Fedor Muratov

tool-evaluation-protocol

Time-bounded protocol for evaluating new tools, libraries, or methodologies before adoption. Use when evaluating a candidate dependency for the project, researching a methodology (e.g. ML/LLM frameworks, testing approaches), or following up on a hype-driven recommendation.

systematic-debugging

4-phase bug-fix protocol (evidence, pattern, hypothesis, verify). Use when investigating any error, failure, or unexpected behavior, handling bug reports from users or monitoring, or diagnosing test failures with non-obvious causes.

skill-template

Canonical template and validation guide for ai-hats skills. Use when creating a new skill, reviewing an existing one for structural compliance, or deciding which pattern fits a new behavior.

skill-optimization

Audit and refactor library components to eliminate redundancy and staleness. Use during periodic library maintenance, after adding multiple new skills or rules, when a role's token budget is too high, or after retrospectives identify component issues.

silent-fallback-gate

PreToolUse hook — refuses an Edit or Write that adds an inert broad except. Use when a Python edit was blocked for a silent fallback.

retro-to-framework

Convert an observation about agent behavior — from a retrospective or from a single incident — into a framework-level change (hook, rule, skill, skill update). Use after a retrospective identifies problems that are not project-specific, after an incident where an agent misapplied a component it had read, when CLAUDE.md band-aids accumulate (more than 3 per-project rules that could be generic), or when the same problem recurs across multiple projects.

py-security-lint

PostToolUse hook — reports the project's own ruff security (`S`) findings on every `.py` you Edit/Write, non-blockingly; fail-open, no manual invocation.

代码审查

prompt-authoring

Choose the mechanism before the wording, then write prose that holds. Use before wording or rewording any rule, injection or skill body, and when a written instruction has failed to change behaviour.

library-editing

Put a component where this project reads it, and wire it to a role. Use before adding or removing a trait, rule or skill, and when a component you wrote is not reaching the agent.

library-change-hypothesis-protocol

After execute-stage commits land (diff finalized) and BEFORE the task transitions to done, file a companion HYP with an explicit verification_protocol — or record an explicit "no behavior change" note for refactors at plan-stage. NEVER file the HYP at plan-stage (precommitment anti-pattern).

composition-verification

Prove a role/trait/rule/skill edit reached the composed prompt. Use after editing any library component, before reporting a composition change landed, and when a rule or skill looks absent from a role that should carry it.

comment-length-lint

PostToolUse hook — flags oversized comment blocks and docstrings on every `.py` you Edit/Write and forwards them non-blockingly; fail-open, no manual invocation.

代码审查

bash-mastery

Shell scripting, Makefile conventions, and modern CLI tooling. Use when writing or reviewing shell scripts, creating or modifying Makefiles, or choosing CLI tools for automation tasks.

agent-cli-guidelines

Design a CLI, gate or hook whose reader is another agent. Use when building or changing one, and when a tool of yours was ignored, misread, or worked around.

worktree-isolation

Isolated development using git worktrees so the main branch stays clean. Use when starting any non-trivial task (execute state), doing parallel work on multiple tasks, making risky changes you might want to discard, or delegating to a sub-agent (automatic via ai-hats agent --isolation).

tool-call-hygiene

Choose between Bash and dedicated Claude Code tools (Grep/Glob/Read/Edit) and edit files efficiently. Triggers on intent to grep/find/cat/sed/rg, multiple sequential reads, session-start context restore, codebase discovery, or multiple sequential Edits to one file.

task-summary

Focused post-task summary covering architectural decisions, decision forks, pitfalls, and plan deviations. Use at `document → review`, before handing the card to the reviewer, on a task that failed, or when the supervisor asks what happened with a task.

task-slicing

Wire a work breakdown into the tracker — depends_on edges, frontier, expand–contract for wide refactors. Use when filling plan.md's Steps section (routed from plan-gate), decomposing an epic into child tasks, or carving a mid-execute remainder into successor tasks.

律师与法务