ValentinNikolaev/agent-skills / agent-plugins/skills/update-memory
update-memory
VerifiedRecord durable facts from completed project work (decisions, fixes, migrations, milestones) in an existing memory store. Not initial bootstrap, external-source ingestion, audits, or temporary notes.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 3.6 KB |
| agents/openai.yaml | config | 205 B |
| references/memory-contract.md | markdown | 12.4 KB |
SKILL.md
View on GitHub---
name: update-memory
description: Record durable facts from completed project work (decisions, fixes, migrations, milestones) in an existing memory store. Not initial bootstrap, external-source ingestion, audits, or temporary notes.
---
# Update Memory
Record the smallest durable change supported by completed project work.
## Load the Contract and Establish Context
1. Read [references/memory-contract.md](references/memory-contract.md) completely before proposing edits.
2. Resolve one project root and memory root from the user's request or current project. Ask for explicit consent before reading or writing any global store, sibling project, unrelated repository, or other root outside that project.
3. Read the active `MEMORY.md` and relevant local files. If no useful baseline exists, route to `create-memory`.
4. Gather evidence from the user's confirmed summary, adopted decisions, completed work, changed interfaces, tests, failures, and available focused repository diff or history. Treat repository state as evidence, not a substitute for user intent.
5. Route a task whose primary purpose is importing or refreshing supplied external content to `ingest-memory`. When a source-backed conclusion arose incidentally during project work, record full structured provenance under the contract.
## Decide Whether to Write
Write only a durable architectural or product decision, workflow or integration constraint, recurring preference, resolved failure and prevention rule, shipped contract or migration, operational milestone, or source-backed fact needed later.
Skip temporary progress, unfinished experiments, routine implementation detail, information obvious from code or existing instructions, secrets, private data, raw logs or conversations, and unadopted ideas.
Use this order:
1. Update an existing focused file only when its durable truth changed.
2. Propose one focused new file when durable knowledge is absent.
3. Skip and explain temporary, redundant, unsupported, or uncertain material.
Preview file-level changes and continue within the authorized store and task. Updating related entries and the index is part of maintaining a coherent store. Ask only when the destination is ambiguous, cross-root access lacks authorization, or a proposed replacement would discard unresolved conflicting evidence or exceed the requested scope. Preserve existing authorization across turns.
## Update and Reconcile
1. State the new durable fact and evidence in one sentence before editing.
2. Search all local memory filenames, names, descriptions, source identities, and distinctive claims for duplicates or conflicts.
3. Determine current truth from the strongest available evidence; do not automatically prefer existing text or newer text.
4. Update the smallest focused section. Preserve dated historical context when it explains the current state, mark superseded details explicitly, and keep unresolved contradictions in a short `Conflicts` or `Open Questions` section.
5. Follow the contract exactly. Preserve or add structured provenance, keep local `related` filenames symmetric, and update `last_updated` only when durable content changes.
6. Update `MEMORY.md` only when a file or its purpose changes, using the exact contract grammar.
## Validate and Report
Resolve `<skill-dir>` to the directory containing this `SKILL.md`, then run:
```text
python "<skill-dir>/scripts/validate_memory.py" <memory-root>
```
Fix only errors introduced within the approved scope. Report evidence used, files changed, facts skipped, duplicate or conflict decisions, validation results, and remaining limitations. Recommend `audit-memory` after cross-file, index, relationship, or provenance changes.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |