AMDResearch
Official@amdresearch
Offers structured project memory management and session orchestration for maintaining technical consistency across complex engineering lifecycles.
Agent Skills by AMDResearch
Showing 28 vetted skills indexed across 2 GitHub repositories.
cleanroom-test
Automates end-to-end cleanroom testing of omniprobe across build, install and relocated trees.
workflow-create
Create a draft workflow packet with dossier, run-log, handoff, and artifacts files.
pm-investigation
Manage investigation workflow packets with templates, run logs, and artifacts.
workflow-refine
Refine rough work requests into structured briefs for workflow-create.
pm-performance
Create structured performance investigation workflows with baselines, targets, and evidence-based progression.
pm-review
Creates/manages structured code, design, or document review workflows with severity-based findings and traceability.
pm-feature
Create feature workflow packets with dossier, run-log, handoff, and artifacts templates.
pm-validate
Validate project memory registries and auto-heal structural inconsistencies.
pm-bugfix
Automate bug-fix workflows from reproduction to regression verification.
session-close
Automates end-of-session orchestration across workflow, PM, and state artifacts with commit and capture procedures.
pm-load
Load minimal Project Memory units from .agents/pm files for the current task.
lessons-forward
Forward generalizable template findings to the Agentic Meta Project.
docs-sync
Aligns documentation indices and cross-reference links after edits to docs/ files.
feedback
Collect user feedback and file GitHub issues or local reports.
discussion-refine
Convert unstructured design discussions into structured briefs with decisions and open questions.
pm-restructure
Convert reflection report recommendations into approved PM unit structural changes.
pm-refactor
Create and manage refactor packets with lifecycle states and verification gates.
pm-reflect
Analyze PM unit structure and generate a reflection report with restructuring recommendations.
session-init
Bootstrap an agent session by reading project metadata and generating a briefing.
pm-update
Persist durable project knowledge into Project Memory after meaningful work.
workflow-readiness-check
Validates draft workflow packets for missing files and conflicts before promotion to active status.
decision-workflow
Record structured design decisions as DW-NNN documents with options and tradeoffs.
workflow-resume
Resume ongoing workflows by restoring context from handoff, dossier, and run-log.
state-check
Cross-reference workflow state files, directories, and run logs to report inconsistencies.
Frequently Asked Questions About AMDResearch
FAQPage SchemaWhat specific tasks are enabled by these project memory capabilities?▼
These capabilities enable the initialization, loading, and structural maintenance of project memory units. Users can perform session bootstrapping, state capture, and reflection-based restructuring to ensure durable knowledge persistence and consistent documentation alignment across complex engineering tasks.
Which personas benefit most from these project memory and session management skills?▼
Technical leads, systems architects, and software engineers managing long-running projects benefit from these skills. They are designed for teams requiring rigorous traceability, structured design decision logging, and automated validation of project state across multiple development sessions.
What are the primary prerequisites for implementing these project memory units?▼
Implementation requires a codebase structured to support .agents/pm metadata files and directory-based packet management. Users must maintain consistent dossier, run-log, and handoff file formats to enable the registry to perform validation, state checking, and session capture procedures effectively.