psi

psi is a skill for Claude Code from Soul-Brews-Studio/arra-oracle-skills-cli. It costs 125 tokens per session (5,860 once invoked), scanned C, original, MIT.

A tool for connecting a code repository’s memory folder to another repository’s shared knowledge store using a symbolic link. A symbolic link is a filesystem shortcut that points to another directory.

In plain words
What is it for?
Use it to check, link, repair, or remove the connection between a repository and its caretaker knowledge store, while verifying ignore and tracking settings.
Why use it?
It lets an ordinary code repository use shared project memory without storing that memory in its public source history.

Skill for Claude Code

Written for Claude Code: argument-hint in frontmatter.

Needs its repository: it reads a path above its own folder, which exists only inside the repository. The line is symlink ../../laris-co/neo-oracle/ψ/family/github.com/soul-brews-studio/arra-oracle-skills-cli.

Part of the oracle-skills plugin — 23 skills shipped together

Good fit Use it to check, link, repair, or remove the connection between a repository and its caretaker knowledge store, while verifying ignore and tracking settings.

Compare 6 skills from other repositories ↓
Install

Getting it into your agent

It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.

Clone the repo
git clone --depth 1 https://github.com/Soul-Brews-Studio/arra-oracle-skills-cli
agentmods
npx agentmods add skills/soul-brews-studio/arra-oracle-skills-cli/psi

Made for: Claude Code.

Or install oracle-skills, the plugin that ships this one along with the rest of its 23 skills.

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 psi

README.md
[![agentmods](https://agentmods.dev/badge/skills/soul-brews-studio/arra-oracle-skills-cli/psi/github.svg)](https://agentmods.dev/skills/soul-brews-studio/arra-oracle-skills-cli/psi)
Your own site
<a href="https://agentmods.dev/skills/soul-brews-studio/arra-oracle-skills-cli/psi"><img src="https://agentmods.dev/badge/skills/soul-brews-studio/arra-oracle-skills-cli/psi/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 psi

Your own site · 80×15
<a href="https://agentmods.dev/skills/soul-brews-studio/arra-oracle-skills-cli/psi"><img src="https://agentmods.dev/badge/skills/soul-brews-studio/arra-oracle-skills-cli/psi.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 125 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 5,860 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 1 finding. 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 $0.00125 $0.05860
Opus 5 $0.00063 $0.02930
Sonnet 5 $0.00025 $0.01172
Haiku 4.5 $0.00013 $0.00586

Measured 11d ago against content hash b470c6d0ac8a, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade C, and why

psi scanned grade C with 1 finding 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 11d 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.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

Never `rm -rf` the vault — see "nothing deleted".
skills/psi/SKILL.md · 425 lines

How it starts

The opening of the file, as written. The whole thing — 425 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/psi

"A code repo does not need to be an oracle to have a memory."

Point a plain code repo's ψ at a caretaker oracle's vault, so the repo keeps a brain without becoming an oracle and without committing vault content into public source history.

/psi          # same as check
/psi check    # what is ψ here, who takes care of it, is it safe
/psi link     # ask which oracle, then absorb → symlink → ignore → verify
/psi heal     # symlink vanished after a checkout, or ignore rules incomplete
/psi unlink   # go back to a standalone real ψ

The model

code repo/ψ  ──symlink──▶  caretaker oracle/ψ
   (ignored, never committed)     (tracked — it IS the brain)
Repo kind ψ git
code repo — software someone else could clone and build symlink to caretaker ignored, never tracked
oracle repo — has an oracle identity real directory tracked

Pick the caretaker — show, don't guess

Never assume which oracle takes care of a repo. Show the fleet and let the human name it. Linking to the wrong vault mixes two oracles' memory.

maw ls

Ask: "Which oracle should take care of this repo's memory?" The human answers with a name (neo, pulse, beta). Resolve it — the short name works:

maw locate "$ORACLE_NAME"   # prints: repo: /path/to/<name>-oracle  and  ψ/: present

Take the repo: line as CARETAKER. If maw locate finds nothing, or its ψ/ is not present, stop and say so — do not fall back to a guess.

Do not loop over repos or scan the filesystem looking for candidates. One repo, one question, one answer.

check

Report what you verified, never what you assume. Every line is the output of a command — readlink -f for the real target, ls-files -s for the mode bits, check-ignore -v for the rule that actually matched. Close with one FOCUS line: the single next action.

REPO=$(git rev-parse --show-toplevel) || exit 1
GI="$REPO/.gitignore"

# ── ψ: type, size, staleness ─────────────────────────────────────────────
if [ -L "$REPO/ψ" ]; then
  TARGET=$(readlink "$REPO/ψ"); REAL=$(readlink -f "$REPO/ψ")
  [ -e "$REPO/ψ/" ] && ALIVE=alive || ALIVE='DANGLING'
  case "$TARGET" in /*) ABS=' ⚠️ absolute';; *) ABS='';; esac
  PSI="symlink → $TARGET$ABS"; PSI2="resolves  $REAL ($ALIVE)"
elif [ -d "$REPO/ψ" ]; then
  N=$(find "$REPO/ψ" -type f ! -name '.DS_Store' | wc -l | tr -d ' ')
  NEW=$(find "$REPO/ψ" -type f ! -name '.DS_Store' -exec stat -f '%m' {} \; 2>/dev/null | sort -rn | head -1)
  AGE=$(( ( $(date +%s) - ${NEW:-$(date +%s)} ) / 86400 ))
  PSI="real dir · $N files · newest $(date -r "${NEW:-0}" +%F 2>/dev/null) (${AGE}d old)"; PSI2=""
else
  PSI="absent"; PSI2=""
fi

