agentsclimarketplace

Brain harness

Skill 0SxD/agent-skills-plugin-v01/skills/brain-harness

Install
npx -y skills add 0SxD/agent-skills-plugin-v01 --skill brain-harness

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

Full session bootup sequence. Triggered when user asks about accessing memory, putting on the harness, checking the brain, or any similar phrase indicating they want full session initialization. Reads all memory files across all active projects and reports session state before asking what we're working on today.

SKILL.md

4.3 KB, ~1.0k tokens by cl100k_base, as published. Nobody here has run it

Brain Harness -- Full Bootup Sequence

Trigger phrases (heuristic -- match the intent, not the exact words)

Any of these should trigger this sequence:

  • "put on the brain harness" / "put on the harness"
  • "put on the brain" / "load the brain"
  • "do you have a brain?" / "can you access your brain?"
  • "load your memory" / "read your memory"
  • "boot up" / "get up to speed" / "get context"
  • Anything that sounds like the user wants you to initialize for a session

When in doubt: if the user seems to be asking you to get acquainted with the current state of the workspace, run this sequence.

Execute this sequence completely before responding.

Step 1 -- Read propagation manifest

Read: <local-path>/.claude/propagation_manifest.md This tells you where all project brain paths and hooks live. If it doesn't exist, note it and continue.

Step 2 -- Read global memory

Read in order:

  • <local-path>/.claude/projects/<project-slug>/memory/MEMORY.md (routing table -- read this first)
  • <local-path>/.claude/OPEN_BRAIN.md (cross-project identity, if exists)

Step 3 -- Read OpenBrainLM project brain

Read in order:

  • <local-path>/OpenBrainLM/memory/short_term.md
  • <local-path>/OpenBrainLM/memory/long_term.md
  • <local-path>/OpenBrainLM/memory/todos.md
  • <local-path>/OpenBrainLM/memory/connections.md
  • <local-path>/OpenBrainLM/memory/decisions.md (if exists)
  • <local-path>/OpenBrainLM/memory/blockers.md (if exists)

Step 4 -- Read Trading Bot project brain

Read in order:

  • <local-path>/trading_bot_build/memory/short_term.md
  • <local-path>/trading_bot_build/memory/long_term.md
  • <local-path>/trading_bot_build/memory/todos.md (if exists)
  • <local-path>/trading_bot_build/memory/connections.md
  • <local-path>/trading_bot_build/memory/decisions.md (if exists)
  • <local-path>/trading_bot_build/memory/blockers.md (if exists)
  • <local-path>/trading_bot_build/memory/indicators.md (if exists)

Step 5 -- Read NautilusTrader project brain

Read in order:

  • <local-path>/nautilus_trader/memory/short_term.md
  • <local-path>/nautilus_trader/memory/long_term.md (if exists)
  • <local-path>/nautilus_trader/memory/connections.md (if exists)
  • <local-path>/nautilus_trader/memory/decisions.md (if exists)
  • <local-path>/nautilus_trader/memory/blockers.md (if exists)

Step 6 -- Check for active handoff

After reading all memory files, look at OpenBrainLM/memory/short_term.md for a section marked ## SESSION HANDOFF. If one exists:

  • It contains specific first-action instructions. Follow them EXACTLY and IN ORDER.
  • The handoff overrides the generic "What are we building today?" question.
  • Exception: if the user gave you a specific task alongside the harness request (e.g. "load the harness then do X"), hold that task in mind -- complete the handoff sequence first, then address the additional task.
  • Do NOT ask "What are we building today?" if a handoff exists. The handoff tells you what to do. Follow it.
  • Only ask "What are we building today?" if there is NO active handoff AND the user gave no additional task.

If no active handoff section exists, go straight to Step 7.

Step 7 -- Report session state

Report a concise summary (no more than 10 lines):

  • Last session: what was being worked on
  • Active blockers: list them (from blockers.md files)
  • In-progress tasks: list them
  • Pending agents: any that were running

Step 8 -- Ask the single question

After the report (and after any handoff actions are complete), ask exactly:

"What are we building today?"

Wait for the answer. Write it as the session header in the relevant project's short_term.md before doing anything else.

What NOT to do

  • Do not start any work before completing steps 1-8
  • Do not skip an active handoff -- if one exists, follow it before asking the generic question
  • Do not ask multiple questions -- the single question is "What are we building today?"
  • Do not skip memory files -- read all of them
  • Do not summarize from memory alone -- read the actual files

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.