talk, seminar, or exam. Not for a written critique; use $devils-advocate or a review…
brief-compliance-check
Check a LaTeX coursework submission against the requirements in a supplied PDF assessment brief. Use when verifying format, required sections, word limits, or deliverables before submission. Not for general prose proofreading; use $proofread.
init-project-course
Bootstrap a university course or module folder with the canonical teaching structure, guidance, and starter files. Use when creating or migrating a course workspace. Not for checking an existing course without modifying it; use $audit-project-course.
dashboard
Launch the learning dashboard web UI to view and edit plans, progress, and spaced repetition data.
evolve
Cluster hardened instincts into a proposed Command / Skill / Agent (Instinct Engine).
learn
Start learning a new topic — asks clarifying questions, researches resources, and creates a structured learning plan.