# ── git: is any of it tracked? is the LINK itself committed? ─────────────
TRACKED=$(git -C "$REPO" ls-files ψ | wc -l | tr -d ' ')
git -C "$REPO" ls-files -s ψ | rg -q '^120000' && BLOB='⚠️ SYMLINK COMMITTED' || BLOB=no

# ── ignore: both forms, and the rule git actually matched ───────────────
rg -qx 'ψ'  "$GI" 2>/dev/null && BARE=yes || BARE='NO ⚠️'
rg -qx 'ψ/' "$GI" 2>/dev/null && SLASH=yes || SLASH=NO
# cut -f1, NOT an awk positional field ref — see "Never use positional parameters"
# below. check-ignore -v separates source from pathname with a TAB, so field 1 is
# the matching rule.
RULE=$(git -C "$REPO" check-ignore -v ψ 2>/dev/null | cut -f1)
[ -n "$RULE" ] || RULE='NOT IGNORED ⚠️'

# ── kind + caretaker ────────────────────────────────────────────────────
[ "$TRACKED" -gt 0 ] && KIND='oracle repo (ψ is tracked — it IS the brain)' \
                     || KIND='code repo (ψ must stay ignored)'
# NOTE: no \K — this rg build rejects it, and the error would silently read as "none".
CARE=$(sed -n 's/^oracle: *//p' "$REPO/.claude/PSI_CARETAKER" 2>/dev/null)
[ -n "$CARE" ] || CARE='none recorded'

# ── render: no borders. two-space margin, aligned columns, grouped by blank
#    lines. identity block first, then the checks, then one FOCUS line.
#    Each check names the command that proved it — the report is its own audit.
#    NOTE: printf inline per row — no row()/fact() helpers, because a helper
#    would need positional parameters and the host rewrites those. See below.
F='  %-14s%s\n'          # identity line
R='  %-14s%-42s%-3s %s\n'  # check line

echo
printf "$F" repo "$(basename "$REPO")"
printf "$F" kind "$KIND"
printf "$F" ψ    "$PSI"
[ -n "$PSI2" ] && printf "$F" '' "$PSI2"
echo
printf "$R" tracked      "$TRACKED entries" "$([ "$TRACKED" -eq 0 ] && echo ✓ || echo ⚠)" 'git ls-files ψ'
printf "$R" symlink-blob "$BLOB"            "$([ "$BLOB" = no ] && echo ✓ || echo ⚠)"     'ls-files -s ψ → 120000'
printf "$R" bare-ψ       "$BARE"            "$([ "$BARE" = yes ] && echo ✓ || echo ⚠)"    'rg -qx ψ .gitignore'
printf "$R" 'ψ/'         "$SLASH"           "$([ "$SLASH" = yes ] && echo ✓ || echo ·)"   'rg -qx ψ/ .gitignore'
printf "$R" ignore-rule  "$RULE"            "$(case $RULE in *NOT*) echo ⚠;; *) echo ✓;; esac)" 'git check-ignore -v ψ'
printf "$R" caretaker    "$CARE"            "$([ "$CARE" = 'none recorded' ] && echo · || echo ✓)" '.claude/PSI_CARETAKER'
echo

Read the full file on GitHub · 425 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. 11d ago First seen · 425 lines · 125 tokens per session scan C b470c6d0ac8a

Subscribe to this mod's changes

psi is a skill published in the GitHub repository Soul-Brews-Studio/arra-oracle-skills-cli (121 stars, last pushed 5d ago), licensed MIT. It adds 125 tokens to every session and 5,860 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

media-ingest

Ingest video, audio, PDF, book, screenshot, and GitHub repo content into the brain. Multi-format handling with entity extraction and backlink propagation. Covers video-ingest, youtube-ingest, and book-ingest subtypes.

garrytan/gbrain · 52 tokens

mem0-oss-to-platform

Plan and then execute a migration of a project from the mem0 open-source / self-hosted SDK (the local Memory class) to the mem0 Platform / hosted / managed SDK (the MemoryClient class). Use this whenever a developer wants to move, switch, or migrate their mem0 usage off OSS/self-hosted to the hosted API — e.g.…

mem0ai/mem0 · 273 tokens

Cortex

Operate Cortex, the LifeOS memory system — the typed Knowledge Archive (People, Companies, Ideas, Research with typed related: links) plus recall of prior work sessions, ISAs, and conversations. Search, add, harvest, develop, ingest, distill, graph-navigate, recall. USE WHEN cortex, knowledge, knowledge base, search…

danielmiessler/LifeOS · 196 tokens

memory

Use when the user asks to remember, recall, forget, update, search, or inspect durable OpenSquilla memory, including profile facts in USER.md and long-term notes in MEMORY.md or memory//.md.

opensquilla/opensquilla · 44 tokens

ha-data-stores

Map of Hope Agent's local data stores and safe read-only query workflow. Use when the user asks where Hope Agent stores data, wants to inspect sessions/messages/memory/logs/background jobs/knowledge indexes/settings, asks the model to query local app data, or debugging requires checking persisted state. Trigger…

shiwenwen/hope-agent · 115 tokens

establishing-project-context

Use when the user asks to establish shared project language, or project work exposes a conflicting, renamed, or deprecated domain term that needs active semantic modeling. Routine small tasks stay on the fast path.

GanyuanRan/Aegis · 45 tokens