Executes these capabilities
Inside this skill
The internal capability tree shows how this skill gets its work done.
- context inventory
- memory file design
- decision logging
- task state tracking
- conflict detection
The contract
Input
- Project conventions and current state
- Recent decisions and open questions
Output
- Structured memory files at the repository root
- Decision log with dates and rationale
- Current task state
Deliverables
- Maintained context files
- Handoff-ready repository state
Use cases
- Multi-agent projects
- Long-running projects with many sessions
- Onboarding new agents to existing work
Dependencies and replacements
Dependencies
- Git
- An agent that loads startup context
Required tools
- A text editor
- Git
Required permissions
- Write access to the repository
Compatible with
- agent-handoff
- validation-checklist
Alternatives
- Issue tracker as sole context source
- Wiki or Notion pages
Replaceable components
- File names and formats
- The specific context file the agent loads
Reliability
Human checkpoints
- Review memory updates before merging
- Approve decisions before they enter the log
Limitations
- Memory is only as current as the last update
- Cannot capture what was never written down
- Too much memory is as bad as none
Failure modes
- Memory files drift from the actual codebase
- Agents overwrite each other's updates
- Decisions logged without context
Recovery steps
- Diff memory files against the repository on each session
- Route updates through a single review step
- Log decisions with date, author, and rationale
Validation methods
- A fresh agent session answers project questions from memory alone
- Memory files pass a diff review on each update
Sources
Entry checked on Jul 31, 2026. Verification status reflects what AgentSkillTree has independently run, not the skill's own claims.
When to use
Use when more than one agent, or more than one session, works on the same project. Without durable context, each session restarts from zero.
How it works
Design memory files for the questions agents actually ask: what is this project, what are the conventions, what decisions were made, what is next. Keep them short enough to read at startup and structured enough to diff.