But this issue doesn't stop at First Street.
Other major companies such as @BlackRock, @Vanguard_Group, and @StateStreet have a stake in platforms that display these ludicrous scores.
That is why we sounded the alarm about major Wall Street firms abusing their power over American homebuyers to push their net-zero fantasies.
this is pure f*cking treasure
10 connected parts for a serious Opus 5.5 harness
context. skills. tools. permissions. hooks. review. effort. goals. state. recovery
(1/10) CLAUDE.md
https://t.co/UHs6BY8lwt
> give the project its facts, conventions, source requirements, and output paths
(2/10) reusable skills
https://t.co/y9Bi4ipFpG
> save the procedure: create the output, check it, record what happens next
(3/10) MCP
https://t.co/MkXDn7Rl95
> connect the tools your task needs and retrieve one known source before relying on them
(4/10) permissions
https://t.co/0K8BNXqKcl
> use /permissions to inspect the effective allow, ask, and deny rules
(5/10) PreToolUse hooks
https://t.co/CFyRQ4dJaH
> inspect the proposed tool call and its arguments before execution
(6/10) evidence reviewer
https://t.co/HI76SvzP47)
> configure a focused read/search worker that returns claim, verdict, source, and correction
(7/10) reasoning effort
https://t.co/6BajPWfqFX
> start the main run at medium; give difficult review work its own assignment at high
(8/10) /goal
https://t.co/JznJdmit2l
> define the required outputs and checks, then surface the evidence in the conversation
(9/10) progress.md
> save completed work, decisions, open issues, and the next action after each meaningful stage
(10/10) checkpoints + Git
https://t.co/9kPcEhWXNB
> use /rewind for tracked file edits and Git for durable history
start the session:
claude --model claude-opus-5-5 --effort medium
the loop:
load context → run the skill → use the tools → review the result → prove completion → save progress → resume
save this, then read the guide for Opus 5.5 Harness Engineering ↓
🌞OM SURYA DEVAY NAMAH🌞
O
M
S
U
R
Y
A
D
E
V
A
Y
N
A
M
A
H
O
M
S
U
R
Y
A
D
E
V
A
Y
N
A
M
A
H
M
S
U
R
Y
A
D
E
V
A
Y
N
A
M
A
H
O
M
S
U
R
Y
A
D
E
V
A
Y
N
A
M
A
H
O
M
S
U
R
Y
A
D
E
V
A
Y
N
A
M
A
H
@grok
ॐ PUSHAN NAMAH
THIS SHT IS A F*CKING CHEAT CODE
the best Claude Code bug-fix trick almost nobody runs yet.
The idea: scout hunts the bug, critic tries to prove it wrong, and a fresh Opus session writes the fix from what survived.
The guesses never reach the code, because the session that writes the fix never sees them. It only gets the claims critic signed off on and a test that's already failing.
You just set it up like this:
/agents
here's the split:
→ scout traces the bug and writes every claim into BUG.md with a file:line
→ critic checks each claim against the code and sends anything unproven back to scout
→ haiku writes a test that HAS to go red on the current code first
→ /clear, a fresh Opus session gets only BUG.md + the failing test and writes the fix
→ haiku runs the full suite so nothing else breaks
Opus only shows up at two moments:
→ writing the fix
→ the final review before you merge
Scout digs. Critic doubts. Opus fixes.
That's the whole rule: the agent that guessed never writes the fix, and the session that writes the fix never sees the guesses.
the full tree
/agents → scout + critic (read-only), tester (haiku)
scout writes BUG.md with a file:line for every claim
critic passes or returns each claim
tester writes a test that fails right now
/clear → Opus fixes from BUG.md + test only
loop until the test is green and the suite is green
paste this prompt into Claude Code ↓
"Fix [BUG] using this setup:
Use the scout subagent to find the root cause. Write BUG.md: cause, repro steps, and a file:line for every claim.
Send BUG.md to the critic subagent. Any claim it can't confirm in the code goes back to scout.
Use the tester subagent to write a test that reproduces the bug and fails on the current code. Show me the failure.
In a new session with only BUG.md and the test, write the smallest fix that makes the test pass without breaking the suite.
Show me BUG.md first. Don't touch the code until I say go."
save it, you'll f*cking need it on your next bug.