memory-onboard

memory-onboard is a skill for Claude Code, Codex from tmj-90/gaffer. It costs 196 tokens per session (3,603 once invoked), scanned A, original, Apache-2.0.

A repository onboarding process for shared project memory, called lore. It reads project materials and proposes draft records about conventions, decisions, migrations, and other facts for a person to review.

In plain words
What is it for?
Use it with `/memory-onboard` or when setting up memory for a repository. It can inspect the README, architecture decisions, recent commits, and deprecation markers.
Why use it?
It creates repo-specific guidance from evidence instead of filling memory with generic or unverified notes.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one. Also seen: reads .claude/ paths; mentions CLAUDE.md; mentions AGENTS.md.

Good fit Use it with /memory-onboard or when setting up memory for a repository. It can inspect the README, architecture decisions, recent commits, and deprecation markers.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/tmj-90/gaffer/memory-onboard
Install

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.

Any agent
npx skills add tmj-90/gaffer --skill memory-onboard
Clone the repo
git clone --depth 1 https://github.com/tmj-90/gaffer

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for memory-onboard

README.md
[![agentmods](https://agentmods.dev/badge/skills/tmj-90/gaffer/memory-onboard/github.svg)](https://agentmods.dev/skills/tmj-90/gaffer/memory-onboard)
Your own site
<a href="https://agentmods.dev/skills/tmj-90/gaffer/memory-onboard"><img src="https://agentmods.dev/badge/skills/tmj-90/gaffer/memory-onboard/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for memory-onboard

Your own site · 80×15
<a href="https://agentmods.dev/skills/tmj-90/gaffer/memory-onboard"><img src="https://agentmods.dev/badge/skills/tmj-90/gaffer/memory-onboard.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 196 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,603 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00196 $0.03603
Opus 5.5 $0.00078 $0.01441
Sonnet 5.5 $0.00039 $0.00721
Haiku 4.5 $0.00020 $0.00360

Measured 1mo ago against content hash 6851503caf34, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

memory-onboard 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 1mo ago.

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.

packages/memory/skills/memory-onboard/SKILL.md · 300 lines

How it starts

The opening of the file, as written. The whole thing — 300 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/memory-onboard — repo-aware onboarding for memory

This skill is the way to seed a repo with lore. It earns its keep by reading the repo first (README, ADRs, recent commits, deprecation markers, in-flight migrations) so the proposed drafts are specific to this codebase, with source citations — rather than inventing memory or mechanically chunking every bullet in the docs (which produces mostly noise and floods the review queue).

Output: DRAFT records with source citations, plus optional absence markers and boundary edges. Everything lands in memory review for a human to ratify — the skill never promotes its own records.

Hard rules

  1. Every lore record this skill produces is a DRAFT. Call suggest_lore (or report_conflict for counter-claims — also drafts), never any approval path. The human reviews with memory review. Exception: record_absence does NOT create a draft — it records a self-expiring "we checked, no policy" marker. Different primitive, different gate (see rule 6).
  2. Every draft carries the tag induction so it can be filtered later. Layer topic-specific extras as appropriate (security, conventions, migrations, incident-lessons, invariants, etc.). report_conflict counter-drafts are auto-tagged conflict-report by the server.
  3. Cite sources in the body. When a draft comes from a file or commit, include a one-line reference at the top of the body — e.g. Source: README.md L42–55 or Source: commit abc1234 "migrate accounts → organisations". This is the trust signal a reviewer needs.
  4. Never put secrets, credentials, personal data, regulated data, or anything that looks like a token into any lore record. If a candidate item contains something like that, skip it and tell the user why.
  5. Skip the obvious. "We use TypeScript", "run the tests before committing", and other things a model already knows about a typical codebase belong in CLAUDE.md (or nowhere), not in lore. Aim for non-obvious, high-consequence knowledge.
  6. Use record_absence sparingly. If the user explicitly confirms "we don't have a policy on X yet, and that's intentional", that's a legitimate absence marker — call record_absence({ query: "<topic>", reason: "<why no policy>", repo: <repo> }). Don't infer absence from "I didn't find anything in the README"; the user has to confirm it. Markers self-expire (14 days default) so they fade automatically if the team takes a stance later.
  7. Use report_conflict when you find a contradiction, not when you find a duplicate. If search_lore returns an existing record and the source-of-truth you just read disagrees with it (e.g. README says "Argon2id", existing lore says "scrypt"), that's a report_conflict moment. If the existing record just covers the same topic without contradicting, leave it alone — possibleConflicts in search results handles benign overlap.

Read the full file on GitHub · 300 lines

Changes

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.

  1. 1mo ago First seen · 300 lines · 196 tokens per session scan A 6851503caf34

Subscribe to this mod's changes

memory-onboard is a skill published in the GitHub repository tmj-90/gaffer (2 stars, last pushed 20d ago), licensed Apache-2.0. It adds 196 tokens to every session and 3,603 once invoked, about $0.0008 per session on Opus 5.5. 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.