Correct Claude Code project memory and check a fresh session
Inspect and correct one owned Claude Code project memory note and its index, then check a fresh local session against the project's authoritative source.
Follow the story
Agent instruction reliability
Explore BIG CHANGE reporting on Agent instruction reliability: what changed, how it works and why it matters.
Latest reporting
35 stories
Build an agent feedback loop and check if it improves
Use Promptfoo, an existing task agent and approved fixtures to compare a baseline with one loaded lesson on development and held-out cases.
Build a checked Claude Code report from GitHub records
Use three Claude Code skills to collect GitHub commits and open issues, draft from saved evidence, check claims, then verify a manual run before scheduling.
Build a Codex skill from work you have already accepted
Trace an accepted artifact back to its real input, write a new Codex skill from the repeatable steps, then check it on a second task and revise an observed defect.
Give an AI agent Redis memory and verify recall across sessions
Save a real authorized user fact in Redis Agent Memory, inspect automatic extraction, then verify a fresh LangGraph agent received and used the scoped memory.
Measure MCP tool context without losing the answer
Build a pinned local MCP server and Claude host, compare five context controls against a broad baseline, and check answers before token totals.
Run parallel Claude Code research subagents and verify their findings
Set up two read-only Claude Code research subagents, check their actual task and model state, then reconcile cited findings, failures and usage in the main session.
Build a checked two-model FAQ benchmark with Promptfoo
Use an approved FAQ, frozen questions and Promptfoo to compare two models, then inspect raw answers, grades, errors and billing before making a decision.
Give two agents the same reviewed memory record
Store one source-approved typed memory record, supply its exact reviewed bytes to two authorized agent sessions, then verify a conditional update and resolve conflicts.
Run Claude Code Ultrareview on one branch and verify its findings
A documentation-based guide to checking repository upload and credits before one Claude Code cloud review, then verifying its findings against the branch before a human merge decision.
Use Superpowers in Claude Code to plan, test and review one code change
A documentation-based guide to taking one owned code change through an approved Superpowers design and plan, observed red-green tests, task and branch review, and an owner decision.
Delegate one Claude Code coding task from Opus to Haiku and verify the result
A documented Claude Code workflow for sending one bounded coding change from an Opus parent to a project Haiku subagent, checking the actual model and diff, and interpreting usage without assuming a saving.
Install Context Mode in Claude Code and check its session output
Install Context Mode in Claude Code, inspect its hooks and session statistics during a bounded coding task, then verify the actual files and checks before accepting the work.
Fix and verify one failing repository check with an AI Goal
Set a bounded Claude Code or Codex CLI Goal for one existing failing check, then verify the original command and final diff before accepting a repair.
Review a Claude Managed Agent memory store with one Dream
Run one manual Claude Dream over authorized sessions, inspect and correct its separate memory store, then attach only reviewed memory to a later session.
Create an owner-approved DESIGN.md for Claude Code
Write a visual specification from approved brand decisions, then use it for two small Claude Code edits and inspect both diffs and renders.
Check Claude Code instruction scope across two client projects
Compare the instructions loaded in two fresh Claude Code project sessions, then verify when a nested instruction file joins context after an authorized read.
Add a bounded project context hook to Claude Code session start
Install a small project SessionStart hook that adds the current Git branch to Claude Code context. Check its direct output, startup and resume, and failure behavior.